People often approach SEO testing with the limiting belief that changing title tags and running them through state-of-the-art A/B Split Testing programs is a niche business.
The reality is that SEO testing is ubiquitous. Everything we do, whether it’s updating a title tag, or spending 12 months on a link acquisition campaign, is an investment.
Some of our SEO investments are small. Some SEO investments are massive.
Some SEO investments are time-only investments. Some are time + financial investments.
If an SEO activity is an investment, then it calls for measurement on our investment.
And if it calls for measurement then it is by nature, a test. Regardless of the flaws in our data analysis techniques (see It’s Time to Re-evaluate Our Love Affair with Statistical Significance), we still need to answer the fundamental questions, “was it worth the effort?” and “should we keep doing it?”
Once we understand that all SEO investments are experiments, an insanely vast world opens up in which we can unchain SEO testing’s attachment to title tags and stat sig requirements to look at all investments as test-worthy hypotheses.
Not covered in this article:
This article is an idea factory, not a how-to. So, if you’re in need of some extra guidance on how to set up, implement, and measure SEO experiments, I implore you to check out some of my other resources on the subject:
Intro: How to Start SEO Testing
6 SEO Testing Techniques that Lead to Better Organic Performance
In-Depth Guide to SEO Title Testing
Without further ado, here is an incomplete, but a BIG list of SEO testing ideas:
1. Title Tests
You didn’t actually think we’d skip title tags, did you?
Although title tags are far from the only SEO testing variable that we can run with, it is generally the best place to start learning the testing process itself, so I recommend beginning your SEO testing journey here.
What I love about title testing is that it teaches us an incredible amount about how Google operates. The title testing process involves some of the most important skills for SEO professionals to learn, including SERP analysis, intent research, indexation, data analysis, and reporting.
Not to mention, there’s a direct correlation between title relevancy to search intent, and keyword rankings that I think most SEOs will agree is stronger than most other page elements (H1s, meta descriptions, images, etc.).
Title testing gets filed under our “SEO quick wins” category. Fast execution time makes for an easier lift than writing a 1,000-word blog post or pitching publishers for a guest post.
Example title testing result on my own website:
2. Featured Snippet Tests
After title tags, featured snippets are my second-favorite page element for testing.
We often hear about featured snippets in the context of optimizations, which is why most SEOs refer to them as “featured snippet optimizations.”
But optimizing something implies that we know our updates will be better than the previous variation.
When we start to benchmark and measure featured snippets in the same way that we do with title tests, we add an extra layer of process and accountability to acquiring even more featured snippets than we might have acquired in “optimization territory.”
The testing process is how we scaled a few hundred featured snippets at Tipalti.
There is one major difference with featured snippet testing.
Results are boolean.
Testing a featured snippet is positive if we capture the featured snippet, and null if we don’t!
3. Content Refresh Tests
Like featured snippets, content refresh projects require follow-up to see if our investment had a payoff.
The follow-up measurements are best-performed using our time-based SEO testing technique where the original content is our control, and the new content is our variant.
4. URL Switch Tests
A little-known secret in the SEO world is that URLs can have a similar algorithmic influence that title tags carry.
Myself, and others have tested the theory in two ways. The first way is optimizing the URL to see whether or not our test generated a positive impact on rankings.
The second test is de-optimizing the URL to measure if the de-optimized URL has a negative impact on rankings.
Testing forwards and backwards helps validate the theory that URLs do have an impact on SEO performance.
Personally, I believe that SEOs should be doing more URL testing, but the reason these tests don’t get run (or talked about) nearly often enough is that they can be quite risky. As I covered in my Moz piece, you’ll need to utilize redirects in order to run URL switch tests correctly. That means that they’re much more tricky to unwind and re-iterate than title tags.
This shouldn’t deter us. It should just make us more precise about when and how to run URL switch tests.
5. Section Rearrange Tests
Another experiment covered in the Moz article is section re-arrange testing. These tests are extremely fast and easy to execute when working in a CMS with block editor capabilities.
Sometimes, just dragging and dropping sections can improve the layout of our page experience by prioritizing content that users are looking for above the fold, or before content that’s less important to users.
Even if you’re not working in a block editor, these experiments are usually very straightforward and easy to execute.
6. Internal Linking Tests
Internal linking projects can sometimes require a very high investment of resources.
And, whenever we have a high investment of resources, we again face accountability for the efficacy of our resources on measurable outcomes.
Of course, it’s well-established that improving internal linking is a good thing. But how good exactly?
Let’s say that we want to build a dynamic internal linking module that appears on thousands of programmatic pages to link them together. That project may take a few dev hours, some database work, and launch within a day or two.
After our modules are launched, we can measure rankings and traffic performance to the linked pages to verify if our investment yielded the desired outcome we had hoped to see. My hunch tells me that our outcome would be positive and that the initiative was a good use of resources.
Now, what if the internal linking project was a manual one? What if our idea was to perfectly organize our topic clusters and internal linking architecture through days and days of manual fine-tuning, followed by one-by-one link insertions in our CMS?
A time-based test for a project of this size may just as easily yield a negative outcome, which could help us avoid making the same mistake in future internal linking projects.
7. Content Removal Tests
CRO professionals occasionally run something called, “a takeaway test.”
In CRO, a takeaway test is exactly as it sounds. Want to see if that demo video really helps generate more form fills? Try taking it away from your variant page.
Want to see if the live chat popup is distracting users from completing form fills? Take it away.
In SEO testing, we can test similarly with content removals.
