Citation Trends Explained: Ramp-Up, Peak Rate and Why Citations Fade
How to read first cited, ramp-up days, peak rate and decay half-life per page, and what to do when a once-cited page starts to slip.
A page is cited. Great. Now what? Most teams stop measuring at that point, which is like stopping at the first sale. Citations have a life cycle: they start, grow, peak and fade. If you can see the shape of that curve for each page, you can plan content work with some confidence instead of reacting to surprises.
Promptwatch's Citation Trends reports exactly this. It classifies every citation by content type and source type, shows citation rate over time, and gives four per-page numbers: first cited, ramp-up days, peak rate, and decay half-life. Here is how to think about each one.
First cited
The date a page first appeared as a source in tracked responses. It sounds trivial, but it is the anchor for everything else. If you published a page on March 3 and it was first cited on March 5, your content was found quickly. If it took two months, something slowed discovery, and the crawler logs are the first place to look.
Pair this number with the date you published. The gap is your time to first citation, and it is a useful benchmark across content types. Comparison pages may be cited sooner than how-to guides, or the reverse. You will only know for your own site by looking.
Ramp-up days
How long a page took to go from first citation to its peak rate. A short ramp means the page found its audience of prompts fast. A long ramp can mean the page is gradually earning citations as models re-retrieve it, or it can mean it is being cited for a narrow prompt set that slowly widens.
What to do with it: compare ramp-ups across your content types. If tutorials ramp in a week and long reference pages take a month, plan reviews accordingly. Do not judge a reference page after ten days.
Peak rate
The highest citation rate a page reached. This is the ceiling you demonstrated. If a page peaked at a strong rate in spring and now sits well below, you have a measurable regression, not a feeling. Peak rate also helps you pick winners when prioritizing refresh work: a page that once reached a high rate has proven demand in the prompt panel.
Decay half-life
The time it takes for a page's citation rate to fall to half its peak. It is the number most people have never seen, and it is arguably the most useful. A page with a short half-life is a news-style asset: it spikes and falls. A page with a long half-life is evergreen in the eyes of the models.
Short half-lives are not automatically bad. A launch announcement should fade. A pricing comparison should not, and if it does, the content is probably stale or a competitor's page overtook it.
Reading the four numbers together
| Pattern | Likely meaning | Typical response |
|---|---|---|
| Fast ramp, high peak, short half-life | News-like or trend-driven | Accept it, or refresh on a schedule |
| Slow ramp, high peak, long half-life | Evergreen authority | Protect it, track competitors |
| Fast ramp, low peak | Narrow prompt match | Broaden coverage or add sections |
| Never ramps | Discovery or access problem | Check crawler logs and internal links |
Why citations fade
There is no single cause, and guessing is expensive. These are the usual suspects, in the order we check them:
- Access changed. A bot started getting errors. Agent Analytics shows real-time, IP-verified crawler logs with failure explanations, which settles this quickly.
- A competitor replaced you. Page Tracker lets you watch a competitor's URL alongside yours, so you can see when their page starts earning the citations you lost.
- The content aged. Facts, prices or screenshots are out of date. Content Refresh in Promptwatch suggests diff-style updates for existing pages.
- The prompt shifted. Query fan-outs changed, so the model now searches for sub-questions your page does not answer.
- A different source type took over. If forum threads or video now supply the answer, Reddit, YouTube and LinkedIn citation reports show it.
A refresh trigger you can use
Pick a rule and write it down. For example: "When a page's citation rate has fallen to half its peak, it goes on the refresh list." That is the decay half-life made operational. You do not have to use a half; the point is that a threshold exists and someone owns it.
Then connect it to your weekly review. The Action Board's Discovery agent investigates your data on a schedule and proposes up to five evidence-backed actions per run, so a fading page can surface as a task instead of waiting for someone to notice.
What not to infer
Do not read the shape as a verdict on quality. A long ramp does not mean a weak page. A short half-life does not mean bad writing. These numbers describe behavior in a tracked prompt panel, and the panel is yours: change the panel, and the curve moves. Keep it stable across the period you compare.
Do not compare pages with very different prompt coverage. A page cited by two prompts and a page cited by forty will produce curves with different noise levels.
Recommendation
If you only track whether a page is cited today, you will be surprised twice a year. If you track its curve, you can schedule maintenance before the dip. Promptwatch publishes ramp-up, peak rate and decay half-life per page, alongside the crawler and traffic data that explain them. Start by adding your ten most important URLs to the Page Tracker, let a month of data accumulate, and write your first refresh threshold from what you see.