SEER AI LABS
Do Author Bylines Influence AI Visibility? Testing EEAT Elements in GEO
Key takeaway: When author bylines were implemented on 123 blog pages, we saw an increase in Googlebot crawl rates (+11.7pp) and citations via Bing Webmaster Tools (+109%). The initial signals were positive but inconclusive, and require deeper analysis of some variances like page topics, recency, and prompt selection.
The Challenge
AI platforms have often cited lower-quality content like listicles over content that adheres to Google’s EEAT (Experience, Expertise, Authority, and Trust) standards. We wondered what elements we could further optimize to increase the likelihood of higher-quality being cited by AI.
As more people turn to AI platforms to find information, trust is the element that will keep users coming back. The platform has to trust the sources it pulls from, and the user has to trust the answer they get. Source authors must also be viewed as trustworthy: both by the user, for them to believe the information, and by the AI platform, for it to view the author as a credible expert.
But adding author bylines isn't always as simple as it sounds. Depending on the CMS, surfacing a complete, structured byline can mean hard-coding fields, adjusting templates, or working around a platform that was never built to display them.
We know Google has named author bylines as a trust signal for high-quality content. The question we wanted to answer: do AI platforms read author bylines as trust signals the same way Google does?
Type
What We Tested
We hypothesized that AI models read author attribution as a signal of credibility and expertise, the same way traditional search does. As low-quality content strategies have historically worked to get cited by AI platforms, we tested to see if adding in additional EEAT signals onto our pages would increase their AI visibility.
We analyzed the blog subfolders of the top 6 digital marketing agencies earning citations in our Scrunch profile to gauge the author-related signals most prevalent on these pages. Six byline elements appeared consistently:
- Verifiable identity: Headshot, full name, link to a dedicated author profile
- Stated title with company affiliation: Specific role tied to the agency
- Specific expertise claim: 3-5 named focus areas
- External credential validation: Publications, conferences, or certifications
- Tenure or experience claim: Numeric statement of years in the relevant domain
- Schema markup: Person schema with worksFor, jobTitle, and sameAs (separately, we’ve confirmed that Schema markup influences AI visibility)
Validation: We know Google’s Search Quality Rater Guidelines (SQRG) emphasize the importance of "high-quality content" and "content expertise" as key ranking factors. These factors also signal positive content creator reputation and creator authority.
Strategy
We added detailed bylines, using a mix of the elements above, to 123 pages on the Seer Insights blog. From there, we formed two primary test groups:
- Treatment group: The 123 treated Insights pages
- Control group: Untreated Insights blog pages as our actual comparison group
We selected a mixture of authors to add the bylines for, ranging from people that AI will 'know' more about (like Seer's founder Wil Reynolds) to Seer team members with less information.
At this point, we developed a secondary hypothesis for the experiment to measure if the author's recognizability mattered. We believed pages with Wil's byline would see less change due to the authority already associated with his name, while team members with lower levels of public authority would experience higher levels of visibility increases.
We sorted 95 solo-byline pages (to account for collabed posts) into three tiers:
- Tier 1: Wil's pages (47 pages)
- Tier 2: Posts by two of Seer’s recognized practice leads (33 pages)
- Tier 3: Posts by three of Seer’s recognized team members (15 pages)
Results
→ Consistent rise in AI bot traffic on all pages: blanked (+22%), placeholder (+12.6%), and control (+21.7%)
→ -1.3% true effect (statistically zero) on AI retrieval crawl once we account for the sitewide trend
→ No AI citation penalty on the de-optimized pages
The bylines didn't deliver the across-the-board citation lift we expected, but some of the signals showed positive indicators.
Methodology note: We analyzed AI Overviews using data sources like Semrush, Bing Webmaster Tools citations, and Googlebot crawl rates due to the wide group of treatment versus control pages. These sources weren't beholden to limitations of the prompts being tracked.
AI Overviews
On AI Overviews, treated pages performed roughly in line with untreated Insights pages over the same window. We modeled the daily citation series against a control group and didn't find a statistically significant difference between the two.
Bing Webmaster Tools
The Bing Webmaster Tools AI citation data tells a cleaner story. While Bing’s AI Performance report is still in beta, it can report on a broader universe of citations rather than a sampled prompt set.
In this Bing data, treated pages roughly doubled in citations period-over-period compared to a ~34% gain on the control group. Of the treated pages with prior citation activity, 44% improved versus 27% of control pages.
Googlebot
The bot data points in the same direction. Googlebot hit our treated pages 11.7 percentage points harder than the control (likely triggered by the page changes). The spike occurred during a two-week window before settling back to baseline performance.
Control Group
In our dataset with a controlled group of prompts, we did see citation declines from all AI platforms - with the exception of ChatGPT. But none of these results were conclusive.
With Bing’s AI performance data, we have an initial signal that author recognizability does matter, but are expanding this test to confirm. Least-known authors showed the largest gain (+113%) and Wil's pages saw the smallest increase (+21%). That being said, each tier has only a handful of pages in the Bing report and the totals are driven by a few high-volume URLs.
The signal is worth digging deeper into; but we’ll admit that some variances, like page topics and recency, could have influenced the initial result.
Next Steps
Our next step is to roll out author bylines across a larger set of Seer team members and pages, expanding the recognizability test.
We're not ready to write off EEAT signals as a non-factor yet. We'll keep monitoring these pages and start testing other trust levers (quotes, reviews, ratings, and similar elements) to see how much trust signals can shape AI visibility.
About the Author
Jamir Ong
AI Optimization Intern
Jamir Ong is an AI Optimization Intern at Seer Interactive, where he's running experiments and research focused on exploring technical decisions that may influence AI visibility. Before Seer, Jamir did hands-on work with AI systems in education to improve the user experience of over 17 schools around the nation.