Methodology

How we score every app and channel, and how we turn 8M+ live snapshots into a revenue forecast. No black box, no hand-waving.

Last updated 2026-05-13

1. Where our data comes from

Every public metric on bumetric comes from one of four sources, all clearly labelled in the snapshot history of each app:

Source	What we read	How often
iTunes Lookup API	iOS app metadata, ratings, icon, screenshots, current price	On-view, then weekly
Google Play Store	Android app metadata, ratings, install band, developer info	On-view, then weekly
YouTube Data API	Channel subscriber count, view count, video counts, category	On-view, then weekly
App Store Connect / Stripe Connect	Real revenue numbers, supplied by app owners who claim their listing	Pulled the moment an owner verifies

We do not scrape App Store Connect (it would violate Apple's terms). We do not buy mobile attribution data. The "estimate" you see on any page is fully derived from public signals plus our calibration set.

2. The BU Score

The BU Score on every app page is a single 0–100 number summarising overall health. It is composed of three weighted components:

Rating quality (up to 50 points). Calculated as rating × 10. An app rated 4.7 stars earns 47 points here.
Audience size (up to ~30 points). Calculated as log10(ratings_count) × 5. This rewards apps with a large rated audience but discounts the marginal gain past ~1M ratings (so a 10M-rating app does not just outscore a 1M one by 10×).
Monetisation signal (5–15 points). Awarded based on the inferred or verified monthly revenue tier. Free-with-no-IAP apps get 0; apps with verified $100K+/mo MRR get the full 15.

The final score is clipped to 100. The tier label ("Transcendent", "Excellent", "Strong", "Solid", "Emerging", "Niche") is a lookup based on score band.

Why so simple? Because every input is auditable. You can pull the same numbers from public APIs and check our math. That is the whole point of a methodology page.

3. The revenue forecast

Each app page shows a forecast monthly revenue. It is produced in two steps:

Base estimate from app metadata: ratings_count, rating velocity (week-over-week growth in reviews), category-typical monetisation curves, in-app purchase signals, ad-supported flag, subscription pricing if visible.
Calibration adjustment against our anchor set (next section). The base estimate is multiplied by the category-specific scaling factor we learned from anchor apps with verified revenue in the same niche.

The forecast does not use the LLM. Gemini is used elsewhere on the site for ASO rewrites and growth recommendations, but the revenue figure is a deterministic calculation.

4. Ground-truth calibration

Without ground-truth data, any app revenue estimate is just an opinion. We solve that with the verified anchor set:

App owners who claim their listing can verify real revenue through one of three paths: Stripe Connect OAuth (read-only), App Store Connect login (read-only), or owner-submitted screenshots that we OCR and audit.
Each verified anchor calibrates our forecast for every similar app in its category and revenue tier.
The anchor set grows weekly. The more anchors per niche, the tighter the confidence band on every other app in that niche.

This is the closest thing to a defensible moat on this site. Anyone can copy a scraper. Nobody can copy a curated verified-MRR set without doing the slow work themselves.

5. Accuracy and confidence bands

We label every forecast with a confidence band derived from how many anchors are available in the same niche:

Anchors in category	Typical accuracy	Confidence
10 or more	±15%	High
3 to 9	±25%	Medium
1 to 2	±40%	Low
0	category median fallback	Indicative only

An app with the "Low" confidence label is still useful for ranking and trend, but the dollar figure should be read with healthy skepticism.

6. How fresh is the data?

Every app page is live. When you open /p/{app_id} for an app we haven't scanned recently, our worker fetches new metadata in the background and saves a fresh snapshot before you finish scrolling. The "Last update" timestamp on the page is the moment we actually wrote that snapshot.

Cron jobs additionally re-scan high-traffic apps weekly to keep the snapshot history smooth.

7. What this doesn't tell you

Things our forecast deliberately ignores:

Paid acquisition spend. An app that buys 100K downloads at $3 CPI shows up the same as an app with 100K organic downloads. The revenue is real either way, but the unit economics are not.
Refund rate. Apple and Google publish gross revenue. Net revenue after refunds is typically 5–10% lower; we report gross.
Geographic split. We give a single global figure. Regional revenue concentration is a paid-tier feature.
Future. The forecast is a current-state estimate, not a prediction of next month. App revenue is volatile.

8. Spotted an error?

If a number on our site is wrong for an app you own, you have two options:

Claim your app and verify the real number through Stripe Connect or App Store Connect (free, 30 seconds, replaces our estimate everywhere).
Email hello@bumetric.com with a screenshot or evidence. We respond within two business days and update the page (or remove it) within a week.

The reputation of this site depends on our numbers being closer to ground truth than competing scrapers. We take corrections seriously.