Skip to main content
Prefer reports seven core metrics: Visibility Score, Share of Voice, Avg Position, Citation Share, Rank, Delta and Sentiment mix. Each is computed from stored AI answers on ChatGPT, Gemini, Google AI Overviews, Google AI Mode and Perplexity, recomputes when you filter, and shows its change vs the previous run. Last updated 3 October 2026 Prefer reports 7 core metrics, and every chart, card and heatmap in the app is built from the same definitions. Learn them once and every Prefer report reads the same way. Each number comes from real AI answers that Prefer stored in full, so you can always click back from a metric to the answer behind it.

​
The 7 metrics, in one table

Every Prefer number is one of 7 metrics, each defined in the Prefer metrics docs in a single sentence. The Overview dashboard adds one more count to its KPI strip: Pages Mentioning You. The strip shows 5 cells (AI Visibility, Share of Voice, Citation Share, Pages Mentioning You and Sentiment), each with its value, a delta chip, a sparkline and a plain sub-line on hover.

​
Where the raw data comes from

Prefer builds every metric from stored answers, not from estimates or panel data. For each prompt in a tracking run, Prefer sends the question to each enabled engine and keeps 4 things: the complete answer text, every brand named in order of appearance, every cited URL, and a sentiment label per brand mention. Every self-serve plan covers Starter, Growth and Scale on the same 5 engines, so the metric set does not change by tier. Because the brand list keeps its order, Avg Position and Rank fall straight out of the stored answers. Because the cited URLs are kept, Citation Share and the source views do too. Nothing on the dashboard is a modelled guess about traffic.

​
Scope: the filter decides the number

Change a filter in Prefer and every metric on the page recomputes for the narrower set of answers. The Filter Bar at the top of every Monitor page sets the scope:
  1. Date range
  2. Engine (pick one or several)
  3. Topic
  4. Persona
  5. Region
  6. Tag
A “N runs, last date” summary next to the filters tells you how much data sits in scope. Small scopes give jumpy numbers, so check that count before you read a swing as a trend.

​
How to read a change

Prefer labels every delta “vs previous run”, never “vs yesterday”, because a run is Prefer’s unit of time. If you trigger two runs in one day, you see two deltas. For a longer comparison, set the date range and compare two runs directly. Read a change in 4 steps:
  1. Check the engine split on the Platform performance heatmap, since a blended score can hide a large move on 1 model.
  2. Open the prompts that moved and read the full answers behind them.
  3. Compare Citation Share to see whether a source change caused the move.
  4. Confirm the run count in scope, so a 2-run window is not read as a trend.
The Answer Engine Insights module shows all of this on one screen, and competitive analysis puts the same metrics side by side for each rival. To see your own numbers once, run a free AI visibility audit.

​
Frequently asked questions

No. In Prefer, Visibility Score is the percentage of tracked answers that name your brand at all. Share of Voice is your brand mentions divided by all brand mentions in those answers. You can appear in many answers and still hold a small share if every answer lists eight rivals next to you.
Because every Prefer metric is computed inside the scope you set. Pick one engine, one topic or one region and the whole page recomputes for that narrower set of answers. There is no hidden global filter and no cached number that ignores your selection.
Both. Prefer computes each metric per engine and blended, and deltas work the same way. That lets you catch a drop on ChatGPT while your blended Visibility Score looks flat, which is why the docs suggest comparing against a competitor one engine at a time.

​
Also asked as

  • What does the Prefer dashboard measure?
  • How does Prefer calculate Visibility Score?
  • Which KPIs can I track in Prefer?

​
Sources