Real failure-rate data for scraping APIs and anti-bot Actors

• 3,032 scraping Actors matched in the top 3,957 of the Apify Store, executing 222,295,388 runs in the last 30 days.
• The market is winner-take-most: the top 10 Actors absorb 42.0% of all scraping runs.
• Median 30-day failure rate among the 1,173 Actors with ≥1,000 runs/month is 0.06% — but the worst general-purpose crawler failed 19.55% of 11,408 runs (apify/ai-web-scraper).
• Dedicated anti-bot / Cloudflare-bypass Actors show real 30-day failure rates from 0.00% to 5.37% — up to a 97x difference vs the general-crawler median between "bypass" products that sound identical. See the report.
• 34 abandoned scrapers (no code update for 6+ months) each still run ≥1,000 times/month — together 655,159 runs/month served by unmaintained code. See the maintenance gap.

Method: full scan of the 3,957 most popular Actors on the Apify public Store API (2026-10-20), keyword-filtered to scraping infrastructure. Failure rate = FAILED runs / TOTAL runs over the trailing 30 days (ABORTED and TIMED-OUT counted separately, not as failures). Details: How we measure. Cite as: ScraperBench, "October 21, 2026 scraping infrastructure benchmarks", https://scraperbench.cloud.

Anti-Bot & Cloudflare Bypass (6)

Actors and APIs whose job is getting past Cloudflare, PerimeterX and friends. The bypass Actors include measured 30-day failure rates; commercial unlockers show claimed pricing.

Scraping API Platforms (11)

Commercial HTML-in/HTML-out APIs. Pricing is claimed from public pricing pages (Oct 2026) — we have not benchmarked them yet. Apify is the exception: we measure its ecosystem directly.

Serverless Scrapers (Measured) (15)

The highest-traffic scraping Actors on the Apify Store, with real 30-day run counts, failure rates and pay-per-result pricing from the public Store API.

Captcha & Proxy Infrastructure (3)

CAPTCHA solvers and residential proxy pools that self-built scrapers pair with. Claimed pricing only.

Data source

We scan the Apify public Store API (no authentication, public data) and keep every Actor whose name, title or description matches scraping-infrastructure keywords (scraper, crawler, bypass, unblocker, captcha, proxy, anti-bot…). This scan: 3,957 Actors scanned out of 93,039 in the Store (popularity-sorted), 3,032 matched, on October 21, 2026.

Metrics

• runs30d — total public Actor runs in the trailing 30 days (publicActorRunStats30Days.TOTAL).
• Failure rate — FAILED / TOTAL. Aborted and timed-out runs are reported separately and are not counted as failures.
• users30d — distinct users who ran the Actor in the trailing 30 days.
• Runs per user — runs30d / users30d; a proxy for how "API-like" (vs casual) the usage is.
• $ per 1k results — for pay-per-event Actors: the FREE-tier price of the primary billing event × 1,000. This is the list price, not our measured cost.
• Abandoned — no code update (modifiedAt) for 6+ months, from our independent Store scan of 3,475 Actors.

What we do NOT claim

We measure the Apify ecosystem directly. We do not yet run black-box benchmarks against commercial APIs — everything marked Claimed comes from vendor pricing pages and must be treated as marketing until verified. Failure rates measure Actor reliability on the Apify platform; they are not a like-for-like comparison with SaaS scraping APIs.

Update cadence

Data is re-fetched weekly. Every page shows its data snapshot date; historical snapshots are kept so trends become citable over time.