Monday, December 30, 2019

Better Site Speed: 4 Outside-the-Box Ideas

Posted by Tom-Anthony

Most of us have done site speed audits, or seen audits done by others. These can be really helpful for businesses, but I often find they're quite narrow in focus. Typically we use well-known tools that throw up a bunch of things to look at, and then we dive into things from there.

However, if we dig deeper, there are often other ideas on how site speed can be improved. I often see plenty of opportunities that are never covered in site speed audits. Most site speed improvements are the result of a bunch of small changes, and so in this post I’m going to cover a few ideas that I’ve never seen in any site speed audit, all of which can make a difference.

A different angle on image optimization

Consider optimized SVGs over PNGs

I was recently looking to book some tickets to see Frozen 2 (because of, erm, my kids...) and so landed on this page. It makes use of three SVG images for transport icons:

SVG images are vector images, so they're well-suited for things like icons; if you have images displayed as PNGs you may want to ask your designers for the original SVGs, as there can be considerable savings. Though not always better, using an SVG can save 60% of the filesize.

In this case, these icons come in at about 1.2k each, so they are quite small. They would probably fly under the radar of site speed audits (and neither Page Speed Insights or GTMetrix mention these images at all for this page).

So you may be thinking, “They're less than 5k combined — you should look for bigger issues!”, but let's take a look. Firstly, we can run them all through Jake Archibald’s SVG compression tool; this is a great free tool and on larger SVGs it can make a big difference.

In this case the files are small, so you may still be thinking "Why bother?" The tool compresses them without any loss in quality from ~1240 bytes to ~630 bytes — a good ratio but not much of an overall saving.

However… now that we've compressed them, we can think differently about delivering them…

Inline images

GTMetrix makes recommendations around inlining small bits of CSS or JS, but doesn’t mention inlining images. Images can also be inlined, and sometimes this can be the right approach.

If you consider that even a very small image file requires a complete round trip (which can have a very real impact on speed), even for small files this can take a long time. In the case of the Cineworld transport images above, I simulated a "Fast 3G" connection and saw:

The site is not using HTTP2 so there is a long wait period, and then the image (which is 1.2kb) takes almost 600ms to load (no HTTP2 also means this is blocking other requests). There are three of these images, so between them they can be having a real impact on page speed.

However, we've now compressed them to only a few hundred bytes each, and SVG images are actually made up of markup in a similar fashion to HTML:

You can actually put SVG markup directly into an HTML document!

If we do this with all three of the transport images, the compressed HTML for this page that is sent from the server to our browser increases from 31,182 bytes to 31,532 bytes — an increase of only 350 bytes for all 3 images!

So to recap:

  • Our HTML request has increased 350 bytes, which is barely anything
  • We can discard three round trips to the server, which we can see were taking considerable time

Some of you may have realized that if the images were not inline they could be cached separately, so future page requests wouldn’t need to refetch them. But if we consider:

  • Each image was originally about 1.5kb over the network (they aren’t gzipping the SVGs), with about 350 bytes of HTTP headers on top for a total of about 5.5kb transferred. So, overall we've reduced the amount of content over the network.
  • This also means that it would take over 20 pageviews to benefit from having them cached.

Takeaway: Consider where there are opportunities to use SVGs instead of PNGs.

Takeaway: Make sure you optimize the SVG images, use the free tool I linked to.

Takeaway: Inlining small images can make sense and bring outsized performance gains.

Note: You can also inline PNGs — see this guide.

Note: For optimized PNG/JPG images, try Kraken.

Back off, JavaScript! HTML can handle this...

So often nowadays, thanks to the prevalence of JavaScript libraries that offer an off-the-shelf solution, I find JavaScript being used for functionality that could be achieved without it. More JS libraries means more to download, maybe more round trips for additional files from the server, and then the JavaScript execution time and costs themselves.

I have a lot of sympathy for how you get to this point. Developers are often given poor briefs/specs that fail to specify anything about performance, only function. They are often time-poor and so it's easy to end up just dropping something in.

However, a lot of progress has been made in terms of the functionality that can be achieved with HTML and or CSS. Let's look at some examples.

Combo box with search

Dropdown boxes that have a text search option are a fairly common interface element nowadays. One recent article I came across described how to use the Select2 Javascript library to make such a list:

It is a useful UI element, and can help your users. However, in the Select2 library is a JavaScript library, which in turn relies on some CSS and the JQuery library. This means three round trips to collect a bunch of files of varying sizes:

  • JQuery - 101kb
  • Select2 JavaScript - 24kb
  • Select2 CSS - 3kb

This is not ideal for site speed, but we could certainly make the case it is worth it in order to have a streamlined interface for users.

However, it is actually possible to have this functionality out of the box with the HTML datalist element:

This allows the user to search through the list or to free type their own response, so provides the same functionality. Furthermore, it has a native interface on smartphones!

You can see this in action in this codepen.

Details/Summary

LonelyPlanet has a beautiful website, and I was looking at this page about Spain, which has a ‘Read More’ link that most web users will be familiar with:

Like almost every implementation of this that I see, they have used a JavaScript library to implement this, and once again this comes with a bunch of overheads.

However, HTML has a pair of built-in tags called details and summary, which are designed to implement this functionality exactly. For free and natively in HTML. No overheads, and more accessible for users needing a screen reader, while also conveying semantic meaning to Google.

These tags can be styled in various flexible ways with CSS and recreate most of the JS versions I have seen out there.

Check out a simple demo here: https://codepen.io/TomAnthony/pen/GRRLrmm

...and more

For more examples of functionality that you can achieve with HTML instead of JS, check out these links:

  • http://youmightnotneedjs.com/
  • https://dev.to/ananyaneogi/html-can-do-that-c0n

Takeaway: Examine the functionality of your sites and see where there may be opportunities to reduce your reliance on large Javascript libraries where there are native HTML/CSS options.

Takeaway: Remember that it isn’t only the size of the JS files that is problematic, but the number of round trips that are required.

Note: There are cases where you should use the JS solution, but it is important to weigh up the pros and cons.

Networking tune-ups

Every time the browser has to collect resources from a server, it has to send a message across the internet and back; the speed of this is limited by the speed of light. This may sound like a ridiculous thing to concern ourselves with, but it means that even small requests add time to the page load. If you didn’t catch the link above, my post explaining HTTP2 discusses this issue in more detail.

There are some things we can do to help either reduce the distance of these requests or to reduce the number of round trips needed. These are a little bit more technical, but can achieve some real wins.

TLS 1.3

TLS (or SSL) is the encryption technology used to secure HTTPS connections. Historically it has taken two round trips between the browser and the server to setup that encryption — if the user is 50ms away from the server, then this means 200ms per connection. Keep in mind that Google historically recommends aiming for 200ms to deliver the HTML (this seems slightly relaxed in more recent updates); you're losing a lot of that time here.

The recently defined TLS 1.3 standard reduces this from two round trips to just one, which can shave some precious time off the users initial connection to your website.

Speak to your tech team about migrating to TLS 1.3; browsers that don’t support it will fallback to TLS 1.2 without issue. All of this is behind the scenes and is not a migration of any sort. There is no reason not to do this.

If you are using a CDN, then it can be as simple as just turning it on.

You can use this tool to check which versions of TLS you have enabled.

QUIC / HTTP 3

Over the last 2-3 years we have seen a number of sites move from HTTP 1.1 to HTTP 2, which is a behind-the-scenes upgrade which can make a real improvement to speed (see my link above if you want to read more).

Right off the back of that, there is an emerging pair of standards known as QUIC + HTTP/3, which further optimize the connection between the browser and the server, further reducing the round trips required.

Support for these is only just beginning to become viable, but if you are a CloudFlare customer you can enable that today and over the coming 6 months as Chrome and Firefox roll support out, your users will get a speed boost.

Read more here: https://blog.cloudflare.com/http3-the-past-present-and-future/

Super routing

When users connect to your website, they have to open network connections from wherever they are to your servers (or your CDN). If you imagine the internet as a series of roads, then you could imagine they need to ‘drive’ to your server across these roads. However, that means congestion and traffic jams.

As it turns out, some of the large cloud companies have their own private roads which have fewer potholes, less traffic, and improved speed limits. If only your website visitors could get access to these roads, they could ‘drive’ to you faster!

Well, guess what? They can!

For CloudFlare, they provide this access via their Argo product, whereas if you are on AWS at all then you can use their Global Accelerator. This allows requests to your website to make use of their private networks and get a potential speed boost. Both are very cheap if you are already customers.

Takeaway: A lot of these sorts of benefits are considerably easier to get if you're using a CDN. If you're not already using a CDN, then you probably should be. CloudFlare is a great choice, as is CloudFront if you are using AWS. Fastly is the most configurable of them if you're more of a pro.

Takeaway: TLS 1.3 is now very widely supported and offers a significant speed improvement for new connections.

