Does Optimizing Structured Data Really Give You a Competitive Edge in Google AI Overviews?
A team conducted a controlled test to analyze the impact of schema markup on search engine performance. They compared three nearly identical web pages. Each page had a different schema quality. One page featured strong, well-implemented schema markup. Another page employed poor schema definitions. The last page had no schema markup at all.
The results of this test were quite telling. Notably, only the page with the well-implemented schema appeared in an AI Overview. This is a significant feature that allows certain pages to be highlighted by algorithmic processes. It leads to increased visibility among search engine results. This page not only received the advantage of being showcased in AI Overview. It also achieved the best organic search ranking compared to the other two pages.
The findings suggest schema markup quality is crucial. It influences the likelihood of appearing in AI-driven features. It also enhances overall organic ranking. Search engines value well-structured and high-quality schema over mere presence. This underscores the importance for webmasters and content creators to invest in proper schema markup implementation. Doing so can improve their pages’ visibility and performance in search results. So, attention to detail in schema usage is a key factor in gaining a competitive edge in search rankings.
Why You Should Question Schema, AI Overviews, and the Demand for Proof
AI Overview visibility is becoming increasingly important to businesses.
A debate within the SEO community has stood out. The question is whether adding schema improves the chances of being cited in an AI Overview.
Schema was created to make webpages more machine-readable. It has also been shown to help large language models – like Microsoft’s better interpret content freshness.
That makes it tempting to assume schema is a best practice for AI visibility.
Still, AI Overviews are the result of complex and layered processes.
It’s difficult to draw firm conclusions from logic alone or from limited glimpses into one part of a model’s behavior.
That uncertainty is what motivated us to run a controlled experiment.
- In earlier work, we analyzed 100 healthcare sites and found a slight correlation between schema use and AI Overview visibility. But the correlation was not statistically significant. The analysis had two limitations. It didn’t assess the quality of the schema. Also, because it wasn’t an experiment, site differences in content, structure, and audience couldn’t be controlled.
- At the same time, other experiments showed that ChatGPT retrieved information more thoroughly and accurately from pages with structured data. Those findings pointed to schema’s role in AI visibility, but they didn’t address Google’s AI Overviews.
With those perspectives in mind, we decided to collaborate on a test. This test builds on the earlier analysis. It also extends the second experiment into Google Search. Our focus is directly on whether schema quality plays a role in AI Overview visibility.
The setup: Three sites, three schema approaches – Which will reign supreme?
We built three single-page sites to compare schema directly:
- One with well-implemented schema.
- One with poorly implemented schema.
- One with none.
Aside from schema, the pages were kept as similar as possible. Keywords were chosen to match in difficulty and search volume.
After publishing, we submitted all three for indexing. We wanted to see whether they would rank. More importantly, we wanted to know if any would appear in an AI Overview.
The Triumph: Only the Page with Flawless Schema Shone in the AI Overview!
The page with well-implemented schema stood out as the only one featured in an AI Overview. It highlighted the significant impact that structured data can have on a page’s visibility. This data can also affect engagement with users. This particular page not only enjoyed the advantage of appearing in this competitive AI context. It also managed to rank for six distinct keywords in traditional search results. The page achieved an impressive peak at Position 3. This Rank 3 designation was remarkable. It was the highest conventional search rank reached by any page during our extensive experiment. The appearance in the AI Overview was directly linked to this query. It further emphasized the symbiotic relationship between effective schema implementation and search engine outcomes.
In contrast, the page that featured poorly implemented schema managed to rank for a larger set of 10 keywords. However, it faltered and peaked only at Position 8 in the search results. The critical distinction here is that none of the queries associated with this page led to an AI Overview appearance. This indicates that despite a broader keyword reach, the lack of robust schema diminished its potential. This made it harder for the page to stand out in an AI-driven environment.
The page did not implement any schema. Despite this, it was promptly crawled by Google within minutes of the others. However, despite this quick crawl, it faced the significant setback of not being indexed. Without proper indexing, this page could not rank for any keywords, effectively sealing its fate in terms of visibility. The absence of indexing meant it was also unable to appear in any AI Overviews. This showcases the importance of not just being crawled, but also being properly indexed. Proper indexing is essential to achieve meaningful search engine results and visibility in AI contexts.
Our findings underscore the crucial role that schema markup plays in enhancing a page’s performance. This applies both to traditional search rankings and in emerging AI-driven search contexts. It illustrates a clear advantage for those who prioritize structured data in their web content strategy.
Methodology: Unmasking the Secrets Behind ‘Good Schema’ Control
To isolate schema as the variable, we kept everything else about the test pages consistent. We ensured consistency in aspects like keyword choice and site setup.
Keyword choice
We used Ahrefs to choose three keywords with identical metrics. Each returned an AI Overview at the time of selection:
- “How much does a marketing team cost.”
- “What are common elements in the promotional mix.”
- “Data pool vs. data lake.”
Metrics (Ahrefs)
- Keyword difficulty: 3
- Monthly search volume: 60
- Traffic potential: 20
We also chose keywords that were qualitatively similar and within the same general industry (marketing/martech).
Site build controls
All three were single-page sites deployed on Vercel, with the following constraints applied consistently:
- No JavaScript.
- No custom domain name or homepage.
- No sitemap.
- No robots.txt file.
- No canonical tags.
Schema treatments
To create a page that exemplified a solid implementation of schema best practices, we included:
- Complete
Articleschema with all required fields. FAQschema for common questions.- Breadcrumb navigation schema.
- Proper date formatting.
- Author and publisher information.
- Educational level and audience targeting.
- Related topics and mentions.
- Word count and reading time.
We deliberately introduced errors into the poor schema page, including:
- Incomplete
Articleschema (missing required fields). - No
FAQschema despite having FAQ-like content. - Missing breadcrumb navigation schema.
- Incorrect date format.
- Missing essential properties.
The third site was built without any schema at all.
All three sites were submitted to Google on Aug. 29 and crawled the same night.
Unraveling the Results: Hopeful Hints or Just Hot Air?
We don’t think these results are proof that good schema helps with AI Overview visibility.
Still, it’s clear: the page with good schema performed the best in our small test. It had the highest organic ranking and was the only page shown in an AI Overview.
We also don’t see any other clear reasons for this outcome.
The page without schema had the fewest words compared to the other pages, but word count shouldn’t be important.
What’s next
There’s still more to do.
Some hidden factors might have confused the results. It’s possible that our findings were just a random occurrence due to the Google algorithm.
Next, we plan to remove the current pages from search results. After that, we will make new pages with the same content. We will then change the schema.
We want to determine whether adding schema to a page helps it get noticed. This page previously had no schema. That would be a very interesting result.
