How to Make Money Selling Scrapers on the Apify Store: Pay-per-Event, the Revenue Share and a 30-Day Plan

You can make money selling scrapers on the Apify Store by publishing an Actor (Apify's name for a cloud tool) and charging users per event, usually per result or per run. Apify keeps 20%, so you get 80% of the revenue, minus the platform costs your Actor burns while it runs for paying users. Payouts go out monthly from $20 via PayPal or Wise. The old rental model ended on 1 October 2026, so pay-per-event is now the way to earn.

The long version is where people lose money: the cost line nobody models, the free-plan users who don't count, the legal side of scraping, and the listing that never gets found. I run two SaaS businesses, so usage-based billing and "why is the server bill bigger than the revenue" are old friends of mine. Every earnings number below is marked as an example, because that's all it is.

What is an Apify Actor, in plain English?

An Actor is a small program that runs in Apify's cloud: it takes an input (a search term, a list of URLs, a date range), does the work (scrapes pages, calls an API, cleans data) and returns the output as a dataset people can download as JSON, CSV or Excel, or pull through an API. Think of it as a serverless function with a storefront attached.

The storefront is the Apify Store, a public marketplace where anyone can find your Actor, run it and pay per use. Apify handles the boring parts that kill most small data businesses: servers, scaling, proxies, billing, invoices and the "your card was declined" emails. You write the code and the documentation; Apify runs it and collects the money.

The market is not small. Apify's own developer page lists more than 87,000 Actors in the Store and about 4,500 community developers, and says it paid out $1.6M to developers in a single month. That last number is a marketing figure for the whole platform, not a promise about your Actor, but it tells you people really do pay for scrapers here.

One more thing makes Actors interesting right now: AI agents. Apify Actors can be called from AI clients such as Claude or Cursor through MCP (Model Context Protocol, the open standard for connecting AI assistants to tools), so your scraper can be used by an AI agent, not only by a human clicking "Start". I covered the protocol itself and the wider business models in how to make money with MCP servers; this article is the Apify-specific deep dive.

How Apify pays developers now: pay-per-event, and what happened to rental

Pay-per-event (PPE) is the main way to monetize an Actor today: you define named events (a result scraped, a page opened, a run started), set a price for each, and charge them from your code. The rental model, where users paid a monthly fee after a free trial, was fully retired on 1 October 2026, and the older pay-per-result model was deprecated and folded into PPE.

Here's how the options line up now:

The rental shutdown happened in two steps: from 1 April 2026 nobody could publish new rental Actors or change rental prices, and on 1 October 2026 the remaining ones were switched over. If you had a rental Actor and didn't set up PPE pricing in time, users now pay only the platform usage and you earn nothing. Check your Actor's monetization tab before you assume anything is earning.

Two built-in "synthetic" events make PPE easy to start with:

• apify-actor-start is charged automatically when a run starts and covers the compute of the first 5 seconds. Its default price is $0.00005 per start ($0.05 per 1,000 starts), charged once per run up to 1 GB of memory and again for each extra GB.
• apify-default-dataset-item charges for every item your Actor writes to its default dataset. This is what pay-per-result became: a price per result, now just one more event.

You can add custom events on top, like "page opened with a browser", which is how you charge more for the expensive work without overpricing the cheap work. Apify highlights one primary event on your Actor's page, so make it the one a buyer understands at a glance: "per 1,000 results" beats "per enrichment step".

Users stay in control of their spend: they set a maximum cost per run, and once a run hits that cap the platform stops charging and stops saving results beyond it. Your code can read the limit (the ACTOR_MAX_TOTAL_CHARGE_USD environment variable) and stop cleanly instead of being cut off halfway through a page. A run that ends gracefully with "limit reached, here are your 3,000 results" earns a five-star review; a crashed run earns a refund request.

The 80% share and the cost line: how your profit is really calculated

Your profit on Apify is 0.8 × revenue minus platform costs, and only users on paid Apify plans count. Revenue and costs from free-plan users are left out completely: you don't earn from them, and Apify covers their platform usage instead of billing you.

That formula hides three details that decide whether an Actor makes money:

1. Platform costs are what your Actor burns for paying users. They're measured in compute units (CU): one CU is roughly 1 GB of memory running for one hour. A lean HTTP scraper uses a fraction of a CU per thousand results; a headless browser clicking through JavaScript pages can use many times more. Proxies and storage add to the bill too.
2. The CU price depends on the user's plan. Apify's docs list $0.20 per CU for Free and Bronze users, $0.16 for Silver and $0.13 for Gold. In plan names, Starter is $19 a month, Scale $199 and Business $999. The same run costs you less when a bigger customer starts it.
3. A loss counts as zero, not as a minus. If one Actor's platform costs exceed your 80% share in a month, Apify sets that Actor's profit to $0, so a badly priced Actor can't eat the earnings of your good ones. It still earns you nothing, though.