Takeaway: QUIC / HTTP3 are only just starting to get support, but over the coming months this will roll out more widely. QUIC includes the benefits of TLS 1.3 as well as more. A typical HTTP2 connection nowadays needs 3 round trips to open; QUIC needs just one!

Takeaway: If you're on CloudFlare or AWS, then there is potential to get speed ups just from flipping a switch to turn on smart routing features.

Let CSS do more

Above I talked about how HTML has built-in functionality that you can leverage to save relying on solutions that are ‘home-rolled’ and thus require more code (and processing on the browsers side) to implement. Here I'll talk about some examples where CSS can do the same for you.

Reuse images

Often you find pages that are using similar images throughout the page in several places. For example, variations on a logo in different colors, or arrows that point in both directions. As unique assets (however similar they may be), each of these needs to be downloaded separately.

Returning to my hunt for cinema tickets above, where I was looking at this page, we can see a carousel that has left and right arrows:

Similarly to the logic used above, while these image files are small, they still require a round trip to fetch from the server.

However, the arrows are identical — just pointing in opposite directions! It's easy for us to use CSS’s transform functionality to use one image for both directions:

You can check out this codepen for an example.

Another example is when the same logo appears in different styles on different parts of the page; often they will load multiple variations, which is not necessary. CSS can re-color logos for you in a variety of ways:

There is a codepen here showing this technique in action. If you want to calculate the CSS filter value required to reach an arbitrary color, then check out this amazing color calculator.

Interactions (e.g. menus & tabs)

Often navigation elements such as menus and tabs are implemented in JavaScript, but these too can be done in pure CSS. Check out this codepen for an example:

Animations

CSS3 introduced a lot of powerful animation capability into CSS. Often these are not only faster than JavaScript versions, but can also be smoother too as they can run in the native code of the operating system rather than having to execute relatively slower Javascript.

Check out Dozing Bird as one example:

You can find plenty more in this article. CSS animations can add a lot of character to pages at a relatively small performance cost.

...and more

For more examples of functionality that you can achieve using pure CSS solutions, take a look at:

  • http://youmightnotneedjs.com/
  • https://dev.to/ananyaneogi/css-can-do-that-18g7m

Takeaway: Use CSS to optimize how many files you have to load using rotations or filters.

Takeaway: CSS animations can add character to pages, and often require less resources than JavaScript.

Takeaway: CSS is perfectly capable of implementing many interactive UI elements.

Wrap up

Hopefully you've found these examples useful in themselves, but the broader point I want to make is that we should all try to think a bit more out of the box with regards to site speed. Of particular importance is reducing the number of round trips needed to the server; even small assets take some time to fetch and can have an appreciable impact on performance (especially mobile).

There are plenty more ideas than we've covered here, so please do jump into the comments if you have other things you have come across.


Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!

Friday, December 27, 2019

The Not-So-Secret Value of Podcast Transcripts - Whiteboard Friday

Posted by NikiMosier

What are the benefits of transcribing your podcasts and what's the best way to go about getting them on your site? Niki Mosier breaks it down into 8 easy steps in this week's episode of Whiteboard Friday.

Click on the whiteboard image above to open a high resolution version in a new tab!

Video Transcription

Hey, Moz fans. Here's another edition of Whiteboard Friday. I'm Niki Mosier, a senior SEO account manager at Two Octobers, and I'm here today to talk to you about the not-so-secret value of podcast transcripts.

I got the idea to play around with podcast transcripts after hearing Moz's Britney Muller talk about machine learning and podcast transcripts at TechSEO Boost last fall. 

+15% increase in organic traffic, +50% keyword lift

I ended up getting the opportunity to play around with this a little bit with a pro bono client we had at a previous job, the Davis Phinney Foundation. They do Parkinson's research and Parkinson's education. They were already podcasting, and then they also had a pretty robust blog, but they weren't adding their podcast transcripts. After about three months of adding a couple of podcast transcripts, we saw some pretty good value for them. We saw a 15% increase in organic traffic to the website and a 50% increase to some keyword lift around the keywords that we were tracking.

Google is now indexing podcasts

Why we think this is relevant right now, as you may know, Google announced, at I/O 2019, that they are indexing podcasts. If you do a search for your favorite podcast, you'll see that come up in the Google search results now. So adding that podcast transcript or any audio transcript to your website, whether that's video, a webinar, or anything, just has some really good value.

How to transcribe & optimize your podcasts

I'm going to walk you through the process that I used for them. It's super easy and you can turn around and apply it to your own website. 

1. Download your audio file

So obviously, download the audio file, whether that's MP3 or MP4 or whatever you have, from your video, podcast, or your webinars if you're doing those. 

2. Transcribe it

You need to be able to get that text transcript, so running it through either Temi or Otter.ai, both two resources that I've used, both really good. Otter.ai seems to be a little cleaner out of the gate, but I would definitely obviously go through and edit and make sure that all of your text and speaker transitions and everything is accurate. 

3. Figure out which keywords the content should rank for

Next up is figuring out what keywords that you want that content to rank for, so doing some search volume research, figuring out what those keywords are, and then benchmarking that keyword data, so whether your website is already ranking for some of those keywords or you have new keywords that you want those pages or those posts to be ranking for.

4. Get a competitive snapshot

Next up is getting a competitive snapshot, so looking at who's ranking for those keywords that you're going to be trying to go after, who has those answer boxes, who has those featured snippets, and then also what are the people also ask features for those keywords. 

5. Get your content on-site

Obviously getting that content on your site, whether that's creating brand-new content, either a blog or a page to go with that podcast, video, webinar, or whatever it is, or adding to it to existing content.

Maybe you have some evergreen content that's not performing well for you anymore. Adding a transcript to that content could really kind of give it a lift and make it work better for you. 

6. Optimize the content

Next up is optimizing the content on your site, so adding in those keywords to your metadata, to your image alt tags, your H1 tags, and then also adding any relevant schema, so whether that's blog post schema most likely or any other schema type that would be helpful, getting that up there on the page as well.

7. Make sure the page is indexed in Search Console

Once you've done all the hard work, you've got the transcript up there, you have your content and you have it optimized, you obviously want to tell Google, so going into Search Console, having them index that page, whether it's a new page or an existing page, either way, dropping that URL in there, making sure Google is crawling it, and then if it is a new page, making sure it's in your sitemap.

8. Annotate the changes in Google Analytics

Then the last thing is you want to be able to track and figure out if it's working for you. So annotating that in Google Analytics so you know what page, when you added it, so you can have that benchmark date, looking at where you're ranking, and then also looking at those SERP features. Have you gotten any featured snippets?

Are you showing up in those answer boxes? Anything like that. So that's kind of the process. Super easy, pretty straightforward. Just play with it, test it out. 

If Google is indexing podcasts, why does this matter?

Then kind of lastly, why is this still important if Google is already indexing podcasts? They may come out and do their own transcription of your podcast or your video or whatever content you have on the site.

Obviously, you want to be in control of what that content is that's going on your site, and then also just having it on there is super important. From an accessibility standpoint, you want Google to be able to know what that content is, and you want anyone else who may have a hearing impairment, they can't listen to the content that you're producing, you want them to be able to access that content. Then, as always, just the more content, the better. So get out there, test it, and have fun. Thanks, Moz fans.

Video transcription by Speechpad.com


Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!

Thursday, December 26, 2019

They're the Best Around: The Top 25 Moz Blog Posts of 2019

Posted by FeliciaCrawford

Well, folks, it's that time of year again. It's hard to believe we've already gone another turn around the ol' sun. But I've consulted my analytics data and made my SQL queries, and I'm here today to present to you the list of the top Moz Blog posts of 2019!

For a little perspective, we published 207 blog posts, averaging out to about 4 per week. Out of those 207, the twenty-five I'm sharing with you below were the most-read pieces of the year. If you're strapped for time (and who isn't in our industry?), survey says these are the articles that aren't to be missed. And bonus — a good chunk of them are videos, so bring out the chocolate popcorn and settle down to watch!

(If chocolate popcorn sounds new and unfamiliar to you, I implore you to check out the Cinerama in Seattle's Belltown neighborhood the next time you're in town for MozCon. It is life-changing. Get the mix of regular and chocolate and never, ever look back.)

I'll be sharing the top keywords each post ranks for according to Keyword Explorer, too, to give you some idea of why these posts have continued to be favorites throughout the year. Gotta love that "Explore by Site" feature — it makes my job way too easy sometimes! ;-)

(For the Keyword Explorer nerds in the audience, I'll be filtering the rankings to positions 1–3 and organizing them by highest monthly search volume. I want to see what we're ranking highly for that gets lots of eyeballs!)

Ready to get started? I sure am. Let's dive in.


