How AI engines perceive your brand

Stored and compared analyses

• The latest analysis shows when the page opens, with its date: refreshing or closing the page no longer makes it disappear. A new analysis only starts when you click.
• The history keeps every analysis; you can reopen any of them.
• The period bar compares the latest analysis of one period with another's: alignment score, net sentiment, and for each AI engine the positioning, tone, adjectives, topics, strengths and caveats gained or lost.
• The analysis can run in the background: start it, switch to another PivotSEO tab, and the « N analyses running » indicator in the top bar tells you when it's done.

An analysis you trigger

The page computes nothing until you click Run analysis. Claude then re-reads every snippet of brand mention your scans collected, engine by engine, and draws a portrait out of it: vocabulary, tone, associations, comparisons. Expect 30 to 60 seconds and roughly $0.05 per analysis — this is not a refresh to repeat daily.

What each engine card contains

• Perceived positioning: the sentence that sums up what the engine believes you are.
• Perceived tone: technical, approachable, premium, institutional…
• Recurring adjectives: the qualifiers that keep coming back, verbatim.
• Associated topics: the themes your name is attached to.
• Compared to: the brands the engine spontaneously names alongside yours — often a better definition of your competition than your own.
• Highlighted strengths: what the engine retains as positive.
• Caveats / criticisms: the reservations it voices. The most uncomfortable section, and the most instructive.

Cross-engine overview

• Consensus across engines: what they all agree on — that is your real image, the one that repeats.
• Divergences detected: where the engines are not telling the same story.
• Engines with problematic perception: the ones worth a targeted correction, through content or PR.
• Alignment score: how closely the engines say the same thing about you. A low score is not a measurement error, it is a symptom: your sources contradict each other.
• Priority action: the first correction to undertake.

Not enough data

The analysis needs material to work with: an engine with fewer than three snippets is shown as « too few mentions » rather than analyzed on thin air, and if there are fewer than three snippets in total the page says so plainly and sends you back to the scans. A rarely cited brand cannot be portrayed — better to know that than to read an invented analysis.

How often to re-run it

Once a month is plenty. Perception moves slowly, at the pace of the sources the models re-read. Re-running after a PR campaign or a major publication makes sense; re-running weekly will only produce noise and cost.