Methodology
Scores are measured via the official APIs of each AI platform with web search enabled. API responses closely approximate — but are not identical to — what consumers see in the apps (personalization and memory differ). We sample each question multiple times and report rates, not one-off answers. Treat scores as directional intelligence, and trends as the signal.
What we ask
Prompts belong to categories, not customers.
For each category we generate around 30 questions people actually ask when they want an app: category questions ("best fasting tracker app for iphone"), job-to-be-done questions ("app to track intermittent fasting windows"), and questions about specific apps by name. Because the questions are shared across everyone tracking that category, one run of a question serves every app in it — which is what makes daily sampling affordable.
How often
Every day. Your highest-demand questions get a daily pulse on Perplexity. Every tracked question is also assigned a fixed weekday, and on that day it runs the full matrix across all three engines with multiple samples — so each question completes three-engine coverage weekly while the data freshens daily. When an answer flips on a high-demand question, we re-ask it across all three engines the same day to confirm before alerting you.
The Visibility Score
0–100, computed over a rolling 14-day window of category and job questions.
| Component | Weight | What it measures |
|---|---|---|
| Mention rate | 50% | The share of sampled answers that named your app. This is the backbone of the score: being named at all is most of the battle. |
| Position quality | 20% | Where you land when the answer is a ranked list. First place scores 1, last place 0. An answer that names you without ranking anything scores mid-list. |
| Engine coverage | 15% | How many of the three engines named you at least once. Always measured out of three, even in a week when one engine ran less. |
| Sentiment accuracy | 15% | On questions about your app by name, how accurately and positively the answer describes it. Shown as pending until there is branded data. |
Failed engine calls are excluded from every denominator. Below 12 samples a score is marked provisional. Only high-confidence app matches count toward metrics.
Bands
| Score | Band |
|---|---|
| 0–20 | Invisible |
| 21–40 | Low |
| 41–60 | Moderate |
| 61–80 | Strong |
| 81–100 | Leading |
Direct and indirect mentions
A direct mention is the AI saying your app's name. An indirect mention is the AI citing your App Store page or website while never naming you — usually a naming clarity problem. Indirect mentions are reported separately and are excluded from the score.
Demand
Questions are grouped into demand tiers so you can prioritise. Tiers are shown as Low, Medium or High and never as absolute numbers.
Demand estimated from App Store and search signals — AI platforms don't publish prompt volumes.
What we don't do
- We never post anything on your behalf, anywhere.
- We never touch the Reddit API or scrape Reddit. Reddit only appears here when an AI engine cites it.
- We don't claim to reproduce your personal ChatGPT session. Personalization and memory differ from the API.
- We don't report single answers as facts. Every number is a rate over repeated samples.