An Apify Actor earns money when someone on a paid Apify plan runs it and triggers a charge you defined: a result delivered, an item processed, a run started. Apify keeps 20% of that revenue and subtracts the platform costs the run generated. What is left is yours, before your own expenses, your time and taxes.
That structure explains why no reliable universal earnings figure exists. The outcome depends on price, volume, cost per run and how much of the usage comes from paying accounts. Anyone quoting a “typical” number is guessing.
What you can do is test the economics of one idea before building it. This article does that with a single illustrative product, a supplier price-and-stock checker, and marks where documented rules end and assumptions begin.
How Selling an Apify Actor Actually Works
An Actor is a program that runs on Apify’s cloud, takes an input and produces an output. Our running example is an Actor that visits a public supplier catalogue and returns a file with SKU, product name, price, stock status, source URL and a checked-at timestamp. The imagined customer is a small online retailer that needs that file refreshed regularly. Treat it as an illustration, not a validated business. Whether anyone would pay for it, and whether you have the right to collect and reuse that supplier’s data, are questions you must answer yourself before writing code.
Four terms are easy to confuse:
- A user is an account that runs your Actor.
- A run is one execution with one input.
- An output is what the run returns, such as one row per product.
- A billable event is the action you decide to charge for, such as “one product row delivered.”
Apify’s documentation lists two pricing models for Store Actors: pay per event and pay per usage. Under pay per event (PPE), you define the events and their prices, and customers pay for those. Pay per usage charges users only the platform usage the run generates. The profit formula in the documentation is written for PPE, so that is the model this article uses. Some older interviews describe rental-style arrangements. They belong to an earlier period, and John Cole’s 2026 profile describes selling by result with no monthly rental, which matches the current model.
Under standard PPE settings, the documented formula is:
profit = 0.8 × revenue − platform costs
Revenue is what users were charged through the event-charging mechanism. Platform costs are the compute, proxy, storage and data-transfer charges the runs generated. Two details change who pays what:
- Free-plan users. Only revenue and costs from paid-plan users count toward your profit. Free users generate no revenue for you, and Apify covers their platform costs.
- Optional usage pass-through. PPE has a “Pay per event + usage” setting that passes platform costs on to users. Apify says it hurts transparency and the Actor quality score, and suggests it only while you work out event prices. It takes 14 days to take effect when you switch it on. In Standby mode, users always pay platform usage on top of events.
Selling your own Actor is a different activity from earning a referral commission by sending people to Apify. Apify runs a separate affiliate programme with its own terms, and this article is only about selling your own product.

conceptual flow · 5 steps, no real figures
Illustration 1. A retailer needs catalogue updates, an Actor produces structured results, eligible paid usage becomes billable events, and the creator receives revenue after commission and platform costs.
How Much Can You Realistically Earn?
The honest answer is a range with a very wide spread, and the public evidence cannot tell you where a beginner lands in it.
Consider what is available. Apify’s own promotional material states large total payouts and says many developers earn over $3,000. It does not define the period or show the distribution. Dividing a platform total by a head count would not produce a median or a beginner’s expectation, so this article does not attempt it. What we do have are named accounts published by Apify itself. They are useful as illustrations, but they are not independent audits.
Case A: Guillaume Lancrenon, 2024. In an Apify blog interview dated October 2, 2024, Guillaume, a software engineer and solopreneur, says Store sales were bringing in more than $2,000 a month. He mentions earning roughly $500 a month from other side projects before. Be careful with that sentence: it is a self-reported figure from two years ago, it does not say which Actors produced it, and it contains no expense account, so it is neither profit nor a 2026 number. His tools target prospecting and lead generation, work he understood from building his own products. The practical details matter more than the headline. He describes using the Issues tab to talk to users, and he says failures usually trace to his own code or to target websites changing, not to Apify’s infrastructure. So a paid tool carries a continuing duty to answer users and chase source changes. The interview does not show what an average creator earns.