The escape hatch is the "Pay per event + usage" toggle. Turn it on and users pay your event prices plus the platform usage their runs generate, so your profit becomes a clean 80% of event revenue. The catch: switching it on takes 14 days to take effect (switching it off is instant), and a visibly higher total price can cost you users against a competitor who absorbs the costs. For browser-heavy Actors with unpredictable costs, I'd still take it. For lean scrapers, absorbing the costs and showing one simple price usually sells better.

Worked examples: what an Actor can earn (examples, not results)

The fastest way to understand Apify economics is to run three scenarios. These are examples to show the math, not results from a real Actor; I ran the arithmetic in a script so the numbers add up. Assume paid users scrape 200,000 results a month and every CU costs the $0.20 Bronze rate.

Scenario B is the trap I see most often in Store listings: a developer copies a competitor's price without knowing what their own Actor costs to run. The break-even price is simple: cost per 1,000 results divided by 0.8. For scenario B that's 4 CU × $0.20 ÷ 0.8 = $1.00 per 1,000, exactly the price charged, so 200,000 results earn precisely nothing. Measure your CU per 1,000 results on real runs first, then price.

Now the free-user effect. Say only 60% of scenario A's results come from paid-plan users and the rest from free accounts. Revenue that counts drops to $360, your share to $288, costs to $9.60, and profit to $278.40. Same Actor, same traffic, nearly half the money gone. That's why the Store's popular Actors often keep a usable free experience but save the high-volume settings or premium fields for paying users (and say so plainly in the README).

And a realistic first month: 25 paying users who each pull 2,000 results at $3 per 1,000. That's $150 of revenue, $120 for you, about $4 of platform costs and $116 of profit. Not quit-your-job money. It is, however, real recurring revenue from a tool you built once, and it clears the $20 PayPal minimum. If you want to play with your own price and volume, the earnings calculator does the same arithmetic for any commission-style income.

When and how do Apify payouts arrive?

Apify pays developers once a month for the previous month's profit, by PayPal or Wise from $20, or by other methods such as a wire transfer from $100. Below the minimum, the money rolls over to the next month.

So money earned in October lands around 21–25 November. Plan your cash flow for that lag, especially if you're paying for proxies or third-party APIs out of pocket in the meantime.

Before the first payout you need complete billing details, a payout method and identity verification (KYC). For individuals, the name on the account must match your ID exactly, and you'll need a clear, high-resolution photo of an ID card or driving licence; screenshots and paper copies get rejected. If verification isn't done when payouts go out, that month's money waits until the next cycle. Do KYC on day one, not the day your first invoice appears.

On international wires you pay the intermediary and receiving bank fees, which Apify's docs put at roughly $10–50, a big bite out of a $100 payout, so PayPal or Wise is usually the smarter pick. And if a user gets credit compensation because your Actor misbehaved, a deduction can hit your invoice (you can dispute it). Reliability is literally money here.

What kinds of Actors actually sell?

The Actors that earn are the ones that save a specific buyer a specific, repeated job: niche data that's annoying to collect, data from sites where generic scrapers break, and tools AI agents call on demand. Generic scrapers for the biggest websites are the most crowded corner of the Store.

My rule of thumb: don't ask "what can I scrape?", ask "who pays someone to copy-paste this every week?". Some directions that fit that question:

• Vertical data for one industry. Public tender notices, building permits, job boards in one sector, real-estate listings in one country, app-store reviews for one category. Buyers are agencies, analysts and sales teams who need it weekly.
• Local and non-English sources. A well-maintained scraper for a national marketplace or directory that big developers ignore. Less competition, and the buyer often has no alternative.
• Monitoring Actors. Price changes, stock status, policy or terms changes, new listings. They run on a schedule, so one customer produces events every day instead of once.
• Enrichment and cleanup. Take a list of companies or URLs and return clean, structured fields. People pay to skip the messy part.
• MCP servers on Apify. You can host an MCP server as an Actor in Standby mode (an always-on web server mode) using Apify's Python or TypeScript templates, and charge per tool call with PPE; Apify's own MCP guide uses a $0.05 per tool call example. Store Actors also show up for AI clients through Apify's MCP server, where Actors with a higher quality score tend to rank higher in its search tool.

One practical detail for AI-agent traffic: agents that pay per request through services like Skyfire can only run Actors with pure pay-per-event pricing (no usage add-on), no Standby mode and limited permissions. If you want agent customers, keep pricing simple and permissions tight.

