How to run SEO split test (easy way)
Made by Google Hundreds of variations Its algorithm is revised every year.
As an SEO, you have two options when dealing with them: make negative changes to the site (if it is negatively affected), or proactively conduct split testing to ensure that it works in the first place.
Split testing based on URL or SEO means dividing pages with similar intents on your website into two groups, making changes to one of them, and monitoring its impact on key SEO indicators (eg organic traffic, click-through rate) (CTR) ) Or keyword ranking position.
The result: minor but important adjustments have been made to your website to improve user access and help ensure that future algorithm updates are good news for your organic traffic, not bad news.
Let’s take a look at how to run an SEO split test without the help of complex code, a team of data scientists or developers. You can get the following from this guide:
- Check your website structure including templated pages
- Set hypothesis
- Divide pages into control and variation test groups
- Change the body page
- Report your test results
1. Check your website structure including templated pages
The first step in conducting an SEO split test is to check whether your website is large enough to ensure that you get statistically significant results.
Tests with too few pages or insufficient monthly organic visits will be unreliable. It’s almost impossible to tell whether the changes you made actually had an impact. Other things, such as slightly increasing search demand to a category, may lead to inflation. If the test is unsuccessful, it may be affected by external factors.
So, what is needed to perform a statistically significant test? Is it a “big enough” website?
To run a reliable SEO A/B test, your website should contain templated pages that have the same repeatable format and similar intent throughout your website. Here are some examples:
- If you are an e-commerce website, your category and product pages will be templated;
- If you are a member website, there may be user-generated content pages that have been templated;
- If you are a catalog website, you may have a large list of businesses or restaurants consisting of templated pages;
- If you are a stock photo website, all stock photo pages will have the same template;
- If you are a content or media website, your blog posts will all use the same template; and
- If you are a market site, there may be a list of vendor pages that all follow the same HTML/CSS template.
Your website should have at least the following two conditions at the same time before it can undergo SEO split testing:
- 300 templated pages; and
- These pages have 30,000 organic sessions per month
2. Make assumptions
Next, decide what to test based on your SEO strategy and priorities.
You can use many SEO experiments as a basis for split testing. Some examples include:
- Title tag And meta description;
- Internal link
- Content structure; and
- Alternate text
Once the focus area has been selected, this framework can be used to set a hypothesis: “We know [something], So we predict [this will happen].We will pass the test [action] And measure [metric]. “
For example: “We know that customers use our free shipping options more than any other delivery method, so we predict that adding’free shipping’ to the meta tag will increase organic traffic. We will pass the meta in the “variant” group Test this by adding “free shipping” to the description and measuring the difference in organic traffic.”
Before starting the test, be sure to remember that the SEO test and Conversion rate optimization (CRO), And user experience testing.
On the one hand, through user experience testing, you can change page elements, such as calls to action (CTA), checkout buttons, and messaging to see how visitors’ behavior changes. On the other hand, with SEO split testing, you can optimize the list to see how the search engine’s crawlers interpret the changes in order to find better results for these users.
The goal of SEO A/B testing is to attract more organic traffic to the webpage, not to increase your live conversion rate. This is why most traditional A/B testing tools (such as Google Optimize, Optimizely, and A/B Delicious) cannot perform SEO testing.
3. Divide pages into control and variant test groups
Once you have a hypothesis and a list of pages to test, you need to divide them into two groups:
- Control group: the original page, which will remain unchanged; and
- Experiment (variant) group: the test page on which the changes will be implemented
This avoids creating different versions of the same page. You will make changes to the pages in the “Variation” group, and then compare them with the pages in the “Control group” to see if you have a positive, neutral, or negative impact.
Therefore, please evenly divide pages with similar intent, such as product/category or review pages. The user is visiting every page with similar goals, whether it is buying goods or searching for business locally.
You can also use filters to run tests on specific pages. For example: if you want to test whether the review mode affects click-through rate, you can exclude pages that are less than one month old.