The top 25 Moz Blog posts of 2019

1. On-Page SEO for 2019 - Whiteboard Friday

Britney Muller, January 4th

57,404 reads

Top keywords according to Keyword Explorer: seo 2019 (#3, 501–850), seo best practices 2019 (#3, 501–850), homepage seo 2019 (#1, 0–10)

On-page SEO has long been a favorite topic for y'all, and the top number-one winner, winner, chicken dinner post of 2019 reflects that loud and proud. In this expert checklist, Britney Muller shares her best tips for doing effective on-page SEO for 2019.

And if you want a hint on one reason this puppy has been so popular, check out #10 in this very list. ;-)

2. The 60 Best Free SEO Tools [100% Free]

Cyrus Shepard, June 10th

51,170 reads

Top keywords according to Keyword Explorer: seo tools (#1, 6.5k–9.3k), free seo tools (#1, 1.7k–2.9k), free seo (#1, 501–850)

This post is a testament to the power of updating and republishing your best content. Cyrus originally authored this post years ago and gave it a sorely needed update in 2019. There are literally hundreds of free SEO tools out there, so this article focused on only the best and most useful to add to your toolbox.

3. The Ultimate Guide to SEO Meta Tags

Kate Morris, July 24th

42,276 reads

Top keywords according to Keyword Explorer: seo meta tags (#1, 501–850), 1-page meta (#2, 501–850), what are meta tags (#3, 501–850)

Here's another vote for the power of republishing really good content that you know your audience craves. Originally published in November 2010, this is the second time we've asked Kate to update this article and it continues to deliver value ten years later. SEO certainly changes, but some topics remain popular and necessary throughout all the ups and downs.

4. The One-Hour Guide to SEO

Rand Fishkin, throughout 2019

41,185 reads for the first post (143,165 for all six combined)

Top keywords according to Keyword Explorer: moz seo guide (#2, 201–500), moz beginners guide to seo (#3, 101–200), moz guide to seo (#2, 11–50)

A "best of the Moz Blog" list wouldn't be complete without Rand! His six-part video series detailing all the most important things to know about SEO was originally published on the Moz Blog as six separate Whiteboard Fridays. We've since redirected those posts to a landing page in our Learning Center, but the first episode on SEO strategy earned over 41k unique pageviews in its time live on the blog.

5. A New Domain Authority Is Coming Soon: What’s Changing, When, & Why

Russ Jones, February 5th

38,947 reads

Top keywords according to Keyword Explorer: moving a 60 da to a 90 da seo (#1, 0–10), moz da update 2019 (#1, 0–10), upcoming domain change (#1, 0–10)

When we upgraded our Domain Authority algorithm in March, we knew it would be a big deal for a lot of people — so we put extra effort into education ahead of the launch. Russ's initial announcement post introducing the coming changes was the foremost source for information, earning ample attention as a result.

6. How Google Evaluates Links for SEO [20 Graphics]

Cyrus Shepard, July 1st

38,715 reads

Top keywords according to Keyword Explorer: free google picture of created equal (#2, 0–10), google 1 page 2 links (#2, 0–10), google top rankingillustrations (#2, 0–10)

All right, I admit it: we did a ton of content updating and republishing this year. And it seriously paid off. Cyrus revamped a perennially popular post by Rand from 2010, bumping it from ten graphics to twenty and giving it a much-needed refresh almost a decade after the original post. The top keywords are kind of weird, right? Check out the title on the original post — looks like we've got a little work to do with this one to get it ranking for more relevant terms!

7. Do Businesses Really Use Google My Business Posts? A Case Study

Ben Fisher, February 12th

32,938 reads

Top keywords according to Keyword Explorer:  google my business posts (#2, 201–500), how to post on google my business (#3, 101–200), google business post (#3, 51–100)

Even a couple of years after Google My Business Posts became an option, it wasn't clear how many businesses are actually using them. Ben Fisher asked the important questions and did the legwork to find the answers in this case study that examined over 2,000 GMB profiles.

8. Announcing the New Moz SEO Essentials Certificate: What It Is & How to Get Certified

Brian Childs, May 1st

32,434 reads

Top keywords according to Keyword Explorer: moz certification (#3, 101–500), moz seo certification (#2, 51–100), moz academy (#3, 51–100)

One of our most-asked questions from time immemorial was "Does Moz offer an SEO certification?" With the launch of our SEO Essentials certificate in May of this year, the answer finally became yes! 

9. Optimizing for Searcher Intent Explained in 7 Visuals

Rand Fishkin, March 23rd

29,636 reads

Top keywords according to Keyword Explorer: user intent moz (#2, 0–10)

What does it mean to target the "intent" of searchers rather than just the keyword(s) they've looked up? These seven short visuals explain the practice of intent-targeting and optimization.

10. 7 SEO Title Tag Hacks for Increased Rankings + Traffic - Best of Whiteboard Friday

Cyrus Shepard, June 7th

26,785 reads

Top keywords according to Keyword Explorer: title tags for landing page (#2, 11–50), moz free hack (#1, 0–10),  title tag hacks (#1, 0–10)

Title tags can have a huge impact on your click-through rates when optimized correctly. In this Whiteboard Friday, Cyrus shares how to use numbers, dates, questions, top referring keywords, and more to boost your CTR, traffic, and rankings.

11. E-A-T and SEO: How to Create Content That Google Wants

Ian Booth, June 4th

25,681 reads

Top keywords according to Keyword Explorer: eat seo (#2, 201–500), eat google (#2, 51–100), eat google seo (#1, 11–50)

Ian Booth covers the three pillars of E-A-T and shares tips on how to incorporate each into your content strategy so that you can rank for the best search terms in your industry.

12. 10 Basic SEO Tips to Index + Rank New Content Faster - Whiteboard Friday

Cyrus Shepard, May 17th

24,463 reads

Top keywords according to Keyword Explorer: how to index a link faster (#2, 11–50), blog seo index (#1, 0–10),  fast on-demand seo (#2, 0–10)

When you publish new content, you want users to find it ranking in search results as fast as possible. Fortunately, there are a number of tips and tricks in the SEO toolbox to help you accomplish this goal. Sit back, turn up your volume, and let the Cyrus Shepard show you exactly how in this episode of Whiteboard Friday.

13. Page Speed Optimization: Metrics, Tools, and How to Improve - Whiteboard Friday

Britney Muller, February 1st

24,265 reads

Top keywords according to Keyword Explorer: page speed optimization (#1, 51–100),  page speed metrics (#3, 11–50), optimize page speed (#1, 0–10)

What are the most crucial things to understand about your site's page speed, and how can you begin to improve? In this edition of Whiteboard Friday, Britney Muller goes over what you need to know to get started.

14. How Google's Nofollow, Sponsored, & UGC Links Impact SEO

Cyrus Shepard, September 10th

24,262 reads

Top keywords according to Keyword Explorer:  how to send my publishers no follow links (#1, 0–10), moz nofollow links (#2, 0–10), rel= sponsored (#2, 0–10)

Google shook up the SEO world by announcing big changes to how publishers should mark nofollow links. The changes — while beneficial to help Google understand the web — nonetheless caused confusion and raised a number of questions. We've got the answers to many of your questions here.

15. How to Identify and Tackle Keyword Cannibalization in 2019

Samuel Mangialavori, February 11th

21,871 reads

Top keywords according to Keyword Explorer: keyword cannibalization (#2, 201–500), ahrefs keyword cannibalization (#3, 11–50), what is keyword cannibalization (#3, 11–50)

Keyword cannibalization is an underrated but significant problem, especially for sites that have been running for several years and end up having lots of pages. In this article, learn how to find and fix keyword cannibalization before it impacts your SEO opportunities.

16. How Bad Was Google's Deindexing Bug?

Dr. Pete, April 11th

17,831 reads

Top keywords according to Keyword Explorer: google de-indexing again (#2, 11–50), google index bug (#3, 11–50)

On Friday, April 5, Google confirmed a bug that was causing pages to be deindexed. Our analysis suggests that roughly 4% of stable URLs fell out of page-1 rankings on April 5, and that deindexing impacted a wide variety of websites.

17. What Is BERT? - Whiteboard Friday

Britney Muller, November 8th

16,797 reads

Top keywords according to Keyword Explorer: what is bert (#2, 11–50), moz wbf (#2, 0–10)

There's a lot of hype and misinformation about the newest Google algorithm update. What actually is BERT, how does it work, and why does it matter to our work as SEOs? Join our own machine learning and natural language processing expert Britney Muller as she breaks down exactly what BERT is and what it means for the search industry.

18. How Do I Improve My Domain Authority (DA)?

Dr. Pete, April 17th

16,478 reads

Top keywords according to Keyword Explorer: how to build domain authority (#2, 501–850), how to increase domain authority (#2, 501–850), how to improve domain authority (#1, 11–50)

Written to help research and inform his MozCon 2019 talk, this article by Dr. Pete covers how and why to improve a Domain Authority score.

