Can you get into media buying with $100 in your pocket? Most players don’t make it past the first month, and it’s almost never the offer’s fault.
The ad account charges you today, and the network pays weeks later. In a cheap country, the real entry point starts near $500, and casino on Facebook with breathing room runs up to $5,000.
The post below splits that bill into 3 parts: what you buy at the start, what each test costs, and how much money has to sit in reserve waiting for the payout.
Table of Contents
- 1 – How much it costs to start media buying
- 2 – How much media buying infrastructure costs
- 3 – How much it costs to test a CPA offer
- 4 – How many tests until a profitable CPA campaign
- 5 – How much working capital to hold in media buying
- 6 – Real small-test cases in CPA
- 7 – How to start media buying on a small budget
- 8 – Mistakes that double the cost of starting in CPA
- 9 – Media buying cost FAQ
- 10 – How much it costs to start in CPA: the bottom line
How much it costs to start media buying
The CPA Lenta team put $300 into a weight-loss offer and got the same $300 back. The campaign, which is the combination of offer, ad, and page, stopped working shortly after.
The next tests burned through what was left in the till, and the operation ended right there, in the first cycle.
Casino on Facebook costs up to $5,000 with the full test sequence and the working capital, and 2 teams landed on that figure in different years. You can enter with less, but finding a campaign that works takes longer.
The 4 figures come from the CPA Lenta report (2024).
| Method | Suggested start |
|---|---|
| Casino on Facebook | up to $5,000 |
| Health and beauty on Facebook | from $3,000 |
| Push and native networks | up to $1,000 |
| Content traffic | under $500 |
- In casino, the test sequence is expensive and the money sits locked in the payout window
- In health and beauty, finding the right campaign costs up to $1,500, and keeping it running takes another $1,000 (CPA Lenta, 2024)
- In push and native networks, the paid action is cheap and you don’t need a stack of accounts
- In content traffic, videos and articles take the place of paid ads
With $3,000 in casino the search is already moving, and the $5,000 buys the breathing room. In a cheap country, a beginner enters with $500 to $2,000 (CPA.live, 2026).
Content traffic was the CPA Lenta team’s own route. That’s how they saved up the money for the paid route.
Whoever enters casino without the $5,000 is betting on luck with the first test.
Why $300 won’t sustain an operation
The team repeated the start with $250 to $300 and ended up in the same place. Both times they found a campaign that responded, even with little money.
What was missing was the stamina for the string of misses that comes before the hit. The offer was never the problem.
In other words: with little money you can find a good campaign, but you can’t survive the string of bad tests it takes to get there.
How much media buying infrastructure costs
Cpagram went shopping before its first casino ad and spent $245. On TikTok, the same shopping list came out near $260.
In that total, everything with a free tier stayed on the free tier, from tracking to ad spying. The figure covers only the start, and the monthly bills arrive later, once the operation grows.
| Item | Cost | Can it wait? | Source |
|---|---|---|---|
| Ad account | $1 each | no | (CPA Mafia, 2024) |
| Virtual card | $2.50 per issue | no | (CPA Mafia, 2024) |
| Proxy (Proxy-Cheap) | $1 to $13 per GB | no | (archive, 2026) |
| Antidetect browser (Dolphin) | free up to 10 profiles | yes | (Cpagram, 2022) |
| Backup domain | $5 each | depends | (CPA Mafia, 2024) |
| Tracker (Keitaro) | from $25 a month | yes | (Cpagram, 2022) |
| Spy tool (BigSpy) | $9 a month | yes | (Cpagram, 2022) |
| Spy tool (AdHeart) | $53 a month | yes | (Cpagram, 2022) |
| Card top-up fee | 2% to 8% | no | (CPA Mafia, 2024) |
A test calls for close to 10 accounts, and nothing runs without them. The card is where the ads charge, and the account goes down without a proxy.
At the start, everything that can wait has a free version: Dolphin‘s paid plan starts at $9, the network’s dashboard covers the first month of tracking, and the free ad library stands in for the spy tool.
The proxy is the service that makes your connection look like it comes from the offer’s country, and Proxy-Cheap sells residential from $2.60 per GB. The tracker is the dashboard that logs every click through the SubID, the tag that identifies the click.
The sale travels back the other way on its own, because the postback is the automatic notice the network sends. Without a tracker and postback, the spend shows up on your statement and you have no idea where the sale came from.
Anyone operating from outside the account’s country also pays a fee on every card top-up, and it hits the bill before the first click.
The alternative: renting ready-made accounts for a commission
- The pre-warmed account takes the place of your own profiles
- Card, page, and ad are still on you
- The commission is a percentage of ad spend, so it goes into the test’s math
An agency account is a service that rents out ready, pre-warmed ad accounts. In exchange, it charges 15% to 20% of everything the campaign spends (CPA Mafia, 2024).
