Methodology

How apifystats measures actors

Every number on this site comes from a daily census of the public Apify Store catalog. Nothing here is paid placement, and no actor can opt in or out. This page defines each metric exactly, including the full formula behind the A–E grade.

The A–E grade

The grade compresses four measured components into one letter. It is a letter and not a score on purpose: the inputs are too coarse to defend a claim like “87.3 beats 86.9”, but they are plenty to defend “A beats D”. Every component is a number shown elsewhere in the same table row, so you can always audit a grade by reading across.

Who gets graded

An actor is graded only if it has at least one direct quality signal: 100+ runs in the last 30 days (so the success rate is meaningful) or 3+ reviews. Actors with neither show “—” — no evidence, no grade. As of the current snapshot that is roughly one in six actors in the catalog.

The components

ComponentPointsHow it is computed
Reliability0–40 30-day run success rate × 40. Requires 100+ runs in the window; with fewer, this component scores a neutral 20 rather than rewarding or punishing silence.
Rating0–30 The review rating smoothed by volume: we add 25 phantom reviews at 4.5★ before averaging — (rating×reviews + 4.5×25) / (reviews + 25) — so a 5.0 from two reviews cannot outrank a 4.7 from five hundred. The smoothed rating maps linearly from 3.0★ = 0 points to 5.0★ = 30 points. No reviews scores the 4.5 prior.
Adoption0–15 Log-scaled monthly users: log10(users30 + 1) / 4 × 15, capped at 10,000+ users. Log scale, so 100 users is worth real credit and the giants do not drown everyone else.
Freshness3–15 Source last modified under 30 days ago = 15, within 180 days = 9, older = 3. Unknown scores the middle 9.

The letters

Points sum to a 0–100 score: A ≥ 90 · B ≥ 80 · C ≥ 65 · D ≥ 50 · E below. The cuts are fixed, not graded on a curve, so a grade means the same thing on every page. On current data about 1% of graded actors earn an A and roughly half sit at C — C is normal, not bad.

Worked example

An actor with a 96.4% success rate, 4.7★ from 120 reviews, 3,400 monthly users, updated 12 days ago: reliability 38.6, rating 25.0 (smoothed 4.67★), adoption 13.2, freshness 15 — total 91.8, grade A.

What the grade deliberately ignores

Price is a preference, not a quality — it gets its own axis, the price gauge, so you can read expensive-but-excellent and cheap-but-shaky at a glance instead of having them cancel out inside one number. Momentum is excluded because it is volatile and would make grades jump week to week. There are no editorial adjustments and no manual overrides.

The other metrics

Price and the gauge

Prices are normalized to $ per 1,000 units (results or events) or $ per month for rentals. The 4-bar gauge compares an actor only against actors with the same pricing model in the same view, by quartile: 4 bars = cheapest quarter, 1 bar = most expensive quarter. It needs at least 5 comparable actors to show at all. Change the filters and the gauge recomputes — it is always relative to what you are looking at.

Success rate

Succeeded runs divided by total runs over the last 30 days, shown only for actors with 100+ runs. Failed runs are not always the actor’s fault — bad inputs count too — but a persistently low rate is worth knowing before you depend on one.

Users and momentum

User counts are unique users over 7-day, 30-day, and lifetime windows, as published by the store. Apify updates them in batches, so they can lag by days — read them as directional. Momentum compares this week’s user pace against the actor’s own monthly average, normalized by the store-median ratio (recurring users make the raw week-to-month ratio exceed 1 for almost every actor, so the median is the honest zero). 0% = a typical week.

Freshness and age

Freshness buckets the days since the actor’s source was last modified: under 30 days, 30–180, over 180. Age counts from first publication.

The data

The census snapshots the public store catalog daily. It sees what a logged-out visitor can see: titles, categories, pricing, user counts, run stats, ratings. It does not see private actors, actual revenue, or run contents. Store search pagination caps what any crawl can reach, so a small share of the long tail is absent from any given snapshot; the actors you can find in store search are the actors you will find here. This project is not affiliated with, endorsed by, or sponsored by Apify.