Does content freshness increase AI citations?

A universal recommendation, untested

Content freshness is graded Hypothesis on the GEO tactic evidence scoreboard. It's widely recommended, and mechanistically plausible: a more recently verified fact is a reasonable thing for a retrieval system to prefer. But no controlled test exists that isolates freshness from every other variable that changes when a page gets updated.

The circumstantial evidence

Some circulating figures claim a high share of frequently-cited pages were recently updated. But these are observational. Recently-updated pages likely differ from stale ones in other ways too: more actively maintained sites, more attention generally. Correlation between freshness and citation doesn't establish that updating a specific page causes its citation rate to rise.

Study design

What counts as a "genuine" update, precisely

A genuine update for this study's purposes must meet at least one criterion, agreed with panel participants before enrollment. Adding at least one new, substantively informative section. Correcting a factual error or outdated figure. Or expanding existing coverage with new examples, data, or detail.

Simply changing a visible "last updated" date, reformatting without content change, or making only cosmetic edits doesn't qualify. That distinction matters. A study testing only date-stamp changes would answer a narrower, less interesting question than the one this page is actually asking.

The registered hypothesis

H1: A genuine content update (new information, corrected facts, expanded coverage) to a stale page causally increases its AI citation rate within one collection cycle, relative to a matched control page left unchanged.

A worked hypothetical example

Here's the measurement made concrete, with invented, illustrative numbers only. Say a stale page on "how to structure a product comparison page" sat at a 6% citation rate across the tracked query set for six months prior to enrollment. It gets randomized into the treatment group, and receives a genuine update: a new section on structuring comparisons for AI answer boxes specifically, two corrected statistics, and an expanded FAQ.

One collection cycle later, its citation rate measures at 14%. Its matched control page, left untouched, moves from 5% to 6% over the same window, noise-level movement consistent with normal variance, not a real change. If this pattern held across the full 300-page panel, not just one invented example, it would count as the kind of clear, pre-registered support the study is designed to detect. A flat or inconsistent pattern across the panel would instead support the null.

How classic SEO tests freshness, for contrast

Classic SEO has studied a version of this question for over a decade, generally under the label "freshness as a ranking signal." The idea: Google's algorithm gives some queries, particularly ones tied to current events or rapidly-changing topics, a boost to recently updated content. That literature is far from settled itself. Freshness appears to matter a great deal for a narrow class of time-sensitive queries, and much less for stable, reference-style ones.

Suppose AI citation behaves similarly: freshness mattering mainly for queries where "current" information is part of what makes an answer correct, and mattering little for queries about stable facts. That would be a genuinely useful finding this study's design is positioned to surface, since the panel spans a range of query types, not only time-sensitive ones.

The confound this design controls for

This design randomly assigns which pages get updated, rather than comparing naturally-updated pages to naturally-stale ones. That breaks the link between "being the kind of site that updates content" and "getting updated content," isolating the update itself as the variable under test.

Plausible mechanisms, either direction

Suppose a positive effect exists. A plausible mechanism: a retrieval system weighting recency signals, whether through a last-modified date, a genuinely different crawl timestamp, or updated content simply displacing an outdated version in a training or retrieval corpus, would favor the freshly updated version.

Suppose no effect exists instead. A plausible explanation: current-generation retrieval leans more heavily on accumulated authority and existing citation patterns than on freshness signals specifically. Or any freshness signal genuinely present gets outweighed by other factors this design holds constant: topic, existing authority, backlink profile.

Schedule

Recruitment alongside the schema RCT panel, targeting the same Q1 2027 collection window as Index v1.

How this relates to the citation half-life study

This study and the citation half-life study examine closely related but distinct questions using overlapping panel infrastructure: half-life asks how long a citation naturally persists without intervention; this study asks whether a deliberate intervention (updating the page) can extend or restore a citation. A future combined analysis — do updated pages show a longer half-life than untouched ones — is a natural extension once both studies have collected enough data, though it is not part of either study's initial pre-registered design.

What to do while you wait

Keep updating your content for the reasons that already hold up on their own: accuracy, user trust, and classic SEO's freshness signal. None of that depends on this study's outcome.

Just be honest about why you're doing it. Say "we correct and expand this page regularly to keep it accurate," not "we update this page because it boosts AI citation rate." The second claim isn't backed by evidence yet, and this study exists specifically to find out whether it ever will be.

Limitations

• "Genuine update" still involves judgment about what counts as substantive — the exact revision criteria will be published alongside the panel recruitment to prevent post-hoc redefinition.
• One collection cycle may be too short to detect an effect that takes longer to propagate through an engine's retrieval index — a limitation the study will flag explicitly if the result is null.