For a data science team, manually splitting hundreds of pages can be a daunting task. It involves exporting data into a spreadsheet and clearing out irrelevant pages, such as noindex or non-canonical URLs, old content, and custom links that redirect to other pages.
Even if you have the list, you will have to run a smaller preliminary test to check whether the “control” and “variant” groups are statistically similar.
The good news is that you can easily complete this tedious task Split signal.
This tool uses the historical data of Google Analytics (GA) and Google Search Console (GSC) to predict the variation group based on the control page, thus completing the hard work of segmenting the page. All you need to do is to connect Google Analytics and Google Search Console to SplitSignal, and SplitSignal will automatically divide them into Control and Variant groups for you.
4. Change the body page
Now it is time to implement the changes on your Variant test page.
Remember that you will only make changes to this group; the control group page will remain the same, so you can compare the two at the end.
URL-based experiments are difficult to perform manually. You need to create a ticket for the development team before you can make any changes on the website. You also need to negotiate with them and prove that it is necessary to prioritize SEO split testing over other tasks. This alone may take several months.
You can also build an internal tool to do this for you (this can take up a considerable part of your annual SEO budget).
The easiest way? SplitSignal. After splitting the page, configure the changes to be made to the Variant group, such as adding “free shipping” to the title tag.
A/B testing best practices apply here: You should only test one hypothesis so that you can be confident that positive or negative changes are the direct result of that test and nothing else.
Does this count as a concealment?
Disguise occurs when people and searchers see different versions of a page.
However, this SEO split test method does not violate Google Webmaster Guidelines cover up. Regardless of whether the viewer is a robot or a human, it will apply the changes to the page, so the robot and the user will always see the same version of the “variant” page.
5. Report your test results
It can take 14 to 42 days to get reliable results from SEO split testing.
The time it takes to get results depends on how quickly Google can index the page and whether the results are meaningful. In general, we recommend that you test within a range of 21 to 28 days.
You will see the actual traffic and forecast traffic of the pages in the Variant group in the SplitSignal dashboard, so there is no need to run complex calculations.Variant traffic forecast is to use Causal impact model And based on the control flow. Then compare it with the actual traffic on the unchanged page.
If the difference is large enough, we consider the test to be statistically significant.
Look at the three examples below. The orange line is the Variant group; the blue line is the control group.
The effect of the variant page in Test A was significantly better than that of the control group.
In Test B, however, the situation was the opposite: the variant group actually performed worse than the unaltered control group.
Finally, Test C showed no significant difference between the two groups.
6. Use test data to fine-tune your SEO strategy
After obtaining the SEO test results, you will have a good understanding of the adjustments that should be continued within the scope of the site.
For example: if your test concludes that adding “free shipping” to the meta title can increase organic visits by 10%, you can make changes on a more templated page.
The key to SEO split testing is to process previous results. Therefore, if the meta title tests positive, please increase the sample size.
Because your confidence in the initial test is high, the risk of extending the change to higher-value pages is low. You have proved this hypothesis in a small test group. Expanding similar changes on a larger range of similar intent pages may produce similar positive (or negative) results.
Are you ready to start SEO split testing?
There is no doubt that SEO split testing is essential to understanding how Google interprets (and ranks) your website. By testing the small adjustments of the page group, you will begin to discover which small adjustments will bring changes to your SEO strategy.
good news? You can use SplitSignal to easily and easily conduct large-scale SEO A/B testing and determine whether the test is successful. You don’t need to ask the data science team to split pages, nor your development team to implement them, nor complex equations to determine statistical significance.
The data will be interpreted; you will be able to see the final result (and how we got there).
The best part? All changes are made through lightweight code, which has no significant impact on Core Web Vitals.
Of course, the adjustment you are testing seems to be small. However, a small change from 10,000 product pages to 10 million product pages may result in a 10% to 20% increase in organic traffic to these pages, which will be worth millions of dollars.
( Exclusive concept, For example, traffic on the variant page increased by 21%. This change will increase the revenue of the entire website by 5% in the next 12 months. )
Start using URL-based free SEO split testing today Split signal.