Content removals are actually quite an ingenious way to cut through the theories we have about what’s working vs. what isn’t.
Most of us would likely agree that multimedia is a good thing for SEO. But a great way to really validate our love of multimedia is to try taking it away. Same with certain schema markups, product reviews, and other best practices that we so often do without measuring.
Once we’ve taken something away, the post-removal performance can tell us a lot about our theories.
8. External Linking Campaigns
We’ve all seen the case studies.
A brilliant Digital PR campaign generates massive volumes of links from high-quality publications. Traffic to the website skyrockets, and the rest is history.
Far be it for me to discredit these case studies.
In fact, I believe them.
What we don’t often see, however, are those case studies where link investment did not generate positive growth outcomes.
The campaigns that don’t get reported on, are the campaigns where we invested time and money into link campaigns and saw nothing on the other side.
In the case of link acquisition, measurement, and testing become more complex than a time-based test because we rarely can point at a single date. Most link campaigns generate links over time, which makes the SEO testing process quite difficult.
Still, where there is an investment, there is also accountability.
My suggestion for SEO testing with external links is to find ways to isolate other variables.
Example:
Send a strong number of quality external links to one page, wait, and see if performance improves.
Focusing your investment in a single direction makes it easier to measure.
Measuring an experiment like this won’t be perfect, but you may be able to reach 60-70% confidence.
Repeat the test multiple times, and your confidence intervals will increase.
Hey, I didn’t say it was easy. But if we’re spending big money on this stuff, the unavoidable truth remains that we’re still accountable for investing wisely.
9. Technical SEO Tests
Once again, in most projects, we invest income with a certain level of accountability, and that’s just as much true of technical projects as it is of content projects.
Sometimes, technical SEO testing is even more critical when we’ve got a dev team piling on more technical debt than Kanye West’s -$53 mil.
SEOs have a long history of sending dev teams on wild goose chases to fix audit issues that get the website absolutely nothing in return.
Got a big CWV project? Measure it as a time-based test.
Switching over from raw JS to pre-rendering? Measure it as a time-based test.
Sometimes a successful test in the tech SEO realm isn’t measured by clicks, but by other success outcomes such as crawlability, server performance, or index bloat.
But if organic performance really is the goal, it helps to test and measure, rather than task the team with issues for no other reason than they showed up in our tech SEO audits.
10. Site Migration Tests
Because site migrations can be massive projects with far-reaching impacts to performance, the time-based SEO testing process is our most secure way of analyzing the impacts that a migration project gives rise to.
Granted, most site migrations are not done to generate lifts in organic performance, which means that SEO testing is not the end goal of site migration.
Still, projects of this magnitude are accountable for performance, and they require a before/after performance measurement where your website before the migration serves as the control, and your website after the migration serves as the variant.
11. Schema Markup Tests
Schema markup testing, like this study from SEOClarity, is one of the most fascinating variables that I’d like to see get tested more often. Some Schema markups result in a rich snippet. Other types of Schema do not. Schema markups that do generate a rich snippet are purported to attract more clicks.
I’ve always been interested in measuring these more accurately. What exactly is the percentage of net-new clicks that we can get from Schema-rich snippets?
And what about Schema that doesn’t generate rich snippets? What exactly is the performance impact of adding Schema markup to pages where rich snippets don’t appear?
Selfishly, I’d really love to see more SEO tests in this category. Schema testing is one of those test hypotheses that I haven’t explored nearly as much as I’d like.
Perhaps there’s an SEO out there who’ll run more Schema tests for me?
12. Topic Clustering Tests
As we go further down this list, we start getting into SEO testing territory that I really have not explored which means two things:
- Uncharted SEO testing territory is the most interesting territory.
- I need more people to help run these tests and report them for our edification.
Topic clustering is another one of those areas within SEO where I’ve not seen a whole lot of clear SEO testing being performed. There are quite a few case studies, like this one from MINUTTIA, which is itself a form of SEO testing.
Still, I’d really love to see some more methodical testing being done around these initiatives.
Personally, I find topic clustering valuable, but difficult to implement due to the number of hours it requires to execute. Anytime I’m working on projects with high investment, I really love to measure a bit more closely to ensure that my high investment is met with high reward.
13. All the Non-SEO SEO Tests
Okay, what is this? Well, I’d like to put it out there that a non-SEO SEO test is any experiment where a non-SEO activity is being deployed which we suspect could have a material impact on performance but requires testing to understand its impact.
This could be a project from another team or department, such as a site-wide popup, a sidebar calculator, or perhaps some load-heavy javascript tag.
Projects that come at our organic pages from outside of SEO are also really great candidates for SEO testing because they can sometimes cause us to worry about performance drops.
Sometimes our fears are validated, other times these changes have no impact whatsoever, which leads us to realize that we’ve caused a ruckus for no real reason.
Wouldn’t it be nice to stop sweating the small stuff for a change?
Conclusion
I’ve said before that almost any controllable element can be tested for SEO impact as long as we’ve got a sound methodology for measuring the experiment.
This, of course, doesn’t mean that any controllable element should be tested.
Use your judgment wisely by testing ideas that either have the potential to impact Organic performance in a meaningful way, or by testing ideas from a place of genuine curiosity to see if they’ll have any impact on organic performance.
With that, I’m sure there are quite a lot of projects and ideas out there that I haven’t thought to experiment with. Please feel free to drop me a message if you’ve got any burning experiment ideas that you’d like to add.
No comments:
Post a Comment