Citation Trends Explained: Ramp-Up Days, Peak Rate and Decay Half-Life

Most citation reporting is a count. Page X was cited 40 times this month. That number is fine for a slide and weak for a decision, because it flattens time. A page that was cited 40 times in its first week and never again looks identical to one that was cited ten times a week for a month.

Citation Trends in Promptwatch tracks a page's life instead of its total. For each page it reports when it was first cited, how many days it took to ramp up, its peak rate, and its decay half-life. Those four numbers describe a curve, and the curve tells you what to do.

The four numbers

First cited. The date an AI answer first listed the page as a source. Useful on its own: it tells you how long publishing-to-citation takes for your site, which varies a lot by topic and by how quickly bots find new pages.

Ramp-up days. How long it took to go from first citation to peak rate. A short ramp suggests the page matched demand that was already there. A long ramp suggests it needed more fetches, more links or more topical support before engines trusted it.

Peak rate. The highest citation rate the page reached. This is the page's ceiling so far, in the prompts you track.

Decay half-life. The time it takes for the citation rate to fall to half after the peak. This is the number most teams do not know they need. A page with a long half-life holds its position. A short half-life flags content that goes stale or gets displaced quickly.

Reading the curve

Four common shapes show up, and each suggests a different response.

Fast ramp, high peak, long half-life. A durable winner. Leave it alone, link to it from new pages, and look at what it has in common with your other winners. It is a template.

Fast ramp, high peak, short half-life. A fashionable page. It answered a question that was hot, or a competitor replaced it. Check whether the topic is time-sensitive. If it is, schedule refreshes on a calendar. If it is not, look at who displaced it.

Slow ramp, low peak. The page exists but engines are lukewarm. Before rewriting, check crawl activity. If bots rarely fetch it, the problem is access or internal linking. If they fetch it often and still skip it, the content needs work.

Never cited. No curve at all. This belongs to the Page Tracker and the crawler logs, not to Citation Trends.

Using half-life to schedule refreshes

Refresh schedules are usually set by habit: every year, every quarter, whenever someone remembers. Half-life offers a better trigger. If your pages decay to half their rate in a short window, a refresh at that point recovers the most. If a page decays slowly, refreshing it early wastes effort you could spend on a page already sliding.

A workable rule: sort cited pages by decay speed, take the five with the shortest half-life and the highest peak, and refresh those first. They had the most to lose and the most to regain.

Promptwatch's Content Refresh helps here. It produces diff-style update suggestions for existing pages, so you change what needs changing and keep the URL, the links and the history. The content gap analysis, with coverage scores, shows which sub-questions the page still does not answer.

Cautions about the data

A curve is only as good as the prompts behind it. If your tracked panel changes, the citation rate changes for reasons unrelated to the page. Note any panel edits in your log, and avoid reading curves across a panel change.

Also keep the sample in mind. AI answers vary, and a rate built on few prompts is jumpy. A page tracked across a dozen prompts gives a steadier curve than one tracked across two. When a curve looks dramatic, check how many prompts feed it before you act.

Finally, a falling rate does not always mean your page got worse. Competitors publish, engines change source preferences, and the question itself can drift. Half-life shows that something changed. It does not say what.

Joining the curve to other evidence

To find the cause, put the curve next to other data.

• Crawler logs. Did bots stop fetching the page when the decay began? Agent Analytics shows fetches, errors and a citation rate per page.
• Competitor pages. Add the page that replaced you to the Page Tracker and compare how it answers the question.
• Visits. The 360 Insights view joins crawler logs, citation clicks and AI-referred visits per page, so you can see whether decay cost traffic or only vanity.

If you have the Professional plan or above, you have crawler logs for this. On Essential, you can still read Citation Trends and 360 Insights but without log evidence, so cause analysis relies more on comparison than on measurement.

A monthly review that uses the curve

1. List pages sorted by decay half-life.
2. Mark the short half-life, high peak pages.
3. Check each against crawl logs and competitor pages.
4. Queue refreshes for the ones where content is the cause.
5. Queue access fixes for the ones where crawling is the cause.
6. Note the date, so next month's curve has a visible event to compare.

The last step is the one people skip, and it is why many teams cannot tell whether a refresh worked. A dated entry turns an ambiguous curve into an experiment.

The recommendation

If you only track counts, you are throwing away the part of the data that tells you when to act. Citation Trends in Promptwatch reports ramp-up, peak and decay half-life per page. Pair it with Content Refresh and you have a refresh calendar based on evidence rather than habit. Start at promptwatch.com and open Citation Trends on the pages you already know are your best.