AI LAB CASE STUDY
Do AI Models Care Where Key Information Is Presented? Testing H1s in GEO
Key takeaway: Reformatting H1 headings to surface key information led to a 47% increase in OpenAI’s GPT-User bot hits.
The Challenge
Our client, a travel aggregator site, is in an extremely competitive industry where brand presence and positioning are essential to their success. The rise in AI-driven search has added further complexity thanks to the expanded playing field outside of traditional SEO. We conducted competitive analyses of their AI traffic to identify gaps in visibility and set a goal to improve their AI presence across lower-performing pages.
So we asked: Could we improve AI performance by surfacing key details in H1 headings?
Hypothesis
We theorized that LLMs and AI search features prefer when key information is easy to parse. By highlighting important details more clearly on the page — in this case, including price points in the H1 heading — we believed our clients would be more likely to see their content surfaced in LLMs. As a result, we would improve their AI traffic and help them outperform competitors.
Validation: After reviewing our client’s log files, we noticed that pages including a price point in the H1 tag received 126% more GPT-User bot log hits compared to pages without a price point in the H1.
Note: For this client in particular, we did not have access to analytics data, so our analysis was focused on their log files.
Strategy
To test our hypothesis, we implemented pricing in H1s on a subset of pages while monitoring AI-driven bot hits before and after the changes (and comparing the results to our control group). We added key pricing details into H1 headings on 81 test pages to see if AI bots would be more likely to check those pages in responses.
Based on the results, we planned to roll out the test to the remaining pages without pricing in the H1 and capture potential AI visibility gains on a larger scale.
Results
+47% increase in AI bot hits on test pages vs. +24% increase on control pages
The test revealed that pages with key information in the H1 heading outperformed our control pages by 23 percentage points. Our test pages saw a 47% increase in AI bot hits, compared to a 24% increase for our control pages.

These results suggest that including key information in H1s (like pricing details for a company in the travel aggregator space) makes it easier for AI models to quickly parse and understand page content. For our client, clearly displayed pricing information likely increased AI’s confidence in surfacing the page for relevant user queries.
Next Steps
Considering the successful outcome of the initial H1 test, our next step is to scale the H1 formatting changes to a broader set of pages. Seer recommends that this low-effort tactic has the potential to have a high impact on AI traffic and visibility. By continuing to validate these patterns, we hope to establish clearer guidance on how low-lift modifications can drive GEO outcomes.