Picking the niche is half the battle, and it's the same skill as picking an affiliate niche: real demand, few serious competitors, buyers with budgets. My breakdown of underrated niches uses that lens, and a lot of it transfers directly to data products.

A side door that pays from week one: sell custom scraper builds as a service. Businesses post requests for "scrape this site into a spreadsheet" on freelance marketplaces like Fiverr every day. Build the job as an Actor, deliver the data, and (with the client's agreement, if the code isn't theirs) publish a generalized version to the Store. My guide to making money on Fiverr covers how gigs get found, and if the clients start asking for full workflows around the data, you're drifting into an AI automation agency, which is a fine place to drift.

How to get your Actor found: quality score and Store SEO

Store visibility comes from two things: Apify's Actor quality score, a 0–100 rating that influences your placement in the Store, and classic search optimization of your title, name, description and README, because Store pages also get found through Google.

The quality score blends eight areas, and almost all of them are in your hands:

The score is recalculated several times a day and is relative to other Actors, so it can move even when you change nothing. Don't panic over a two-point dip; watch the trend.

For search, Apify's own SEO advice boils down to a few rules I'd follow to the letter:

• The title is your best real estate. Put the key terms first, in the words buyers type: "Amazon Reviews Scraper", not "A Tool That Extracts Reviews From Amazon".
• The technical name becomes your URL. It's prefixed with your username and is a search signal, so make it readable and keyword-matched.
• The description skips boilerplate. No "This Actor…" openers; include the popular variations of your title.
• The README is your sales page. Use the keywords and their variants in headings, show example output, list limits honestly, and link to your related Actors at the end.

Here's my favorite trick, borrowed from affiliate SEO: type your target site's name plus "scraper", "API", "export" and "data" into Google and note the autocomplete suggestions. Apify's own help article recommends Google Suggest for finding phrase variations. Those suggestions are how buyers think, and "export Etsy reviews to CSV" might be a better description line than anything a developer would write. Then share the Actor's link where those buyers hang out; early links help Google find the page.

Is it legal to sell web scrapers?

Selling a scraper is legal in most places, but what it collects and how it collects it can create legal risk: breaking a site's terms of service, getting around logins or blocks, and collecting personal data without a lawful basis are the three big ones. This section is general information, not legal advice; if you plan to build around personal data, talk to a lawyer in your market.

The case everyone quotes is hiQ Labs v. LinkedIn, and it's usually quoted wrong. In the US, the Ninth Circuit ruled in 2019, and again on 18 April 2022 after the Supreme Court sent the case back in light of Van Buren v. United States, that scraping publicly visible profiles likely wasn't "without authorization" under the federal anti-hacking law (the CFAA). Scrapers celebrated. Then in November 2022 the district court found that hiQ had breached LinkedIn's User Agreement, and the case ended with a December 2022 consent judgment: $500,000 against hiQ, a permanent ban on scraping LinkedIn or using fake accounts, and an order to delete the data and the code. The lesson isn't "public data is fair game". It's "the anti-hacking law may not get you, but contract claims and fake accounts can".

Personal data is the bigger risk for anyone selling to a worldwide audience. Under the GDPR in the EU and the UK GDPR, information about identifiable people stays personal data even when it's publicly visible. In August 2023, twelve data protection authorities, including the UK's ICO, signed a joint statement on data scraping making exactly that point: organisations need a lawful reason to collect and use people's data, even when it's publicly available. The extreme example of what goes wrong is Clearview AI, which the Dutch data protection authority hit with a €30.5 million fine, announced in September 2024, over its illegal database of face photos collected from the internet.

What that means for your Actor in practice:

1. Prefer non-personal data. Prices, products, listings, company-level facts, tenders and public statistics are a far safer business than profiles of people.
2. Stay out of logged-in areas. Scraping behind a login means you've accepted the site's terms, and fake accounts were exactly what sank hiQ.
3. Don't defeat technical blocks aggressively. Respect rate limits; don't hammer small sites into downtime.
4. If personal data is involved, minimize it. Collect only the fields the use case needs, say in your README what's collected, and remind users that they're responsible for having a lawful basis.
5. Read the terms of the site you scrape, and if a site sends you a cease-and-desist, take it seriously instead of quietly carrying on.

A boring, clean data source is worth more over five years than an exciting, risky one that gets taken down in month three.

"Someone will just clone it" and "nobody will use it": the fears that stop developers

The two fears that keep developers from publishing are opposites: that a successful Actor will be copied within a week, and that a published Actor will sit there with zero users. Both are real risks, and both have practical answers that don't involve "just believe in yourself".