19. How to Get Into Google News - Whiteboard Friday

Barry Adams, January 11th

16,265 reads

Top keywords according to Keyword Explorer: how to get on google news (#3, 101–200), google news inclusion (#3, 51–100), getting into google news (#3, 11–50)

How do you increase your chances of getting your content into Google News? Barry Adams shares the absolute requirements and the nice-to-have extras that can increase your chances of appearing in the much-coveted news carousel.

20. Topical SEO: 7 Concepts of Link Relevance & Google Rankings

Cyrus Shepard, April 1st

15,579 reads

Top keywords according to Keyword Explorer: link relevance (#2, 0–10), read more on seo (#2, 0–10),relevant links (#2, 0–10)

To rank in Google, it’s not simply the number of votes you receive from popular pages, but the relevance and authority of those links as well.

21. The 5 SEO Recommendations That Matter in the End

Paola Didone, March 26th

13,879 reads

Top keywords according to Keyword Explorer: seo recommendations (#1, 11–50), 10 seo recommend (#1, 0–10), seo recommendations report (#1, 0–10)

What are the most steadfast, evergreen SEO recommendations you can make for your clients? These are the top five that this SEO has encountered that consistently deliver positive results.

22. An SEO’s Guide to Writing Structured Data (JSON-LD)

Brian Gorman, May 9th

13,862 reads

Top keywords according to Keyword Explorer: json structured data (#3, 0–10), seo json content (#3, 0–10), seomoz structured data (#3, 0–10)

This guide will help you understand JSON-LD and structured data markup. Go beyond the online generators and prepare your web pages for the future of search!

23. A Comprehensive Analysis of the New Domain Authority

Russ Jones, March 5th

13,333 reads

Top keywords according to Keyword Explorer: does post clustering build domain authority (#2, 11–50), who invented domain authority (#3, 11–50), domain authority curve (#1, 0–10)

A statistical look at Moz's much-improved Domain Authority. Find out how it performs vs previous versions of Domain Authority, competitor metrics, and more.

24. The Practical Guide to Finding Anyone's Email Address

David Farkas, November 26th

13,263 reads

Top keywords according to Keyword Explorer: N/A in positions #1–3

The never-ending struggle with link building begins with finding contact info. David Farkas outlines a few simple and easy ways to discover the right person to reach out to, plus some tips on which tools and strategies work best.

25. How to Use Domain Authority 2.0 for SEO - Whiteboard Friday

Cyrus Shepard, March 8th

12,940 reads

Top keywords according to Keyword Explorer: domain authority 2.0 (#2, 11–50), thought domain authority keywords (#1, 0–10), domain authority for seo (#2, 0–10)

Domain Authority is a well-known metric throughout the SEO industry, but what exactly is the right way to use it? In this Whiteboard Friday, Cyrus Shepard explains what's new with the new Domain Authority 2.0 update and how to best harness its power for your own SEO success.


That's a wrap for the top posts of 2019! Did we miss any that were on your own must-read list? Let us know in the comments below. We can't wait to see what 2020 has in store!


Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!

Monday, December 23, 2019

Three Cheers for Holiday Lighting!

Never has there been a better time to enliven darkness with celebratory holiday lighting. Nationally and globally, the human condition is improving. And great progress has been made with each of the most affordable, plentiful, reliable energies to meet present and future needs: natural gas, oil, and coal. These are the dense energies, the mineral energies formed over the ages, which consumers voluntary choose and which require no taxpayer subsidy.

Holiday lighting was once just the wealthy putting wax candle holders on their Christmas trees. The first electric lights in the U.S. were displayed by an executive of the Edison Electric Light Company in 1882, and a new tradition was born. By the mid-twentieth century, strings of electric lights became commonplace.

A Positive Externality

Environmentalists critical of electrified America must have mixed emotions this time of the year. It may be the season of good cheer and goodwill toward all, but it is also the time of the most conspicuous energy consumption. America the Beautiful is at her best when billions of strung lights turn nights into magnificent glory, border to border, sea to shining sea.

Holiday lighting is a wondrous social offering—a positive externality in the jargon of economics—given by many to all.

Are we running out of energy? Is energy less secure? Is air or water pollution increasing?

The answer across the board is no. Today’s energy comes from an ever-growing capital stock thanks to resourceship in an incentivized economy. Record growth of oil and gas production and reserves in the U.S., as recently documented in Hard Facts, is just part of a happy story about how so-called depletable resources can increase faster than they are consumed.

Not only production, the statistics of price, usage, and reliability indicate that consumers and the general economy are well served as fossil fuels enter into its fourth century.

Hooey the Humbugs

Dense, reliable, consumer-preferred, taxpayer-neutral energies are an engine of economic betterment, which is opposed by deep ecologists. And so celebratory lighting comes in for criticism.

“With the holiday season, the intensity of light pollution only increases,” said a representative of the United Green Alliance.

Ideally, there would be a public outcry against bigger causes of light pollution, such as street lamps and tall buildings. But until that happens, the simplest thing that one can do is to not put up any Christmas or Hanukkah lights.

Another Scrooge wrote in the HuffPost:

Do you love those displays of Christmas (or Hanukah or Kwanza or …) lights? Are you awed by those so impassioned that they string up 1000s of lights in awesome displays worthy of a city center? I once did, pausing on cold winter nights, white clouds issuing from my mouth, enjoying being in the glow of beautiful displays…. But … no longer … I’ve reached the point of feeling like a Scrooge; feeling outrage over the tons of C02 going into the atmosphere via neighbors’ 10,000 light displays rather than feeling ‘joyous’.”

These cries for darkness are fringe and misplaced. Traditional pollutants associated with fossil-fuel burning have been declining for decades with more improvement on the way. Carbon dioxide emissions, on the other hand, while increasing, are greening Planet Earth, as exhaustively documented by the CO2 Coalition. The climate scare more generally is exaggerated and misunderstood given that fossil fuels have made the world safer, not more dangerous, over recorded history.

Conclusion

The climate nannies at the New York Times, while urging less consumption of meat and many other lifestyle alterations to reduce greenhouse gas emissions, recommend being “smart” about holiday lights. Replace your incandescent bulbs with LED ones. Turn off the lights after midnight. And “promptly” take them down “when the holidays are over.”

Instead, how about a “thank you” and encouragement of holiday lighting? The displays can be left on for the whole night—and up into the New Year.

Here’s to ever greater holiday celebrations in the years ahead in the energy superpower, the United States of America, and around the world.

The post Three Cheers for Holiday Lighting! appeared first on IER.

Friday, December 20, 2019

Bright Bulb Award: Administration Shuts Off Unnecessary Incandescent Light Bulb Ruling

WASHINGTON DC (December 20, 2019) – The Institute for Energy Research (IER) applauded a ruling from the Department of Energy (DOE) today concerning General Service Incandescent Lamps, also known to average consumers and business owners as incandescent light bulbs.

DOE announced today it would maintain the current standards for incandescent light bulbs. As indicated in their ruling, DOE affirmed it is not allowed to increase energy efficiency standards unless doing so is economically justified – and in this case, DOE determined the costs outweigh the benefits.

According to DOE’s analysis, more stringent standards would raise the price of incandescent bulbs by 300 percent and the bulbs wouldn’t last long enough for consumers to recoup the higher, upfront cost. This may make incandescent light bulbs so expensive that no one would buy them, and potentially no one would manufacture them for U.S. consumers or businesses going forward. More stringent standards would effectively regulate incandescent light bulbs out of existence.

“This Administration will stop at nothing in its effort to make America great again, and that includes making light bulbs great again. The Department of Energy studied this issue extensively, determined the cost outweighed the benefits, and turned the lights out on a regulation that simply isn’t needed – what a refreshing concept coming out of Washington.”

“This is about following the rule of law and making a making a decision based on common sense and allowing consumers a choice. The ruling does not prohibit any cost-effective, innovative solutions for lighting, but it doesn’t outlaw the standard incandescent light bulb, either. When competition exists, free markets thrive and Americans win, every time. I applaud this rule because it puts innovation over government intervention in the marketplace.”

For media inquiries, please contact Jon Haubert
jon@hblegacy.com
303.396.5996

 

###

The post Bright Bulb Award: Administration Shuts Off Unnecessary Incandescent Light Bulb Ruling appeared first on IER.

IER Releases Hard Facts: An Energy Primer

WASHINGTON DC (December 20, 2019) – Today, the Institute for Energy Research (IER) released its fourth edition of Hard Facts: An Energy Primer. The report provides an in-depth analysis of America’s energy resources and its significance to the economy. Like in previous editions, IER aims to ground policy conversations with facts and counter misconceptions about our domestic energy resources.