Case C: John Cole, 2026. An August 27, 2026 builder spotlight reports that John Cole took second place in Apify’s $1 Million Challenge and won $20,000. That is a prize for a portfolio of nearly 100 Actors, judged on factors such as monthly active users and Actor quality. It is not monthly revenue and says nothing about a recurring product’s income. What carries over is his method. He says nearly everything he ships starts as a fix for a problem he hit in another project, and he runs a test script on every new Actor before release. The same article says his profile shows a run success rate above 99% and an issue response time of 16 hours. Those are strong operating habits. They are not earnings.
Keep the distinction between a prize, a user count and net income in mind whenever you read a headline figure.
Work backwards from what you keep
Start from the formula above and treat its result as calculated earnings, a platform-side number. Business profit sits further down:
- Gross event revenue is what paying users were charged.
- After commission, you hold 80% of it.
- After platform costs, you reach the number Apify calculates for your payout.
- After external costs, you reach something closer to profit. Apify’s Store terms put third-party costs on you, including AI model APIs and data-provider fees, and say your pricing must account for them.
- After your time and taxes, you reach what the project is actually worth to you.
One more documented rule affects the outcome: an Actor with negative profit in a month counts as $0 for that month and does not reduce the profit of your other Actors. It limits how much one bad month hurts. It also means a money-losing Actor can look harmless in the payout total while still consuming your attention.
A worked example at $5 per 1,000 paid results
Illustrative example. The price and the cost below are assumptions, not market rates or measured run costs.
Assume 10,000 chargeable results come from eligible paid-plan usage:
- Price: $5 per 1,000 results, so gross event revenue is 10 × $5 = $50.
- Commission at 20%: $50 × 0.20 = $10, leaving $40.
- Assumed platform costs: $15 (that is $1.50 per 1,000 results).
- Calculated earnings: $40 − $15 = $25.
To reach $500 in those earnings with identical economics, you would need 20 times as much: 200,000 paid results. That is a mathematical requirement, not a sales forecast.
Count development and maintenance time
Suppose the first version takes 10 hours, which you value at $25 an hour. That is a $250 opportunity cost, meaning the value of your time rather than cash you paid out. At $25 a month in calculated earnings, the simplified payback period is 10 months, assuming no further costs and no change in income.
Now add support. Two hours a month, valued at the same rate, equals $50 of your time against $25 of earnings. The project loses $25 a month in labour terms, and the initial $250 is never recovered at that income. If you enjoy the work or it advances other goals, the choice may still be reasonable, but call it what it is. Cash outlay and time cost answer different questions: the first tells you whether you can afford the experiment, the second tells you whether the experiment pays you fairly. Our guide to what freelancers in a neighbouring field actually earn, n8n freelance earnings, is a useful reference point for setting your own hourly value.

illustrative example · typed assumptions, not Console data
Illustration 2. In this illustrative example, $25 of calculated earnings is not take-home profit: external expenses, your time and taxes still come out of it.
Find a Problem People Need Solved Repeatedly
Identify the buyer and the recurring job
Fill in this sentence before anything else:
I help [buyer] get [output] every [interval] so they can [business task].
For the example: I help small online retailers get a clean file of supplier prices and stock status every morning so they can update their own listings before orders arrive. If you can’t complete the sentence with a specific buyer, the project is not ready. “Anyone who needs data” is not a buyer.
The word “every” carries the economics. A one-time export earns once. A task repeated weekly or daily earns again each time it runs. That explains why the strongest accounts above revolve around recurring jobs like prospecting, flight price lookups and monitoring.
Compare existing Actors before building
Search Apify Store for direct alternatives and read a handful closely. Note the output fields, required inputs, stated limitations, pricing, recent issues and how the author replies to complaints. A gap can be concrete: a field everyone omits, an output format that imports cleanly into a retailer’s spreadsheet, a visible failure pattern in reviews. Apify’s publishing terms explicitly allow similar functionality, so competing is legitimate, but copying another creator’s code, README or description is prohibited.
