Do more SEO testing and experimenting

Olga says: “Start doing SEO tests and experiments, in a nutshell.

If you do tests, even simple ones, you gain valuable SEO experience. You will actually learn how SEO works. Even if you are a beginner, this allows you to gain that little bit of experience. It also allows you to make decisions based on evidence instead of just assumptions or guesswork.

The purpose of that should be to optimise your SEO performance by understanding, through your tests, what works, what doesn't work, and what may work for your specific website.

It is also something that allows you to manage risks. If you test on a smaller site that you don't really care about, if something goes wrong, you won't pay for it. You can be sure to only put the implementations into production that turned out well during tests.

This is also a way to stay updated. We all know that Google changes its algorithm basically every day and, by doing those tests, you can learn what Google is valuing more now than it was a week ago. Of course, it all depends very much on the scale and type of tests you are doing.”

How do you decide what to test?

“That depends on the KPIs that you have for a specific project, and what you want to achieve. You will be able to determine what it is that you want to test.

In most cases, it will be related to what you can do to increase rankings, increase traffic, increase conversions, and increase revenue. There are endless ways of doing SEO tests and experiments.

A/B testing, or split testing, is one of the ways you can test. For example, an initial test could be changing the format of your page titles on a certain number of your pages and then split testing that against your existing pages. You can do the same kind of testing with meta descriptions, content layouts, and basically any element on the page.

Of course, there are other types of tests you can do, like split URL testing. That is when you test different URLs, implement the change, and check how those URLs react to that change. There are also technical SEO tests, like whether better mobile speed for one URL really made a difference. That is one of the famous Core Web Vitals that people were optimizing like crazy, but then a lot of tests showed that they didn't really move the needle.

You can also do simple content experiments, like putting the introduction high above the fold instead of below the fold, because it was below an image before. Those types of things can bring very interesting results.

Another test would be a schema markup test, where you only add schema and nothing else. Will adding that schema change anything? Will it change rankings? Will it change traffic? It may change the presentation in SERPs, but will that different presentation lead to better CTR? I also love internal linking tests, where you change internal links and observe.

Those are just some examples of tests, and they can be very, very simple. You don't need to take a one-million-page website and do huge tests across every page. You can simply put up a basic WordPress site and just test one thing on that new test site. I recommend having a bunch of test sites so that you can test a few things simultaneously, or have a control group.”

Is there any difference between a test and an experiment?

“I was using them interchangeably, but the only real difference is that you don’t need an alternative to compare against with an experiment. You can run an experiment without a split test.”

What's your favourite software to use for testing?

“Again, it depends. To be as minimalistic about it as possible, you can simply use a spreadsheet where you list the URLs, the rankings for specific keywords, the changes you made, and the date. That can be all you have: just a simple spreadsheet. You can also use more advanced SEO testing tools that allow you to set up more advanced tests, and have a better eye on how they are being implemented, what the results are, and what the changes are. If you are using spreadsheets, you will have to monitor them on your own or with your rank trackers.

The tool I'm also using (which is not necessarily strictly for SEO testing) is Cora, which basically measures tens of thousands of factors and, for a specific query, it will show you which factors statistically have more meaning. For example, in the case of one URL, it might show you that the title is the most important factor. That basis is run against all 100 pages that rank for a specific keyword, and then you can test whether this is actually true for your site. In many cases, it will be. Essentially, you will have better grounds for what you want to test.

When you’re tracking rankings, you have to be efficient, because rankings like to fluctuate. I use a tool called SERP Volatility so that, every time I look at rankings, it runs that check 10 times. That way, I know what the position will be in 10 separate instances, so I can make a better decision regarding the results of my tests. Sometimes, the position may go up and I might assume that was thanks to my test, but it could just have been a one-time thing. How much of the time a specific keyword is on a specific position matters. Is it 100% of the time?

Of course, there are other things to keep an eye on, like Google Core updates. If your test started the day before an update, you have to take that into account as well.”

How do you determine when a test was successful? Is there a certain amount of traffic that you need or a certain percentage improvement over the initial version?

“If you carry out the same test on multiple pages, on multiple sites, and all of those instances show similar results, then you can be pretty sure that the results are reliable. You want to have a few instances of that running, including the control group and the actual pages you are testing.

You don't have to be super scientific about this – just play with it and it will start to give you a lot of meaningful insights, and a different outlook on how SEO works.”

Are SEOs testing things that they don’t need to be testing?

“That is not the biggest issue; the biggest issue is that SEOs aren't testing enough. I was that type of SEO for a long time as well. We read a lot of stuff, we see a lot of articles, and we watch a lot of videos where people claim different things – and we often just accept that those things are true and keep preaching about them without actually checking for ourselves.

Try testing anything. You can even test something that may not make much sense, like testing Core Web Vitals. There have been so many tests run that we can be pretty sure it won't change much, but even running a Core Web Vitals test can teach you a lot about how to run a test, what you can learn, and the entire process. It’s not that SEOs are testing the wrong things, but that they are not testing enough.”

What don't you test yet that you intend to test in the future?

“There are hundreds of things I still haven't tested. The one thing I haven't tested yet is schema – like setting up a bunch of pages where I add a sentence description with schema and, within that sentence, I add entities that I think would make sense for a specific query. Those are the types of things I want to isolate. I could just be doing that, but I have to do my job as well. That is something I intend to do in the future.”

Should SEOs carry out testing in one batch or continue their testing over time?

“I would advise SEOs to have active tests going all the time, and you can even run some tests on your live site. You don’t always have to do it on a test site. Of course, it depends on what type of live site you have, but it is totally okay to test a title change on your site, for example.

Change the title of a blog post, start tracking it in your spreadsheet, and see what happens. There will be other factors in that type of test which may influence the rankings but even this is a type of a test: you do something, you document that, and you check what's going on afterwards.”

If an SEO is struggling for time, what should they stop doing right now so they can spend more time doing testing in 2024?

“Stop reading about and following shiny objects.

A huge example of that in the past was Core Web Vitals, and I was a victim of that as well. I went crazy about it, created tons of guides and tutorials, and studied the Google documentation – and it didn't really change anything for my site.

Go back to the fundamental rules of SEO and test them for yourself by simply creating a small test website (if you don't have one already). A lot of SEOs don't want to create their own sites because they think they don't have the skills and they don't know how. They want to learn the theory first, but it doesn't really matter. You can find a quick tutorial on how to set up a WordPress site in five minutes, and it is important that you do that.

You can start by testing something like how long it takes Google to index your new site when you publish it with 100 pages versus when you publish it with 5 pages. You can do anything on earth, and it can bring you valuable experience. Do more practice and less theory.”

Olga Zarr is the SEO Consultant at SEOSLY, and you can find her over at SEOSLY.com.