Apify Review: 4.8/5, and Where the Billing Surprises You
Apify rates 4.8 out of 5 on Capterra from 571 reviews, and 4.8 on Trustpilot from 727 reviews. Both checked 9 September 2026. It is a cloud platform for web scraping and browser automation. You run prebuilt tools called Actors from a store of 70,000+, or you deploy your own Crawlee code to the same runtime. Paid plans start at $19/month subject to change · verified 2026-09-09. The platform is SOC 2 Type II compliant and reported 99.86% Actor uptime over the last 90 days.
The billing is where evaluations go wrong. Compute and per-result Actor fees are two separate meters, and only one of them shows up where you expect it.
Apify was launched by Jan Čurn and Jakub Balada in 2015 from the Y Combinator Fellowship in Mountain View, and the team moved to the Czech Republic in 2016 (apify.com/about, checked 2026-09-09). An earlier version of this page said "founded 2015 (Prague)". That was wrong, and it is corrected here.
What do independent reviewers actually rate Apify?
Two aggregators publish a rating we could fetch on 2026-09-09. A third, G2, returned HTTP 403 to an automated request, so we are not quoting a number we could not read ourselves.
| Source | Rating | Reviews | Sub-scores | Checked |
|---|---|---|---|---|
| Capterra | 4.8 / 5 | 571 | Features 4.8 · Ease of use 4.7 · Customer service 4.6 (526) · Value for money 4.6 (533) | 2026-09-09 |
| Trustpilot | 4.8 / 5 | 727 | 90% five-star, 2% one-star | 2026-09-09 |
| G2 | not quoted | — | Blocked our fetch (HTTP 403) | 2026-09-09 |
The interesting part is which sub-scores are lowest. On Capterra, the two 4.6s are Customer Service and Value for Money, while Features scores 4.8. Reviewers like what the product does more than they like what it costs to find out.
Worth knowing before you use an aggregator as a pricing source: Capterra's own Apify pricing page listed a $29.00/month starting price on 2026-09-09, while Apify's pricing page listed $19. The rating is current. The price beside it is not.
Is Apify a good platform for production web scraping?
Yes, if your work is recurring extraction from many sites and you would rather not run browsers yourself. The platform gives you scheduling, retries, dataset storage, proxies, and webhooks behind one API. It is a weaker fit when you need raw proxy bandwidth, one-off Markdown crawls, or an air-gapped deployment.
Does Apify have good uptime and infrastructure reliability?
Apify publishes component-level uptime at status.apify.com. Over the trailing 90 days, read on 2026-09-09:
| Component | 90-day uptime |
|---|---|
| Web (apify.com) | 100.0% |
| Console | 100.0% |
| Actors | 99.86% |
| API | 99.87% |
| Storage, scheduler, webhooks, MCP server | 100.0% |
| Proxy | 99.84% |
| SERP | 99.52% |
Platform uptime is not the number that will bite you, though. Individual Actors carry their own failure rate, and Apify exposes it. The most-used Google Maps Actor, compass/crawler-google-places, logged 3,770,275 public runs in the last 30 days: 95.7% succeeded, 1.9% failed, 1.6% were aborted, 0.8% timed out (api.apify.com/v2/acts/compass~crawler-google-places, read 2026-09-09).
So the honest reliability answer is two numbers, not one. The runtime is up 99.86% of the time. Roughly one run in twenty-three does not return a clean dataset, for reasons that live in the target site and the Actor, not in Apify. Build the retry into your schedule.
What does Apify actually cost?
Plan prices, verified against apify.com/pricing on 2026-09-09:
| Plan | Monthly | Annual, per month | Included usage | CU price | Max RAM |
|---|---|---|---|---|---|
| Free | $0 | $0 | $5 | $0.20 / CU | 16 GB |
| Starter | $19 | $17 | $19 | $0.20 / CU | 64 GB |
| Scale | $199 | $179 | $199 | $0.16 / CU | 256 GB |
| Business | $999 | $899 | $999 | $0.13 / CU | 512 GB |
| Enterprise | Custom | Custom | Custom | Custom | Custom |
One compute unit is 1 GB of RAM running for one hour. The monthly fee is prepaid usage, not an access fee, and it does not roll over. Exceed it on a paid plan and the overage lands on your next invoice; exceed it on Free and access is blocked until the next cycle.
Concurrency on the Free plan: Apify's two sources disagree, and we are not picking a winner. apify.com/pricing lists 5 max concurrent runs on Free. docs.apify.com/platform/limits lists 25 concurrent Actor runs per user on Free. Both read 2026-09-09. The 16 GB memory ceiling is consistent across both. Plan against 5 if the answer matters to your architecture, and check the limit your own Console account shows before you commit.