An empty search result proves very little. It may mean no one has asked for the thing. Treat the absence of competitors as a warning to verify demand, not as evidence of an opening.
Test the proposal with a sample output
Before building, produce a mock or hand-made sample of the file and put it in front of real prospects. Ask how they do the task today, how often, what it costs them now in hours or subscriptions, which fields they need, and why existing tools fall short. A short message you can adapt:
Hi [name], I’m exploring a small tool that checks [supplier]’s catalogue each morning and returns SKU, price, stock status and a timestamp as a CSV. Here is a sample with ten rows. How do you handle this today, and what would make a file like this worth paying for? I’m not selling anything yet; I’m trying to find out whether it solves a real problem.
Polite interest (“sounds cool”) is cheap. Paid validation looks like a prospect sharing their current workflow, naming a price they would accept, or asking when it will be ready. Even a modest commitment tells you more than enthusiasm.
Caleb David’s story shows why this step pays off. In a May 6, 2024 Apify interview, the freelance engineer describes taking custom scraping projects from clients while also publishing his own Actors in the Store. He says over 50 Actors came out of this work, some now embedded in clients’ workflows. His Twitter scraper reportedly passed 10,000 users, over half regularly active. He also says it had months with no income at all, a period where “you might feel as if all your efforts have been watered down.” No monthly earnings figure is given, and a user count is not a count of paying customers. It also dates from the rental era, so recheck platform mechanics before copying anything from it. The lesson holds regardless: real client requests reveal problems worth turning into products, some work stays bespoke, and a good Actor can wait a long time for usage.

illustrative example · invented sample rows, no real customer
Illustration 3. A defined buyer and a repeatable job come before development: the sample file should answer the retailer’s inventory-update decision.
Measure Costs Before You Set a Price
Prices set by feel often lose money. Run three tests first:
- A small routine input, such as 50 SKUs from a normal catalogue page.
- A larger input, such as 2,000 SKUs, to see how cost scales.
- A difficult or incomplete case: missing pages, changed layouts, out-of-stock items, a 404.
For each run, record input size, usable outputs, runtime, platform costs, external API costs and errors. Platform costs appear in Console. Apify’s documented unit rates (compute units, proxy traffic, storage operations) vary by customer tier, so your cost per run can differ by who runs it, and higher tiers cost you less per unit.
Use this worksheet and fill it only with figures you measured:
| Run | Input size | Usable outputs | Runtime | Platform cost | External cost | Errors / retries |
|---|---|---|---|---|---|---|
| Small routine | measure | measure | measure | measure | measure | measure |
| Larger input | measure | measure | measure | measure | measure | measure |
| Difficult case | measure | measure | measure | measure | measure | measure |
Then calculate:
cost per 1,000 usable results = (all costs, including failed runs and retries ÷ usable results) × 1,000
A quick illustration, using assumed numbers: a run costs $12 in total and returns 9,000 usable rows out of 10,000 attempted. Cost per 1,000 usable results is $12 ÷ 9 = about $1.33, higher than the $1.20 you would have calculated per 1,000 attempted. Retries and failures raise the real figure, so count them.
Some costs scale with volume (compute time, proxy traffic, storage writes), while others do not (the time you spend fixing a broken selector). Apify’s guidance also covers charging for inputs that look valid but fail, such as a URL returning a 404, since the platform resources were still spent. It adds that this may not recover the full cost, so test invalid inputs too.
Set a Price Customers Can Understand
Choose a billing event that matches visible value. For the example, “one product row delivered with price and stock” fits better than “one run started,” because the retailer can see and count the rows. Apify’s own advice is to keep the number of events small and to make each event correspond to something the user can see in their results.
Compare prices against alternatives on a like-for-like basis. One Actor might charge per product, another per page, another per run, and each is only comparable once you translate it into the customer’s total monthly cost for the same task.