As outlined in the introduction of the latest edition of Hard Facts, nothing is more fundamental to unlocking human creativity and our full potential than the availability of affordable and reliable energy. Energy heats our homes, fuels our transportation, and powers the technology that contributes to our overall well-being. To put it simply, affordable energy enriches our lives and those of our loved ones, neighbors, friends, and fellow citizens by allowing us to do more.

Thomas Pyle, President of the Institute of Energy Research (IER), released the following statement along with the latest issue of Hard Facts:

“Too few people in America, be it policymaker, journalist, or everyday citizen, fully grasp the nation’s true energy story. Fewer still fail to understand the significance that affordable energy plays in our society and all the incredible benefits it delivers. With our country at a crossroads in terms of energy policy, Hard Facts sets the record straight.”

Energy has been the driving force behind the massive improvements in human living standards that began nearly 200 years ago. Affordable, reliable energy is the foundation of a free and prosperous society because it is essential to everything that makes progress and opportunity possible.

Hard Facts: An Energy Primer is a genuine reference tool for the news reporter, the policymaker, and everyday citizens. Key takeaways include:

  • At the end of 2017, the U.S. had slightly more than 39.1 billion barrels of proved crude oil reserves—84 percent higher than 10 years ago.
  • U.S. oil production in 2018 averaged almost 11 million barrels per day, with production in November and December hitting almost 12 million barrels per day. Oil production in 2018 was 1.6 million barrels per day higher than in 2017 (a 17 percent increase).
  • In the past decade, U.S. companies have drilled 114,000 wells in the Permian Basin of Texas alone; many of them are profitable at crude oil prices as low as $30 a barrel.
  • The renewable energy sector, including hydro, was responsible for 17 percent of the net electricity generated in the U.S. in 2018.
  • In 2018, the U.S. produced an average of 101.3 billion cubic feet of natural gas per day, the highest volume on record.
  • In 2018, U.S. coal exports rose for the second consecutive year to 116 million short tons.
  • In 2018, the U.S. nuclear power fleet generated its highest capacity factor ever at 92.6 percent.
  • In 2018, the renewable energy industry experienced an unexpected flattening of growth as renewables added about the same net capacity worldwide in 2018 as they did in 2017.

___________________

Read the full edition of Hard Facts: An Energy Primer here.

___________________

For media inquiries, please contact Jon Haubert
jon@hblegacy.com
303.396.5996

 

###

The post IER Releases Hard Facts: An Energy Primer appeared first on IER.

Should We Trust the Climate Models? Part 1 of 3

As an economist who writes on climate change policy, my usual stance is to stipulate the so-called “consensus” physical science (as codified for example in UN and US government reports), and show how the calls for massive carbon taxes and direct regulations still do not follow. For an example of this discrepancy that I often mention, William Nordhaus recently won the Nobel Prize for his work on climate change economics, and he calibrated his model to match what the UN’s IPCC said about various facts of physical science, such as the global temperature’s sensitivity to greenhouse gas emissions, etc. Even so, Nordhaus’ Nobel-winning work shows that governments doing nothing would be better for human welfare than trying to hit the UN’s latest goal of limiting warming to 1.5 degrees Celsius.

However, even though extreme government measures wouldn’t be justified even if we had complete faith in the standard computer models used to simulate the global climate, the fact is that we shouldn’t put too much faith into such models. You sometimes see establishment outlets slip up and let this truth out. For example, as I documented here at IER at the time, last summer the New York Times ran an op-ed from a physicist who said we needed faster computers to run the climate projections, because, with our current state of knowledge, climate change might turn out to be “a mere annoyance” (her actual term).

In this and two following posts, I will provide evidence from expert sources to demonstrate that the popular computer projections of future climate change are far less reliable than the public has been led to believe. Americans should better understand the limits of the current climate models before allowing political officials to impose massive taxes and regulations that we know will raise energy prices.

An Analogy With Flipping a Coin

In this first post, I am ultimately going to present a chart from a mainstream scientist who endorses the “consensus” approach to climate change policy in order to show just how poorly the current batch of climate models has performed in recent years. But in order for the average reader to understand what the chart is showing, it will help if we first consider a hypothetical scenario involving a coin toss. (I urge the reader to trust me on this one; the silly story I’m about to spin will be a low-impact way for me to get across the statistical concept that you’ll need to know, in order to understand the climate projection chart at the end of the article.)

Suppose you’re walking through a carnival and a man at one of the booths grabs your attention. “Hey there, have I got a great opportunity for you! Here in my hand is a regular quarter. I’ll let you flip it 100 times, and we’ll both keep track of how many ‘Heads’ turn up. When you’re all done flipping, if the total number of Heads that turned up is 40 or higher, then I’ll pay you $100. However, if the total number of Heads that turns up is 39 or lower, then you pay me $100. How does that grab you?”

Now at first blush, this sounds like a great wager. With a fair coin, when you flip it 100 times, you would expect to see Heads about half the time, i.e. 50 of them. And so if the carnival showman is going to pay you so long as you see at least 40 Heads, while you only lose if you see 39 or fewer Heads, then this seems to be a pretty advantageous bet. So long as it’s a fair coin, the fact that you are going to flip it 100 times really helps solidify your edge: Vaguely remembering your high school stats class, you pull out your phone and go to a website that calculates such things, and you realize that the chance of a fair coin coming up Heads at least 40 times out of 100 flips is a whopping 98%.

Ah but of course, there’s a difference between being smart and being wise. Your smarts tell you that this is a great wager, but your wisdom tells you the carnival guy isn’t holding a fair coin. You politely decline his offer and start to walk away.

“Wait a second!” he exclaims. “I know you don’t trust my coin. But before we do the wager, I’ll let you test it! You can flip it 25 times as a test run, and see whether you think it’s a fair coin. Then—but only if you want—we can go ahead and flip it another 100 times for the wager.”

This new twist stops you in your tracks. What would it hurt to at least flip the coin 25 times and see what happens? But before you begin the test, the man gives you a quick statistics refresher:

“Now I don’t mean to patronize you,” he begins, “but I just don’t want there to be any hard feelings or accusations of dishonesty. I know this is a fair coin, but I realize that you are skeptical. So we’re going to treat these 25 flips as a ‘hypothesis test,’ and specifically what we’re testing is the hypothesis that I am indeed giving you a fair coin. Now let’s think this through for a moment. Even if I’m telling you the truth, and this really is a fair coin, it is possible that the first flip comes up Tails, right?”

You nod your head. “Sure,” you say. “I mean, there’s a 50% chance it comes up Tails on the first flip.”

“That’s right!” the man says. “And let’s keep going. Even if I’m telling you the truth, and this is a fair coin, it’s entirely possible that during this test, it comes up Tails on the first and the second flip, isn’t it?”

“Sure,” you agree. “There’s actually a 25% chance that we’d see two Tails in a row, even if it’s a fair coin.”

“Great,” the man continues. “And let’s push it again, in fact it’s possible you will see three Tails in a row?”

“Yeeeeeeeah,” you say, now getting a little hesitant. “But at this point, we’re down to a one-eighth probability, or 12.5% chance, of seeing all Tails on the first three flips.”

“I know, I know, don’t worry, just making a point here,” the man says, trying to reassure you. “Now you look like an educated man,” he says. “Did you ever take a statistics class in school? Do you remember what the convention is for what they called a ‘confidence interval’ when doing a hypothesis test?”

You vaguely remember the details, but can’t quite pull them up. After all, this was a dark period in your life that you would rather put behind you.

“Let me jog your memory,” the carnival man continues. “If the hypothesis we’re testing is whether this coin is fair, then let’s use a 95% confidence interval—this is standard stuff in statistics. So we want to know how many Heads/Tails we would need to see, after a given number of flips of the coin, for us to be pretty sure that the coin isn’t fair.”

The man then pulls out a piece of paper with the following diagram:[1]

Figure 1. Coin Tossing Experiment, Initial Setup

Coin Toss Graph

The man explains, “This diagram shows that at any given time, the expected number of Heads is equal to one-half of the total number of flips you’ve already done by that point. For example, after 8 flips, you would have expected to see a total of 4 Heads, which is why the black line goes through the point (8,4). Now the gray region is what statisticians call the ‘95% envelope’ surrounding the mean. So long as the actual observations of Heads fall within that envelope, then we can’t reject the hypothesis that it’s a fair coin. In contrast, if the observations did end up outside that shaded region, then we would be able to say, ‘There’s less than a 5% chance that these results could’ve occurred with a fair coin, so we’re pretty sure it’s not a fair coin.’ OK? Ready to take my coin for a test run?”

