# Apify Store after launch: the quality score, the daily test, and the 31-day path to "deprecated"


# Apify Store after launch: the quality score, the daily test, and the 31-day path to "deprecated"

*Not legal or investment advice. Mechanics come from docs.apify.com (Actor quality score, Actor testing in Apify Store, Automated tests for Actors), archived 2026-10-15 (Asia/Shanghai). Apify says the scoring algorithm keeps evolving, so re-open the pages before you rely on any detail. This article links none of our Actors.*

Publishing an Actor is the start, not the finish. Two systems keep judging it afterwards: a **quality score** that feeds search ranking, and a **daily automated test** that can label it `under maintenance` and eventually `deprecated`. Neither needs you to do anything wrong. A target site changing its HTML is enough.

## 1) Lead answer

1. **Quality score: 0–100**, shown in Console > Insights > Actor quality. It correlates strongly with ranking in Apify Store search **and** in the Apify MCP server's `search-actors` tool used by AI agents. No position is guaranteed. Console search is personalized per user.
2. **It moves without you.** It is recalculated several times a day, it is relative to other Actors, and the algorithm changes. Recommendations may linger after you've fixed the issue.
3. **Daily Store test:** Apify runs your Actor with its **prefill** input. The run must end **Succeeded**, with a **non-empty default dataset**, **within 5 minutes**.
4. **Failure calendar:** 3 consecutive failing days → you're notified and the Actor is marked `under maintenance`. 14 more days → a second notice. 14 more → `deprecated`.
5. **Recovery:** fix and rebuild, and the test picks it up within 24 hours. Or, if the target site was at fault, pass the **majority** of test runs over the next 7 days and it is marked healthy automatically.
6. **Actors that need auth, or that legitimately run more than 5 minutes, always fail.** Contact support to ask for an exclusion. Don't game the prefill instead.

## 2) The eight categories and what you actually control

| Category | Docs say it reflects | Your lever |
|---|---|---|
| Reliability | Run success rate + passing automated QA tests ("poor reliability significantly impacts") | Input schema validation; pass the daily test |
| Popularity | Users running, saves, return usage | A clear use case. You can't honestly buy this |
| Feedback and community | Reviews/ratings from repeat runners; negative reviews hurt | First-run experience; reply to issues |
| Ease of use | Title, description, input field descriptions, README | Self-explanatory inputs |
| Pricing transparency | How predictable cost is; PPE named as the predictable model; Bronze/Silver/Gold discounts suggested | Explicit events, honest per-result wording |
| Trustworthiness | **Limited permissions**. Their absence "will negatively impact" | Use limited permissions unless support approves more |
| History of success | Your track record across Actors | Slow, cumulative |
| Congruency | Title, description, README, input/dataset/KVS schemas use consistent terms | One vocabulary everywhere |

**Weights: Not verified.** Apify doesn't publish them, so don't write "pricing is 20% of the score" or anything like it.

## 3) The daily test, step by step

- **Prefill is the test input.** It's the `prefill` in your input schema, not `default`. A prefill that scrapes 10,000 pages or needs a cookie will fail every day.
- **Keep the default run under 5 minutes, with at least one dataset item.** A run that succeeds but pushes nothing still fails.
- Treat the prefill as the demo, the test, and the first-run experience all at once. The same run that decides `under maintenance` is what a new user sees first.

**Hypothetical calendar** (first failing test on 2026-10-15):

| Date | Event |
|---|---|
| 2026-10-15 | Day 1 failing |
| 2026-10-17 | 3 consecutive days → notified, `under maintenance` |
| 2026-10-31 | +14 days still failing → second notice |
| 2026-11-14 | +14 more → `deprecated` |

So you have about a month from first failure to deprecation, but the public `under maintenance` label arrives on **day 3**. The intro of the same page says "broken for more than a month." Whether that and the 3+14+14 count run on the same clock: **Not verified**.

**Majority over 7 days:** 4 of 7 passing (0.571) qualifies and 3 of 7 (0.429) doesn't, assuming one test per day. Whether the count is "7 runs" or "all runs in the window": **Not verified**.

## 4) Your own tests (so you hear first)

Apify's guide recommends the public Actor Testing Actor:

1. Create **1–5 test tasks**, including one with the default configuration.
2. Set a low `maxItems` to save credits; run heavy-data tests less often.
3. Assert status `SUCCEEDED`, no `ReferenceError`/`TypeError` in the log, retries below a threshold, runtime within a range, a non-empty dataset, and the fields in the expected shape.
4. Schedule it, then review the results **weekly**.

A scheduled test that mirrors the Store test (prefill, 5-minute budget, non-empty dataset) is the cheapest early warning you can get.

## 5) Honest language while the score or status is bad

Say:
- "Temporarily under maintenance: the target site changed; fix in progress" (if that's true).
- "Requires your own login; excluded from automated tests by arrangement with Apify" (only after support confirms).

Don't say:
- "Top-ranked" or "#1 in Store." Ranking is personalized in Console and not guaranteed anywhere.
- "Quality score 90+" as a lasting badge. It moves daily, relative to other Actors.
- "Always works." The daily test is public evidence against you the moment it fails.

## 6) Self-check

- [ ] Prefill run finishes Succeeded, with ≥1 dataset item, in < 5 min.
- [ ] The input schema validates (Reliability).
- [ ] Limited permissions are on (Trustworthiness).
- [ ] Title, README, and schemas use the same terms (Congruency).
- [ ] Your own scheduled test runs and is reviewed weekly.
- [ ] Auth/long-run Actors: exclusion requested from support, not hacked around.
- [ ] Listing copy makes no ranking or score promises.

## Not verified

- Category weights inside the quality score.
- How "majority of test runs in the next 7 days" is counted.
- Whether "broken for more than a month" and the 3+14+14 schedule are the same clock.
- Whether a `deprecated` Actor returns to healthy after a fix, or must be republished.
- Whether Actors excluded from the daily test lose Reliability points.