The second meter: per-event Actor fees
Most Store Actors bill you separately from compute. Using the live event prices for compass/crawler-google-places (read from api.apify.com on 2026-09-09), scraping 1,000 places costs:
| Your plan | Discount tier | Price per place | 1,000 places |
|---|---|---|---|
| Free | FREE | $0.004 | $4.00 |
| Starter | BRONZE | $0.003 | $3.00 |
| Scale | SILVER | $0.002 | $2.00 |
| Business | GOLD | $0.0015 | $1.50 |
Add-ons stack on top of that base event. Contact enrichment adds $0.002 per place on Starter, social profile enrichment $0.008, email verification $0.004, and an AI competitor analysis $0.02 each. Reviews and images are $0.0005 apiece, which is cheap per unit and expensive per venue once you pull 200 reviews from each of 1,000 places.
Two consequences worth internalising. Your plan changes your per-result price, so a cost estimate from a Free-plan trial overstates a Starter-plan bill by a third. And the Actor's own store page advertises "from $1.50 / 1,000 scraped places" — the Business-tier rate, not the one a $19 customer pays. Full method in our Apify pricing breakdown and pay-per-event guide.
Residential proxy bandwidth is a third meter again, at $8/GB on Free and Starter, $7.50 on Scale, $7 on Business. A 10,000-request run at 50 KB per response is 500 MB, or $4 on Starter. That is not included in either of the numbers above.
Should I use Apify or self-hosted Crawlee?
Crawlee is Apify's own open-source crawling library, and it is genuinely free. If you already run containers, have a scheduler, and can absorb proxy procurement, self-hosted Crawlee wins on unit cost and on control. It loses on the hours you spend keeping it alive.
Apify is the managed version of that same library plus a marketplace. You are paying for the storage contracts, the run history, the retry semantics, and the 70,000 Actors you did not have to write. At $19 a month, the platform is cheaper than one engineer-hour; the question is never the subscription, it is the per-result fees at your volume.
The dividing line we would use: if a Store Actor already covers your target sites, take Apify. If your targets are unusual enough that you are writing custom code anyway, price out both, because at that point you are buying operations rather than scrapers.
What changed recently, and when
Most reviews of this platform are rewrites of the vendor's marketing page from a year ago. Here is what moved, with dates we checked ourselves.
| Date | What changed | How we know |
|---|---|---|
| 2026-09-09 | autofacts/shopify switched from a $5/month rental to pay-per-event at 03:25 UTC | api.apify.com pricingInfos |
| 2026-09-15 | bebity/linkedin-jobs-scraper leaves $29.99/month rental for pay-per-event | scheduled pricingInfos entry |
| 2026-09-17 | anchor/email-phone-extractor leaves $30/month rental for pay-per-event | scheduled pricingInfos entry |
| 2026-09-21 | trudax/reddit-scraper, already flagged deprecated, leaves $45/month rental | scheduled pricingInfos entry |
| Current | Apollo scrapers are withdrawn from the Store | apify.com/scrapers/apollo: "Apollo scrapers are not available right now because of the company's strict terms of use" |
| Current | Apify's MCP support is a hosted endpoint at mcp.apify.com, not an Actor you run | Apify MCP docs |
The pattern behind the first four rows matters more than any single Actor. Monthly rental pricing is disappearing from the Store. We sampled 2,651 Actors from the public Store API across ten offsets on 2026-09-09: 2,613 were pay-per-event, 38 were free, and none was a monthly rental. Any review, ours included, that quotes "$X per month for this scraper" is quoting a model on its way out. Our Reddit scraper roundup and Google Maps roundup track the replacements.
One more discrepancy to hold lightly: Apify's store page advertises "70,000+ ready-to-run tools", while api.apify.com/v2/store reported 57,177 total on 2026-09-09. The two count different things, the public searchable listing being the smaller. We quote Apify's published figure and flag the gap rather than inventing a precise number.
Where Apify is the wrong choice
This is where a first-party review earns its keep, so we will be direct about it.
If you are buying proxies, buy proxies. Bright Data's network is the market leader for raw IP access, and if your only requirement is high-quality residential bandwidth to feed a scraper you already wrote, Apify's $8/GB is not the reason to be here. See Apify vs Bright Data for the split.
If you want a URL turned into clean Markdown for a RAG pipeline, Firecrawl is purpose-built for that and has a simpler mental model than Actors and datasets. Details in Apify vs Firecrawl. Plenty of teams run both.
If compliance forbids third-party processing, no amount of SOC 2 Type II helps you. Self-hosted Crawlee on your own infrastructure is the answer, and it is Apify's own library, so you are not fighting the vendor to get there.
If the target site's terms forbid it, the platform will not save you either. Apollo is the live example: Apify pulled those scrapers rather than fight the terms. Our notes on the legal side of scraping cover the boundary.
What breaks in practice
Five things we have watched go wrong, in rough order of how much money they cost.
- Per-result fees dwarf compute. A run that burns $0.10 of CUs can carry $3.00 of place-scraped events. The usage chart in Console shows the first number prominently. Read the Actor's Pricing tab before you schedule anything.
- Memory defaults are set for the worst case. Store Actors routinely default to 2–4 GB when 512 MB finishes the job. Since a CU is GB-hours, that is a 4× to 8× multiplier on compute until you override it. See cutting Apify costs.
