Although not a confirmed ranking factor, meta descriptions can influence search snippets and how AI search engines preview content.
Back in 2024, I proposed a Python framework to analyse the similarity between meta descriptions and search queries in order to influence rankings.
If you ask me today whether it’s worth the effort, the answer depends on the cost and resources required compared to the potential SEO gain.
While you delegate to the account the dirty work of sorting out the SOW, you can run small tests to gather insights and build a case for implementing meta descriptions.
I decided to hone my testing skills and I ran a test on a few old Python for SEO posts to see whether removing meta descriptions would lead to a drop in clicks.
TL;DR – Clicks decreased by 6% over a 29-day test.
In this post, I will outline the methodologies adopted and the results observed.
Honourable Mention
The testing methodology and reporting style were adapted from Giulia Panozzo’s resources on SEO tests.
I also drew on insights from a Sistrix-sponsored webinar in which Giulia appeared, which served as a key reference for this experiment.
Time-based tests (or pre/posts)
Following Giulia’s advice, I opted for a time-based test where you compare the performance of a group of pages against their previous performance.
Hypothesis
As with any test, you should first define the subject of your analysis (dependent variable), as well as any external factors that may influence it (independent variables)
Will removing custom meta descriptions from blog pages result in a 5% decrease in clicks?
Dependent variable: -5% decrease in clicks
Independent variable: removal of custom meta descriptions
Groups and Treatment
Testing Group: the test ran the oldest 7 posts from the Python for SEO subfolder
Control Group: the test considered the next 14 posts within the same folder
⚠️ Note – I considered pages with similar traffic share, and removed potential clicks anomalies
Control Group Randomisation
I then asked ChatGPT to randomise the control group batch (14 URLs) and separate them in 2 groups with the slightest difference as possible of clicks.
There are a bunch of more orthodox ways of randomising your control group using R, but automation worked like charm and saved me some time.
Next, I setup my 3 group of pages (7 testing group, 7 control group A, 7 control group B) and I removed all custom meta descriptions from the URLs in the testing group.
Finally, I requested indexing straight in Google Search Console for all testing URLs so that Google could pick up on the changes.
Test Dates
Launch Date: as soon as Googlebot crawled the changes, 1st August 2025, with check-ins scheduled 14 and 28 days post-launch.
Full Test Timeframe: (including pre and post periods): 30th July 2024 – 29th August 2025
⚠️ Note – Clicks were retrieved using the Google Search Console API via Screaming Frog and averaged across the last 12 months for both the pre and the post periods.
Results and Classifications
Over a period of 2 weeks, the dependent variable (clicks) didn’t show any visible improvements or decline. So I extended the test for 2 extra weeks and I observed a -6% decrease compared to pre-launch period.
Results: the test was labelled as a Winner as it confirmed my assumptions advanced in the hypothesis.
⚠️ Note – A new blog post on how to automate AI Overview tracking was published on 18th August. No meta description were implemented in the effort to rule out potential outliers in the testing batch.
⚠️ ⚠️ Note – By the end of the testing timeframe (26th August), there was an industry change aimed at tackling spam results, leading to potential turbolences in rankings.
Next Steps
Given the size of seodepths and the roll out of the Spam Update, the SEO gain from reinstating meta descriptions will likely be modest – yet proportional to the effort required for the implementation.
In fact, it will likely take me less than 30 minutes to reinstate meta descriptions across 7 URLs.
It’s simply a matter of accessing WordPress and asking ChatGPT to help me draw up a few snippets to manually submit via RankMath on each post using.
Meta Descriptions are as Valuable as the Effort Required
Are you familiar with the principle of economies of scale?
In traditional production, as output grows, the cost per unit typically falls because fixed costs are spread across more units.
In SEO, the same principle applies: if you can automate a workflow, the effort required to optimise hundreds of pages drops dramatically per page, while the potential return grows*
⚠️ *VIP Note – the expected SEO gain is largely dependent on the website and the industry it belongs to!
Winning Support: The First Step in SEO Testing
It may sound obvious, but getting stakeholder buy-in is a critical first step.
Whether you work in-house or at an agency, part of the job is persuading stakeholders to grant access to tools and testing space.
Technical barriers – like discrepancies between staging and production environments – or strained client relationships can slow you down.
Remember: SEO is as much about soft skills as technical skills.
Get the green light first, and the technical requirements will follow suit on your journey to SEO testing.
Simone De Palma
Simone De Palma is an SEO Manager at TUI and the founder of SEO Depths.
He graduated in Marketing and Management from Università IULM before completing a degree in Digital Marketing and Data Science at Leeds Beckett University.
Simone has worked as an SEO Specialist in digital agencies in Italy and the United Kingdom and he’s a contributor for the Search Engine Land and Majestic SEO podcast
When he’s away from his double screens, he enjoys cooling down with a refreshing swim at the pool. You could find him exploring art museums or enjoying the company of a classic romance
