The 11-Day AI Citation Half-Life: What It Means for Your Content

The widely shared 11-day AI citation half-life is a measure of relative citation-share decline after a peak. It is not a deadline after which a page stops being cited or a brand loses its credibility.

For content teams, the useful response is better monitoring and diagnosis—not automatic rewriting every fortnight.

What did the study track?

Profound’s September 2026 research tracked approximately 883,000 pages across seven answer engines over twelve months. Measuring URLs separately on each engine produced 1,188,816 qualifying page–engine combinations.

It calculated a page’s daily citation share within its category, using a fixed eligible prompt denominator, and smoothed the series with a trailing 14-day average. Half-life was the time from the smoothed peak to half that level, sustained for 14 consecutive days.

Those choices are part of the metric, not incidental technical detail.

What were the headline findings?

The median time from peak to half-peak was 11 days. Among combinations with sufficient follow-up for the short-term split, 78% reached half-peak within two weeks.

The rolling average matters. Profound explains that a brief burst can remain in the smoothing window for roughly two weeks, helping produce that observed pattern.

A decline in relative share is also not identical to an equivalent decline in absolute citation count. What other pages receive and how the denominator behaves matter to the result.

What does the study suggest about durability?

Pages cited by five or more engines were reported as 5.5 times as likely as single-engine pages to remain above half-peak after eight weeks. The underlying estimates were 6.8% and 1.2%.

That is an association. The analysis does not prove that securing an additional engine citation causes longer survival.

The research also found different patterns by content family, with comparisons, explanations and commercial pages retaining share better than reporting in its sustained-page analysis.

What does it not establish?

The study did not test a scheduled-refresh intervention against unchanged comparison pages. It therefore cannot establish that rewriting every few weeks prevents citation decay.

It also measured URL-level share, not the expiration of company expertise or brand reputation. A page falling from its peak does not tell you whether another page from the same business gained visibility.

What should a team do?

Track priority questions repeatedly and investigate material changes. Check whether the issue is factual accuracy, topic relevance, engine variability or a measurement change.

Maintain time-sensitive pages when information changes. Improve evergreen pages when there is a useful contribution to make. Continue publishing original work when buyer needs justify it.

The correct lesson is not “publish or disappear in eleven days”. It is “do not mistake one citation screenshot for a durable visibility position”.