A sample translation for the buyer, again illustrative: a retailer checking 2,000 SKUs weekly generates about 8,000 results a month. At $5 per 1,000, that costs them $40 a month. At the assumed $1.50 per 1,000 platform cost, your calculated earnings on that customer are 0.8 × $40 − $12 = $20 a month. Reaching $500 in calculated earnings at that rate would take 25 such customers. This is arithmetic about scale, not a prediction.
Check the pricing floor
Assumption-based illustration for the standard PPE formula only. If platform costs are $1.50 per 1,000 results and the commission is 20%, the break-even event price is $1.50 ÷ 0.80 = $1.875 per 1,000 results. At that price the calculated earnings are zero. The floor leaves out external API costs, your labour and taxes, so a sustainable price needs margin above it. The figure is a worked example and not an Apify market rate, and it changes when the cost setting changes, for example under the usage pass-through option or in Standby mode.
Free users, cheap successful tests and a busy dashboard do not prove profitability. Free-plan usage is excluded from your profit entirely. Apify’s documentation also warns against setting high prices only for free-plan users, because people who test on enough data often become paying users after they see the value. A high price may be justified only to deter abuse or to cover third-party costs, such as LLM calls.
On price changes: Apify treats increasing prices, adding a new paid event and changing the pricing model as significant changes. Users are notified and the change takes effect after 14 days. You can submit one such change per month per Actor, and you cannot cancel it once submitted. If your Actor has no paying users, the waiting period does not apply. Lowering prices takes effect immediately. The practical consequence is to price carefully the first time.

assumption-based calculation · $1.50 ÷ 0.80 = $1.875
Illustration 4. The $1.875 floor covers only platform costs and the commission; external costs, labour and taxes need additional margin.
Publish a Listing That Makes the Product Easy to Evaluate
Customers decide in seconds, so the listing has to do the selling. The practical minimum:
- A specific title that names the job (“Supplier catalogue price and stock checker”) rather than a vague label.
- A sample input a visitor can run unchanged.
- A useful sample output showing real field names, so a buyer sees the file they would receive.
- Transparent charging. State the primary event, what it costs and what limits apply. Apify requires free-plan limits to be communicated up front in the README and input schema, and a limit that stops a free user should end the run with a clear status message, never a crash or something that looks like a bug.
- Error explanations for common failure cases, and documentation a non-developer can follow.
Two Store rules matter here. Names must not suggest an official link to a supplier or a trademark owner. And you may not use the README, description, issues or reviews to promote products or services outside the platform. Ask a person who has never seen the Actor to follow the README from scratch and watch where they stumble. The listing is ready when they succeed without asking you anything.
Get Your First Paying Users
Pick channels that match the buyer. Four concrete options:
- A narrow demonstration, such as a short video or example file showing the retailer’s morning routine with and without the file. It works when the buyer is easy to describe, as in our example.
- A tutorial answering the buyer’s actual problem, for instance “How to keep your store prices in sync with a supplier catalogue,” with the Actor presented as one solution. It works when people search for the problem, not the tool.
- A relevant professional community, where you answer questions first and mention the tool when it fits. Apify’s own guidance lists Reddit, Quora, tutorial videos and articles as channels, and asks promoters to follow each community’s rules.
- Conversations with existing contacts who run online shops. This often produces the first paid feedback, because you can ask about the file directly.
Social posts with no target reader rarely work, and the Store rules above limit what you can say inside your own listing, so promote through channels that sit outside it.
Once users arrive, diagnose before building more features. Treat the following as hypotheses to test:
- Little discovery: check whether your listing matches a task people search for.
- Visits without trials: inspect the sample output, the price and the input requirements.
- Trials without repeat usage: look at output quality, usefulness and how often the job comes up.
- Usage without meaningful earnings: examine how much comes from paid plans, the price per event and the costs per run.
Apify’s documentation says the Actor Analytics dashboard (under Development, then Insights, then Analytics) reports revenue, costs and profit trends, paid and free user growth, cost per 1,000 results, run success rate and acquisition funnel data. Those are the metrics to use. Don’t assume others exist.