- Actors go quiet before they go broken. An Actor with no commits in six months usually keeps returning 200 and empty arrays after the target site changes its markup. Sort the Store by users, then check the last-modified date.
- Over-quota behaviour flips at $19. Free blocks you until next month. Paid plans accept the overage and invoice it. Set a billing limit in Console before your first schedule, not after your first surprise.
- The rental-to-per-event switch changes your unit economics overnight. Four Actors change model in the next twelve days alone. If you budgeted a fixed monthly line item, re-derive it per result.
Our score, and how we got it
This is one operator's score against five stated criteria, not an aggregate of user reviews. For aggregate sentiment, use the Capterra and Trustpilot numbers at the top of this page. Whole and half points only, because a second decimal place would be false precision on a qualitative judgment.
| Criterion | Score | Why |
|---|---|---|
| Developer experience | 5 / 5 | Crawlee, a CLI, and a versioned REST API match how backend teams already ship. |
| Catalogue and time to first result | 4.5 / 5 | 70,000+ Actors means most common targets are a config problem. Quality varies by maintainer. |
| Reliability | 4.5 / 5 | 99.86% platform uptime; per-Actor run success closer to 95.7%. |
| Billing transparency | 3 / 5 | Three meters, plan-dependent per-event rates, and a store page quoting the enterprise-tier price. |
| Documentation and support | 4 / 5 | Docs are thorough. Two Apify pages still contradict each other on Free-plan concurrency. |
| Overall | 4 / 5 | Strong product, weakest at telling you what it will cost. |
Our 4 sits below the 4.8 aggregate because we weight billing transparency heavily and Apify loses points there. Reviewers on Capterra reach a similar conclusion by a different route: Value for Money is one of the two lowest sub-scores they give it.
Run this before you commit. Take the Free plan, pick the Actor you would actually use, and run 20 records against your real target URLs. Read both the compute charge and the event charge on that run, multiply by your monthly volume, then double it. If the number is still acceptable, the rest of this page is detail.
Start on the Apify free plan → · Check the current plan prices →
No credit card is required for the free plan, and Free blocks rather than bills when the $5 runs out, so a trial cannot generate an invoice (apify.com/pricing, checked 2026-09-09).
Yes. Apify was launched in 2015 through the Y Combinator Fellowship in Mountain View and has been headquartered in Prague since 2016. It is SOC 2 Type II compliant, publishes component-level uptime at status.apify.com, and holds 4.8/5 on Capterra (571 reviews) and 4.8/5 on Trustpilot (727 reviews), both checked 9 September 2026.
The free plan is $0 with $5 of included platform usage. Starter is $19/month ($17 billed annually) with $19 included, Scale is $199, and Business is $999, verified on apify.com/pricing on 2026-09-09. Those figures cover compute only. Most Store Actors add a per-event fee, and residential proxy bandwidth is billed separately at $7 to $8 per GB.
Because they are stale. Capterra's Apify pricing page still listed a $29.00/month starting price on 2026-09-09, while Apify's own pricing page listed $19. Check apify.com/pricing rather than an aggregator when the number matters.
Apify advertises 70,000+ ready-to-run Actors on its store page. The public store API reported 57,177 total on 2026-09-09; the two figures count different things, with the API covering the searchable public listing. Use 70,000+ as Apify's published figure and treat any more precise number with suspicion.
The platform is. Apify's status page showed 99.86% Actor uptime and 99.87% API uptime over the trailing 90 days on 2026-09-09. Individual Actors are the weaker link: the most-used Google Maps Actor succeeded on 95.7% of 3.77 million public runs in the last 30 days. Budget for retries at the Actor level, not the platform level.
Crawlee is Apify's open-source crawling library, free to run anywhere. Apify is the managed platform around it: hosted execution, the Store marketplace, proxies, scheduling, storage, and billing. Choose Crawlee if you already operate containers and want the lowest unit cost. Choose Apify if a Store Actor already covers your targets.
Rarely, and less each month. Of 2,651 Actors sampled from the public store API on 2026-09-09, 2,613 used pay-per-event pricing and 38 were free; none used a monthly rental. Four Actors we track leave rental pricing between 2026-09-09 and 2026-09-21. Treat any monthly Actor price you read as provisional.
When you only need raw proxy bandwidth, where Bright Data leads on network scale. When you want a URL converted to clean Markdown for a RAG pipeline, where Firecrawl is purpose-built. And when compliance forbids third-party processing, where self-hosted Crawlee on your own infrastructure is the answer.
Common mistakes and fixes
The bill is larger than the compute usage chart suggested
Most Store Actors charge per event on top of compute. Open the Actor's Pricing tab, add up the events your run triggers, then set a spending limit in Console → Billing.
An Actor that worked last month now returns empty results
Check the Actor's last-modified date and its issues tab. If the maintainer has gone quiet, move to the highest-adoption alternative in the same category before your schedule runs again.
The concurrency limit does not match what the docs said
Apify's pricing page and its platform limits page disagree on the Free plan. Trust the limit shown in your own Console account.