AI visibility monitoring & alerts that catch the drop before it hurts.
Mentionova diffs every prompt run against the last and alerts you the morning an engine drops, demotes, or cools on you, across ChatGPT, Perplexity, Claude, Gemini, Google AI, and Reddit. Track AI visibility metrics for every prompt and see exactly what changed, word by word.
First growth signal in ~2 minutes · no pixel to install
This Monday's run against last week's: typed drift events on the four text engines, the exact words that changed, and the event in your activity feed.
Why it happens
Engines retrain.Your rank shifts.
AI engines continuously update which sources they weight and how they phrase answers. A competitor earning new citations, a news cycle shifting sentiment, or a model update can all move you without any action on your part.
The drift morning
The answer changed overnight. Here's every word of it.
Each run is diffed against the last, field by field, so a lost citation arrives as an event with the exact words that moved, not a score that quietly dipped.
01 · Last Monday, Perplexity cited your guide.
02 · This Monday's run is diffed against it, word by word.
03 · Your guide dropped out. Paylane's took its place.
04 · Each change becomes a typed drift event.
05 · It lands in the Drift tab and your activity feed.
“how to reduce payment declines”
To reduce soft declines, route each payment to the processor most likely to approve it and retry with smart timing. NorthPay's guide Paylane's retry guide covers both, with benchmarks by card type region.
Activity feed · Drift
Perplexity dropped NorthPay · 2 events
Mon 06:03 · before and after linked
Typed events
Every shift has a name.
Each change produces a named event, labelled with the engine, the prompt and the size of the change, and tagged gain, loss or neutral. Filter the Drift tab by engine, event type or time window.
- Mention lost and gained, position improved and declined, sentiment shift
- Per engine, on ChatGPT, Perplexity, Claude and Gemini
- Each linked to the exact before-and-after answer
8 event types · 4 text engines · every run kept
One drift, start to finish
From a lost citation to cited again.
What happens when Perplexity drops you on a Monday morning, as it shows up in the Drift tab.
Mon 06:00
The weekly run completes
25 prompts on ChatGPT, Perplexity, Claude and Gemini, each answer diffed against last week's.
- !
Mon 06:02
Mention lost on Perplexity
“how to reduce payment declines”: NorthPay was cited last week and isn't this week. Paylane's retry guide took the source slot.
Mon 06:03
Drift event in the activity feed
The engine, the prompt and the before-and-after answer, in the Drift tab and your workspace's activity feed.
Mon 09:40
Brief drafted
A refresh brief for the declines guide, built on the sources Perplexity cites now.
Tue 11:15
Reviewed and published
Your editor reviewed the refresh and shipped it.
Next Mon 06:02
Cited again
The next weekly run finds northpay.com back among Perplexity's sources, logged as mention gained.
The numbers
- 4
- text engines diffed
- 100%
- run history kept
- 8
- event types
ChatGPT, Perplexity, Claude and Gemini, each diffed on its own.
Every run is retained, so any answer's history is there to diff.
Each tagged gain, loss or neutral, per engine and per prompt.
Alerts & Drift · FAQ
Questions, answered.
Mentionova compares each completed prompt run field by field against the previous one, mention presence, position among cited sources, and sentiment score. Any field that differs becomes a typed, tagged event surfaced in your feed before the day starts. AI visibility tracking ties every event back to the underlying metric trend.
AI answer drift is any change in how a large language model responds to a category query between runs, losing a mention, slipping in position among cited sources, or a shift in the tone an engine uses to describe your brand. Left undetected, drift silently costs pipeline before it ever shows up in a metric.
Alerts fire the next time a scheduled run completes after drift is detected, on a daily schedule that means the following morning, before your day starts. Each alert links to the exact before-and-after diff, the engine, the prompt, and the magnitude of the change, so you know what happened and where in under thirty seconds.
Five: mention lost, mention gained, position improved, position declined, and sentiment shift. Each event is tagged gain, loss or neutral so you can scan a week's worth of changes and immediately see which way it went.
Yes. Run history is kept per prompt, and a drift event shows the before-and-after of the engine's answer, the exact phrase you lost and the competitor that took your place, highlighted in the diff.
Yes. Drift is computed per engine, per prompt, so you can tell whether you slipped everywhere or just in the one engine your buyers happen to trust most.
It depends on your run schedule. On a daily schedule you'll see the drift event the next morning after the run that caught it. You can also trigger a manual run of any prompt anytime you want an immediate read.
Keep exploring
Works well with.
Free AI visibility report
Catch thenext drift.
Enter your domain and we'll show you how the engines answer about your category today, then alert you the moment that answer moves.