On cloning: yes, scraper code is easy to copy, and the Store has tens of thousands of Actors, so for big sites you'll have competitors. But the moat on Apify isn't code, it's maintenance and trust. Websites change their layouts constantly, and the Actor that's fixed within a day keeps its success rate, reviews and quality score, while the clone that broke last Tuesday slides down the rankings. My SaaS businesses taught me the same thing: customers rarely leave for a slightly cheaper copy, they leave when something breaks and nobody answers.

On "nobody will use it": this is where the sunk-cost trap bites. Developers spend six weeks polishing an Actor in private, publish it, get silence, and then either give up or keep polishing. The fix is to make the first version embarrassingly small (one site, one output, one price) and put it in front of buyers in week two, not week six. Psychologists call the pull to keep investing in something because you've already invested the sunk-cost effect, and the cheapest way to beat it is to decide in advance what result in 30 days means "continue" and what means "kill it or pivot". The plan below builds that in.

And keep a healthy distance from income screenshots. Apify's partner page says many developers earn over $3k; that's the platform's own marketing, and the median Actor earns far less. Treat a first Actor as a cheap experiment that might become a product, not as a salary.

A 30-day plan to launch your first paid Actor

The goal of the first month isn't income; it's one published, priced, documented Actor with real users and real cost data, so you can decide with numbers whether to keep going.

Two tricks make this month cheaper. First, Apify's Creator plan costs $1 a month, paid six months upfront ($6 in total), and adds $500 of platform usage for those six months, which is plenty for building and testing. It's a one-time offer, it restricts which Store Actors you can use while you're on it, and if you cancel you can't get it again, so start it the week you actually begin building, not the week you "plan to". Second, you don't need your own server at all: Apify hosts and bills. If you want a cheap always-on box for side scripts or monitoring your Actors from outside, I use InterServer for that kind of thing, and its homepage promo code gives the first month for one cent.

One rule for after launch: significant price changes (raising prices, adding paid events, changing the model) need 14 days' notice and can happen only once a month per Actor, while lowering prices is instant. So launch at a price you're comfortable keeping, and cut it if needed, rather than launching cheap and planning to "raise later".

Apify's affiliate program: a second income line from the same Actor

Apify also runs an affiliate program that pays 20% of what a referred customer pays for the first three months and 30% after that, with no time limit, up to $2,500 per customer on the starter tier ($5,000 per customer once you have 50+ active customers). For an Actor developer that's a neat second stream: the README, tutorial or blog post that sends people to your Actor can also earn on their spending.

Example, not a result: one referral on the $19 Starter plan earns $3.80 a month for three months ($11.40), then $5.70 a month, or $62.70 over the first year. A referral on the $199 Scale plan earns $119.40 in the first three months and $59.70 a month after that, $656.70 in year one. You can see why tutorials aimed at teams and agencies pay better than tutorials for hobbyists.

The fine print: payouts start at $100 and require at least three referred customers, they go out via PayPal, Wise or bank transfer, and paid advertising (Google, Facebook and other PPC) is banned, with paid-search referrals blocked automatically. One clever option: you can take your commission as Apify platform credit instead and get 15% more ($100 becomes $115), which effectively subsidizes your own Actors' test runs. If you write about scraping anyway, you can sign up through Apify and apply to the program from your account. And if you like the idea of stacking a product, a service and an affiliate line around one skill, that's the whole argument of my piece on multiple streams of income.

Mistakes that cost Actor developers money

• Pricing without measuring. If you don't know your CU per 1,000 results, you don't know whether you're making money. Scenario B above earns exactly zero.
• Forgetting free-plan users. They make your user count look great and your payout look sad. Plan the paid-user experience on purpose.
• Ignoring the maintenance load. Every site you scrape will change. Budget a few hours a month per Actor, the same way I warned about in hidden business costs: the costs you don't write down are the ones that sink small online businesses.
• Skipping limited permissions. It hurts your quality score and shuts you out of some AI-agent traffic.
• Scraping personal data "because it's public". Regulators have said clearly that public doesn't mean free to use.

The bottom line

Selling Actors on the Apify Store is one of the cleanest ways for a developer to earn from scraping: Apify hosts, bills and pays, you keep 80% of event revenue minus your platform costs, and payouts start at $20. The money is decided by three things most guides skip: knowing your cost per result before you price, building for paying users rather than free ones, and choosing a niche data source that's both in demand and legally boring.

Start with one small Actor, measure everything for 30 days, and let the numbers tell you whether to build the second one. If you'd rather earn from AI tools without running a product, my list of AI side hustles compares the other options by hours, startup cost and difficulty.