All of his remarks seem consistent with what you remember from your stats class. So you go ahead and flip the coin 25 times. He hands you a red Sharpie for you to mark the results on the paper as they occur. Specifically, you observe a Head on the 4th, 6th, 13th, 14th, 16th, 17th, 19th, 20th, and 21st tosses. The chart ends up looking like this:

Figure 2. Coin Tossing Experiment, With Results

Another coin Toss Graph

At this point, your Spidey Sense is tingling. During the trial, the coin was definitely “Head shy,” consistently falling below the expected number of Heads. Indeed, if you had stopped the trial after the 12th toss, the number of observed Heads at that point was so low that it fell outside the 95% envelope, and the red line hovered near that boundary for most of the other flips as well.

The carnival man can see your anxiety, and tries to reassure you by arguing, “What’s the problem? We have every reason to trust that this is a fair coin. Look: its performance has been well within the 95% spread.”

As you ponder the chart, you realize something: To say that the observed number of Heads “falls within the 95% envelope” does not mean, “We are 95% confident that this is a fair coin.” Rather, what it’s really saying is that if this were a fair coin, then the probability of it generating such a small number of Heads is at least 2.5%. (Note the decimal point: that’s two-and-a-half percent.) That is entirely a different claim, and one that—by itself—doesn’t inspire much confidence at all!

Now as I wrap up this tale, I don’t want to put the final words in your mouth. I just ask you, dear reader, whether you would take the carnival bet based on the above graph?

Now Looking at Climate Projections

The reason I walked through the long story above is that it illustrates a crucial concept: a 95% envelope or “spread,” showing how far observations can stray from the mean “expectation” value without us definitely rejecting the underlying model.

In the case of climate change, we aren’t testing whether a coin is fair, instead, we are testing whether a suite (or “ensemble”) of leading climate models gave accurate predictions of global temperature. So we use the computer models to generate their (averaged) mean projections of temperature, along with a 95% spread around the mean in order to account for the internal variability of the projections. (Even if you run the same computer simulation over and over with the same parameter settings, you will get different outputs.)

Then, after time passes, we can look at actual observations of global temperature, and plot them against the predictions. The following chart is taken from the RealClimate blog, which is one of the leading outposts giving “real climate scientists” (their term) a platform to give accurate information to the public. I stress that these folks are definitely part of the climate change orthodoxy, and are by no means “deniers” or even “skeptics.” With that in mind, look at the following chart that they themselves posted, in an evaluation of the climate models:

Figure 3. Climate Model Projections vs. Surface Observations, as of Feb. 2019

Temp Anomalies
Source: RealClimate

In Figure 3, the dark black line shows the mean global temperature prediction from an ensemble of leading climate models. (CMIP5 stands for Coupled Model Intercomparison Project, Phase 5; details are here.) The gray band is the 95% spread around the mean, accounting for the internal variability in the models.

The colored lines represent different series of global temperature observations. For example, the dark blue line represents HadCRUT4, the temperature dataset compiled by the Climatic Research Unit (CRUT) of the University of East Anglia, while the dark red line is GISTEMP, a global surface temperature dataset maintained by NASA.

Note that up to 2005, the models are giving a “hindcast”—meaning they were calibrated after these observations had been made. So the fact that the black line goes through the middle (roughly speaking) of the observations up through 2005 isn’t evidence of the models’ effectiveness.

The only true forecasts were made by these models from 2005 onward. And as we see, the observations are consistently below the mean of the model projections, with only one year on one dataset—namely, GISTEMP in 2016—actually hitting the model mean.

Furthermore, many of the other observations flirt with the lower edge of the 95% envelope, meaning that on those occasions, it would be correct to say, “If the climate models were a correct simulation of the actual climate, natural variation would give us such low observed temperatures just barely 2.5% of the time.”

Now if the surface temperature datasets are iffy, it’s really not even close with the satellite observations. The next figure is again taken from RealClimate:

Figure 4. Climate Model Projections vs. Satellite Observations, as of Feb. 2019

Projections vs. Satellite Data
Source: RealClimate

In Figure 4, we see that once we move into the genuine Forecast period, not a single data point in any of the three satellite series ever hit the mean projection of global temperature. Furthermore, as of the most recent observation, two out of three of the datasets lie below the 95% spread—and one of these series (UAH v6.0) spent much of the entire forecast period below the spread.

If the above results don’t make the establishment climate scientists admit to the public, “Yep, our models have been running hot, making projections of temperature increases that haven’t been observed,” then what would?

To repeat, Figures 3 and 4 are not taken from the Heartland Institute or Richard Lindzen’s website. Instead, they are taken from the people who think those figures vindicate the orthodox perspective on climate change. I invite the reader to peruse the relevant blog posts (here and here for an earlier, fuller discussion), where the RealClimate blogger interprets these charts. There really isn’t much commentary at all on the awkward satellite observations, and regarding the surface datasets, we are told that, “The current temperatures are well within the model envelope.”

Conclusion

In this post, I developed a long but hopefully intuitive fable about a carnival wager involving coin tosses. In that context, we saw quite clearly what it means for observations to “fall within the 95% spread,” even if they are flirting with its lower boundary.

There is an analogous situation with respect to the climate models. A few vocal scientists and many activists are pressuring political officials and the public to support massive tax hikes and new regulations in order to mitigate the threat of catastrophic climate change. Yet one of the crucial planks in their argument—namely, that we can trust the alarming computer simulations of global warming—rely on models that have clearly been overpredicting global temperature. The fact that the defenders of the orthodoxy won’t even acknowledge their own damning charts should give any honest citizen pause.

_________________________________

[1] Thanks to Derrick Reiner for helping me construct the two graphs illustrating the coin-flipping example.

The post Should We Trust the Climate Models? Part 1 of 3 appeared first on IER.

Actually Accurate Analytics - Whiteboard Friday

Posted by RuthBurrReedy

Clean, useful Google Analytics data is all-important — both for you, and for the clients and colleagues that will be working on the site in the future. Ruth Burr Reedy shares her absolute best tips for getting your Analytics data accurate, consistent, and future-proof in this week's Whiteboard Friday.

Click on the whiteboard image above to open a high resolution version in a new tab!

Video Transcription

Hi, Moz fans. I'm Ruth Burr Reedy, and I am the Vice President of Strategy at UpBuild. We're a technical marketing agency specializing in technical SEO and advanced web analytics. One of the things I wanted to talk about today, Whiteboard Friday, is about analytics.

So when I talk to SEOs about analytics and ask them, "When it comes to analytics, what do you do? What do you do first? When you're taking on a new client, what do you do?" SEOs are often really eager to tell me, "I dive into the data. Here's what I look like.Here are the views that I set up. Here's how I filter things. Here's where I go to gain insights."

But what I often don't hear people talk about, that I think is a super important first step with a new client or a new Analytics account, or really any time if you haven't done it, is making sure your Analytics data is accurate and consistent. Taking the time to do some basic Analytics housekeeping is going to serve you so far into the future and even beyond your time at that given client or company.

The people who come after you will be so, so, so thankful that you did these things. So today we're going to talk about actually accurate analytics. 

Is your Analytics code on every page?

So the first question that you should ask yourself is: Is your Analytics code on every page? Is it?

Are you sure? There are a lot of different things that can contribute to your Analytics code not actually being on every single page of your website. One of them is if portions of your site have a different CMS from the main CMS that's driving your site. 

Forums, subdomains, landing pages

We see this a lot with things like subdomains, with things like forums. A really common culprit is if you're using a tool like Marketo or HubSpot or Unbounce to build landing pages, it's really easy to forget to put Analytics on those pages.

Over time those pages are out there in the world. Maybe it's just one or two pages. You're not seeing them in Analytics at all, which means you're probably not thinking about them, especially if they're old. But that doesn't mean that they don't still exist and that they aren't still getting views and visits. 

Find orphan pages

So, okay, how do we know about these pages? Well, before you do anything, it's important to remember that, because of the existence of orphan pages, you can't only rely on a tool like Screaming Frog or DeepCrawl to do a crawl of your site and make sure that code is on every page, because if the crawler can't reach the page and your code is not on the page, it's kind of in an unseeable, shrouded in mystery area and we don't want that.

Export all pages

The best way, the most sure way to make sure that you are finding every page is to go to your dev team, to go to your developers and ask them to give you an export of every single URL in your database. If you're using WordPress, there's actually a really simple tool you can use. It's called Export All URLs in the grand tradition of very specifically named WordPress tools.

But depending on your CMS and how your site is set up, this is something that you can almost certainly do. I need a list of every single URL on the website, every single URL in our database. Your dev team can almost certainly do this. When you get this, what you can do, you could, if you wanted, simply load that list of URLs. You'd want to filter out things like images and make sure you're just looking at the HTML documents.

Dedupe with Screaming Frog

Once you had that, you could load that whole thing into Screaming Frog as a list. That would take a while. What you could do instead, if you wanted, is run a Screaming Frog crawl and then dedupe that with Screaming Frog. So now you've got a list of your orphan pages, and then you've got a list of all of the pages that Screaming Frog can find. So now we have a list of every single page on the website.