Reliability and support keep customers. John Cole’s habit of testing every release and answering issues quickly is one example. Apify’s terms add enforcement: you must address a reported issue within 14 days, and an Actor that stays faulty can be marked under maintenance and eventually removed.

diagnostic aid · 4 stages, 1 question each, no data
Illustration 5. Locate the weakest stage of the funnel before adding features; each stage has one question to answer first.
The Work That Continues After Launch
Target websites change layouts, rename fields or block requests, and a scraper that worked in March can return nothing in June. Guillaume’s interview says as much. Failures come mostly from his own code or from sites that change, so the maintenance burden follows the data source. Beyond breakage, expect bad data (a price shown as “call us”), higher resource consumption on large inputs, user requests for one more field, and support questions.
Apify’s Store terms make maintenance an obligation: you agree to use best efforts to keep the Actor working. Payout entitlement for an Actor also pauses while it counts as faulty, so downtime costs money directly.
Maintenance lowers your income per hour. If a tool earns $40 a month and eats three hours of repair time, it pays about $13 an hour before costs. There is no universal number of maintenance hours, so track your own for the first few months. Then decide between two paths. A reusable product suits a task many buyers share and a source that rarely changes. Custom services suit one customer with specific needs, where you charge for the hours rather than hoping usage covers them. Caleb’s mix of client work and public Actors is one version of this split, and our comparison of product and service income in making money with n8n and of reusable products in selling n8n templates covers the same trade-off in a different market.
When and How Creator Payouts Arrive
Three different things get confused here: money earned, an invoice issued and money received.
To be eligible, complete your billing details, choose a payment method and pass identity (KYC) verification. If you have not completed verification by the time invoices are issued, your earnings roll over to the next month. Under the Store terms, balances held back over incomplete verification for twelve months are forfeited, so do this early.
The documented monthly cycle runs by calendar day:
- Days 1–10: Apify prepares invoices.
- Day 11: invoices are issued, covering the previous month’s profit across your Actors.
- Days 11–14: you review the invoice. If you take no action, it is approved automatically on the 14th.
- Days 21–25: payouts are sent.
The minimum payout is $20 for PayPal and Wise and $100 for other methods. Below the threshold, funds roll over. Only funds from users who have already paid are included, and Apify may withhold a payout during an investigation of suspicious usage, or deduct amounts if a user’s compensation request related to your Actor is approved.
Transfers can cost money. For international bank wires, Apify describes intermediary fees typically between $10 and $50, charged under a shared-cost method, and suggests PayPal for lower, more transparent fees. A $60 payout by wire could therefore lose a large part of its value. Banking partners and methods can change without notice, so check the payout page before you choose. Tax treatment depends on where you live and how you operate, so consult a qualified adviser.

editorial decision aid · 4 gates, 3 outcomes
Illustration 6. Keep investing only while each condition holds: a buyer, a reliable output, a price above costs and paid repeat usage.
A Small Launch Test Before You Commit
Work through the steps in order and stop wherever the evidence fails:
- Choose a buyer and a recurring task, using the one-sentence exercise above.
- Inspect existing Actors for gaps in fields, format, reliability or support.
- Show a sample of the output to prospects.
- Validate interest with questions about current cost and frequency, and look for commitments, not compliments.
- Build a limited version that does one job well.
- Measure costs with the three-run test and calculate cost per 1,000 usable results.
- Publish at a defensible price, above your floor, with transparent charging.
- Review paid repeat usage and economics after a few weeks, using the diagnostic sequence.
Revise or stop if you meet any of these conditions:
- You cannot name an identifiable buyer.
- The prices people say they would accept sit below your costs.
- Output reliability is weak and fixing it keeps failing.
- Bespoke support is consuming the margin.
No income milestone is promised here, because nothing in the documentation or the cases supports one. Your next action takes about an hour. Write the one-sentence buyer statement for your idea, then send the sample-output message above to five people who fit it.





Leave a Reply