We can use either a combination of crawler and list or just the list, depending on how you want to do it, to run the following custom search. 

What to do in Screaming Frog

Configuration > Custom > Search

So in Screaming Frog, what you can do is you can go to Configuration and then you go to Custom Search. It will pop up a custom search field. What this will allow you to do is while the crawler is crawling, it will search for a given piece of information on a page and then fill that in a custom field within the crawler so that you can then go back and look at all of the pages that have this piece of information.

What I like to do when I'm looking for Analytics information is set up two filters actually — one for all of the pages that contain my UA identifier and one for all of the pages that don't contain it. Because if I just have a list of all the pages that contain it, I still don't know which pages don't contain it. So you can do this with your unique Google Analytics identifier.



If you're deploying Google Analytics through Google Tag Manager, instead you would look for your GTM Number, your GTM ID. So it just depends how you've implemented Analytics. You're going to be looking for one of those two numbers. Almost every website I've worked on has at least a few pages that don't have Analytics on them.

What you'll sometimes also find is that there are pages that have the code or that should have the code on them, but that still aren't being picked up. So if you start seeing these errors as you're crawling, you can use a tool like Tag Assistant to go in and see, "Okay, why isn't this actually sending information back to Google Analytics?" So that's the best way to make sure that you have code on every single page. 

Is your code in the <head> and as high as possible?

The other thing you want to take a look at is whether or not your Analytics code is in the head of every page and as close to the top of the head as possible. Now I know some of you are thinking like, "Yeah, that's Analytics implementation 101." But when you're implementing Analytics, especially if you're doing so via a plug-in or via GTM, and, of course, if you're doing it via GTM, the implementation rules for that are a little bit different, but it's really easy for over time, especially if your site is old, other things to get added to the head by other people who aren't you and to push that code down.

Now that's not necessarily the end of the world. If it's going to be very difficult or time-consuming or expensive to fix, you may decide it's not worth your time if everything seems like it's firing correctly. But the farther down that code gets pushed, the higher the likelihood that something is going to go wrong, that something is going to fire before the tracker that the tracker is not going to pick up, that something is going to fire that's going to prevent the tracker from firing.

It could be a lot of different things, and that's why the best practice is to have it as high up in the head as possible. Again, whether or not you want to fix that is up to you. 

Update your settings:

Once you've gotten your code firing correctly on every single page of your website, I like to go into Google Analytics and change a few basic settings. 

1. Site Speed Sample Rate

The first one is the Site Speed Sample Rate.

So this is when you're running site speed reports in Google Analytics. Typically they're not giving you site timings or page timings for the site as a whole because that's a lot of data. It's more data than GA really wants to store, especially in the free version of the tool. So instead they use a sample, a sample set of pages to give you page timings. I think typically it's around 1%.

That can be a very, very small sample if you don't have a lot of traffic. It can become so small that the sample size is skewed and it's not relevant. So I usually like to bump up that sample size to more like 10%. Don't do 100%. That's more data than you need. But bump it up to a number that's high enough that you're going to get relevant data.

2. Session and Campaign Timeout

The other thing that I like to take a look at when I first get my hands on a GA account is the Session and Campaign Timeout. So session timeout is basically how long somebody would have to stay on your website before their first session is over and now they've begun a new session if they come back and do something on your site where now they're not being registered as part of their original visit.

Historically, GA automatically determined session timeout at 30 minutes. But this is a world where people have a million tabs open. I bet you right now are watching this video in one of a million tabs. The longer you have a tab open, the more likely it is that your session will time out. So I like to increase that timeout to at least 60 minutes.

The other thing that Google automatically does is set a campaign timeout. So if you're using UTM parameters to do campaign tracking, Google will automatically set that campaign timeout at six months. So six months after somebody first clicks that UTM parameter, if they come back, they're no longer considered part of that same campaign.

They're now a new, fresh user. Your customer lifecycle might not be six months. If you're like a B2B or a SaaS company, sometimes your customer lifecycle can be two years. Sometimes if you're like an e-com company, six months is a really long time and you only need 30 days. Whatever your actual customer lifecycle is, you can set your campaign timeout to reflect that.

I know very few people who are actually going to make that window shorter. But you can certainly make that longer to reflect the actual lifecycle of your customers. 

3. Annotations

Then the third thing that I like to do when I go into a Google Analytics account is annotate what I can. I know a lot of SEOs, when you first get into a GA account, you're like, "Well, no one has been annotating.Ho-hum. I guess going forward, as of today, we're going to annotate changes going forward."

That's great. You should definitely be annotating changes. However, you can also take a look at overall traffic trends and do what you can to ask your coworkers or your client or whatever your relationship is to this account, "What happened here?" Do you remember what happened here? Can I get a timeline of major events in the company, major product releases, press releases, coverage in the press?

Things that might have driven traffic or seen a spike in traffic, product launches. You can annotate those things historically going back in time. Just because you weren't there doesn't mean it didn't happen. All right. So our data is complete. It's being collected the way that we want to, and we're tracking what's happening.

Account setup

Cool. Now let's talk about account setup. I have found that many, many people do not take the time to be intentional and deliberate when it comes to how they set up their Google Analytics account. It's something that just kind of happens organically over time. A lot of people are constrained by defaults. They don't really get what they're doing.

What we can do, even if this is not a brand-new GA account, is try to impose some structure, order, consistency, and especially some clarity, not only for ourselves as marketers, but for anybody else who might be using this GA account either now or in the future. So starting out with just your basic GA structure, you start with your account.

Your Account Name is usually just your company name. It doesn't totally matter what your Account Name is. However, if you're working with a vendor, I know they'd prefer that it be your company name as opposed to something random that only makes sense to you internally, because that's going to make it easier for them. But if you don't care about that, you could conceivably name your account whatever you want. Most of the time it is your company name.

Then you've got your property, and you might have various properties. A good rule of thumb is that you should have one property per website or per group of sites with the same experience. So if you have one experience that goes on and off of a subdomain, maybe you have mysite.com and then you also have store.mysite.com, but as far as the user experience is concerned it's one website, that could be one property.

That's kind of where you want to delineate properties is based on site experiences. Then drilling down to views, you can have as many views as you want. When it comes to naming views, the convention that I like to use is to have the site or section name that you're tracking in that specific view and then information about how that view is set up and how it's intending to be used.

Don't assume that you're going to remember what you were doing last year a year from now. Write it down. Make it clear. Make it easy for people who aren't you to use. You can have as many views as you want. You can set up views for very small sections of your site, for very specific and weird filters if there are some customizations you want to do. You can set up as many views as you need to use.

Must-have views

1. Raw data - Unfiltered, Don't Touch

But I think there are three views that you should make sure you have. The first is a Raw Data view. This is a view with no filters on it at all. If you don't already have one of these, then all of your data in the past is suspect. Having a view that is completely raw and unfiltered means if you do something to mess up the filtering on all your other views, you at least have one source of total raw data.

I know this is not new information for SEOs when it comes to GA account setup, but so many people don't do it. I know this because I go into your accounts and I see that you don't have it. If you don't have it, set it up right now. Pause this video. Go set it up right now and then come back and watch the rest, because it's going to be good. In addition to naming it "Raw Data Unfiltered," I like to also add something like "Don't Touch" or "For Historical Purposes Only," if you're not into the whole brevity thing, something that makes it really clear that not only is this the raw data, but also no one should touch it.

This is not the data we're using. This is not the data we're make decisions by. This is just our backup. This is our backup data. Don't touch it. 

2. Primary view - Filtered, Use This One

Then you're going to want to have your Primary view. So however many views you as a marketer set up, there are going to be other people in your organization who just kind of want the data.

So pick a view that's your primary filtered view. You're going to have a lot of your basic filters on this, things like filtering out your internal IP range, filtering out known bots. You might set up some filtering to capture the full hostname if you're tracking between subdomains, things like that. But it's your primary view with basic filtering. You're going to want to name that something like "Use This One."

Sometimes if there's like one person and they won't stop touching your raw data, you can even say like, "Nicole Use This One." Whatever you need to label it so that even if you got sick and were in the hospital and unreachable, you won the lottery, you're on an island, no one can reach you, people can still say, "Which of these 17 views that are set up should I use? Oh, perhaps it's the one called 'Use This One.'" It's a clue. 

3. Test view - Unfiltered

Then I like to always have at least one view that is a Test view. That's usually unfiltered in its base state. But it's where I might test out filters or custom dimensions or other things that I'm not ready to roll out to the primary view. You may have additional views on top of those, but those are the three that, in my opinion, you absolutely need to have.

4. All Website Data

What you should not have is a view called "All Website Data." "All Website Data" is what Google will automatically call a view when you're first setting up GA. A lot of times people don't change that as they're setting up their Analytics. The problem with that is that "All Website Data" means different things to different people. For some people, "All Website Data" means the raw data.

For some people, "All Website Data" means that this is the "Use This One" view. It's unclear. If I get into a GA account and I see that there is a view named "All Website Data," I know that this company has not thought about how they're setting up views and how they're communicating that internally. Likely there's going to be some filtering on stuff that shouldn't have been filtered, some historical mishmash.

It's a sign that you haven't taken the time to do it right. In my opinion, a good SEO should never have a view called "All Website Data." All right. Great. So we've got our views set up. Everything is configured the way that we want it. How that's configured may be up to you, but we've got these basic tenets in place.

Goals

Let's talk about goals. Goals are really interesting. I don't love this about Google Analytics, but goals are forever. Once you set a goal in GA, information that is tracked to that number or that goal number within that goal set will always be tracked back to that. What that means is that say you have a goal that's "Blue Widget Sales" and you're tracking blue widget sales.

Goals are forever

Over time you discontinue the blue widget and now you're only tracking red widget sales. So you rename the "Blue Widget Sales" widget to now it's called "Red Widget Sales." The problem is renaming the goal doesn't change the goal itself. All of that historical blue widget data will still be associated with that goal. Unless you're annotating carefully, you may not have a good idea of when this goal switched from tracking one thing to be tracking another thing.

This is a huge problem when it comes to data governance and making decisions based on historical data. 

The other problem is you have a limited number of goals. So you need to be really thoughtful about how you set up your goals because they're forever. 

Set goals based on what makes you money

A basic rule is that you should set goals based on what makes you money.

You might have a lot of micro conversions. You might have things like newsletter sign-ups or white paper downloads or things like that. If those things don't make you money, you might want to track those as events instead. More on that in a minute. Whatever you're tracking as a goal should be related to how you make money. Now if you're a lead gen biz, things like white paper downloads may still be valuable enough that you want to track them as a goal.

It just depends on your business. Think about goals as money. What's the site here to do? When you think about goals, again, remember that they're forever and you don't get that many of them. 

Group goals efficiently

So any time you can group goals efficiently, take some time to think about how you're going to do that. If you have three different forms and they're all going to be scheduling a demo in some way or another, but they're different forms, is there a way that you can have one goal that's "Schedule a Demo" and then differentiate between which form it was in another way?

Say you have an event category that's "Schedule a Demo" and then you use the label to differentiate between the forms. It's one goal that you can then drill down. A classic mistake that I see with people setting up goals is they have the same goal in different places on the website and they're tracking that differently. When I say, "Hey, this is the same goal and you're tracking it in three different places," they often say, "Oh, well, that's because we want to be able to drill down into that data."

Great. You can do that in Google Analytics. You can do that via Google Analytics reporting. You can look at what URLs and what site sections people completed a given goal on. You don't have to build that into the goal. So try to group as efficiently as possible and think long term. If it at any time you're setting up a goal that you know is someday going to be part of a group of goals, try to set it up in such a way that you can add to that and then drill down into the individual reports rather than setting up new goals, because those 20 slots go quick.

Name goals clearly

The other thing you're going to want to do with goals and with everything — this is clearly the thesis for my presentation — is name them clearly. Name them things where it would be impossible not to understand exactly what it is. Don't name your goal "Download." Don't name your goal "Thank You Page."

Name your goal something specific enough that people can look at it at a glance. Even people who don't work there right now, people in the future, the future people can look at your goals and know exactly what they were. But again, name them not so specifically that you can't then encompass that goal wherever it exists on the site. So "Download" might be too broad.

"Blue Widget White Paper Download" might be too specific. "White Paper Download" might be a good middle ground there. Whatever it is for you, think about how you're going to name it in such a way that it'll make sense to somebody else, even if you don't work there anymore and they can't ask you. Now from talking about goals it kind of segues naturally into talking about events, event tracking.

Events

Event tracking is one of the best things about Google Analytics now. It used to be that to track an event you had to add code directly to a page or directly to a link. That was hard to do at scale and difficult to get implemented alongside conflicting dev possibilities. But now, with Google Tag Manager, you can track as many events as you want whenever you want to do them.

You can set them up all by yourself, which means that now you, as the marketer, as the Analytics person, become the person who is in charge of Google Analytics events. You should take that seriously, because the other side of that coin is that it's very possible to get event creep where now you're tracking way too many events and you're tracking them inefficiently and inconsistently in ways that make it difficult to extract insights from them on a macro level.

What do you want and why?

So with events, think about what you want and why. Any time somebody is like, "I want to track this," ask them, "Okay, what are we going to do with that information?" If they're like, "I don't know. I just want to know it." That might not be a good case to make to track an event. Understand what you're going to do with the data. Resist the urge to track just for tracking's sake.

Resist data for data's sake. I know it's hard, because data is cool, but try your best. 

Naming conventions

As you take over, now that you are the person in charge of events, which you are, you're taking this on, this is yours now, develop naming conventions for your events and then become the absolute arbiter of those conventions. Do not let anybody name anything unless it adheres to your conventions.

Category

Now how you name things is up to you. Some suggestions, for category, I like that to be the site section that something is in or maybe the item type. So maybe it's product pages. Maybe it's forms. Maybe it's videos. However you are going to group these events on a macro level, that should be your category.

Action

The action is the action. So that's click, submit, play, whatever the action is doing. 

Label

Then the label is where I like to get unique and make sure that I'm drilling down to just this one thing. So maybe that's where I'll have the actual CTA of the button, or which form it was that people filled out, or what product it was that they purchased. Again, think about information that you can get from other reports.

So for example, you don't need to capture the URL that the event was recorded on as part of the label, because you can actually go in and look at all of your events by URL and see where that happened without having to capture it in that way. The important thing is that you have rules, that those rules are something that you can communicate to other people, and that they would then be able to name their own categories, actions, and labels in ways that were consistent with yours.

Over time, as you do this and as you rename old events, you're going to have a more and more usable body of data. You're going to be increasingly comparing apples to apples. You're not going to have some things where Click is the action and some things where Click is the label, or things that should be in one category that are in two or three categories. Over time you're going to have a much more usable and controllable body of event data.

Be consistent

Then you need to be ruthless about consistency with usage of these naming conventions. There will be no just setting up an event real quick. Or, in fact, there will be just setting up an event real quick, but it will be using these rules that you have very thoroughly outlined and communicated to everybody, and that you are then checking up to make sure everything is still tracking the same way. A big thing to watch for when you're being ruthless about consistency is capitalization.

Capitalization in category action and label and event tracking will come back as two different things. Capital "C" and lowercase "c" category are two different things. So make sure as you're creating new events that you have some kind of standardization. Maybe it's the first letter is always capitalized. Maybe it's nothing is ever capitalized.

It doesn't matter what it is as long as it's all the same. 

Think about the future!

Then think about the future. Think about the day when you win the lottery and you move to a beautiful island in the middle of the sea and you turn off your phone and you never think about Google Analytics again and you're lying in the sand and no one who works with you now can reach you. If you never came back to work again, could the people who work there continue the tracking work that you've worked so hard to set up?

If not, work harder to make sure that's the case. Create documentation. Communicate your rules. Get everybody on the same page. Doing so will make this whole organization's data collection better, more actionable, more usable for years to come. If you do come back to work tomorrow, if in fact you work here for the next 10 years, you've just set yourself up for success for the next decade.

Congratulations. So these are the things that I like to do when I first get into a GA account. Obviously, there are a lot of other things that you can do in GA. That's why we all love GA so much. 

Homework

But to break it down and give you all some homework that you can do right now.

Check for orphan pages

Tonight, go in and check for orphan pages.

When it comes to Analytics, those might be different or they might be the same as orphan pages in the traditional sense. Make sure your code is on every page. 

Rename confusing goals and views (and remove unused ones)

Rename all your confusing stuff. Remove the views that you're not using. Turn off the goals that you're not using. Make sure everything is as up to date as possible. 

Guard your raw data

Don't let anybody touch that raw data. Rename it "Do Not Touch" and then don't touch it. 

Enforce your naming conventions

Create them. Enforce them. Protect them. They're yours now.

You are the police of naming conventions. 

Annotate everything

Annotate as much as you can. Going forward you're going to annotate all the time, because you can because you're there, but you can still go back in time and annotate. 

Remove old users

One thing that I didn't really talk about today but you should also do, when it comes to the general health of your Analytics, is go in and check who has user permissions to all of your different Analytics accounts.

Remove old users. Take a look at that once a quarter. Just it's good governance to do. 

Update sampling and timeouts

Then you're going to update your sampling and your timeouts. If you can do all of these things and check back in on them regularly, you're going to have a healthy, robust, and extremely usable Analytics ecosystem. Let me know what your favorite things to do in Analytics are. Let me know how you're tracking events in GTM.

I want to hear all about everything you all are doing in Analytics. So come holler at me in the comments. Thanks.

Video transcription by Speechpad.com


Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!