A warmed account has a history of normal use, and the platform takes longer to shut it down. What it doesn’t solve is the rest: card, page, and ad you still pay for and build yourself.
How much it costs to test a CPA offer
In India, Cpagram tested a casino offer paying $50 per deposit, and the network asked for 30 approved deposits to close the test.
In the Germany scenario, the same math with Justbit takes $4,500 just to break even, that is, for the deposits to pay back the ad money.
| Test scenario | India offer | Germany offer |
|---|---|---|
| Payout per deposit | $50 | $225 |
| Deposits required in the test | 30 | 20 |
| Double the spend | $750 | $2,250 |
| Break even | $1,500 | $4,500 |
| Lose half | $3,000 | $9,000 |
The teams’ rule is to spend at most 10 times the lead’s value on a test, the lead being the action the network pays for. With a $50 lead, the test’s ceiling sits at $500.
The country is the first thing that moves the budget. Swapping Germany for India cuts the same test’s cost by 3.
What to test first to spend less
The advice that repeats across the reports is to test the offer with one page and one ad, and leave the polish for later. The lowest real figure in a cheap country, already in profit, was $925 before subscriptions (Cpagram, 2022).
- 3 or 4 days of continuous spend, with the budget stretched to the end of the window
- 2 lead values per ad group, with several groups running at once
- The first cut at 3 or 4 lead values, before hitting the limit
- The network’s own page and ad, swapped only after the offer responds
A misleading ad takes the account down before the test ends, and what an ad is allowed to say changes from country to country.
How many tests until a profitable CPA campaign
In the CPA Lenta reports, every test that works arrives after 4 or 5 that don’t. If you plan for a single test, you’re betting everything on the first result.
That’s why teams budget a sequence of 5 to 6 tests. The tests alone start at $2,500, not counting the tools and the money sitting idle waiting on the network’s payout.
After each test, the decision follows the return on spend (ROI), which is profit divided by what went out. The rule is to stop putting money into the range where the campaign almost never turns around.
The whole rule comes from the CPA Lenta report (2024).
| Test result | What to do |
|---|---|
| Positive return | increase the spend |
| Loss up to 20% | optimize |
| Loss of 20% to 50% | push on while it improves |
| Loss beyond 50% | almost never turns around |
| Loss beyond 70% | shut it off |
With a positive return, you increase the spend without touching the page or the ad. Up to a 20% loss, the turnaround is likely, and optimizing is worth it. Between 20% and 50%, push on only while things keep improving, up to the test’s limit.
When the loss passes half and optimization changes nothing, the rule says shut it off and look for another offer. Pushing on there burns more money than the campaign gives back.
The risk the dashboard doesn’t show
One advertiser, for example, rejected more than half the leads of an entire network, and the loss rolled down to the affiliates. The contract allows it: a lead outside the agreed quality target (KPI) comes back unpaid.
The reserve beyond the testing budget exists for that rejection. Ask about the quality target in your first conversation with the network’s manager.
- The quality target comes up in conversation before the first click
- The reserve absorbs the rejection without stopping the good campaign
- Payout window and quality target, together, set the size of the till
How much working capital to hold in media buying
Most players go broke at the start and never try again. The fall comes with the offer still working, because the money runs out while the network hasn’t paid.
Working capital is the money that pays for ads while the network’s payout hasn’t arrived. The hold is the length of that wait, and in casino it runs 2 weeks.
The locked-money math is simple: daily spend times the days of hold. The card charges for ads every day, and the revenue only lands when the window closes.
| Daily spend | Network’s wait | Money locked up |
|---|---|---|
| $50 | 14 days | $700 |
| $100 | 20 days | $2,000 |
| $500 | 20 days | $10,000 |
At $50 a day and a 14-day window, $700 sits locked up before the first payout. That’s 2 weeks paying for ads without collecting a thing.
On the classic route, working capital takes half the total or more. Stopping a profitable campaign for lack of cash is expensive too, because the platform loses what it learned about the audience.
How to split the money once the profit arrives
The CPA Lenta report splits income across 4 destinations, and reinvestment takes at least half. The order exists to protect the working capital before any withdrawal.
- Reinvestment in the campaign that’s already profitable, with at least half the income
- New tests, to find the next campaign before the current one stops working
- The operation’s services and subscriptions
- Personal withdrawal, last and with a fixed cap
The report’s classic mistake is withdrawing past the cap on the first good streak. The bad streak comes later, and the hole in the till stops the campaign.
Real small-test cases in CPA
In Australia, an affiliate put $59 into casino ads and got $92 back. In Poland, a men’s health offer turned $105 into $390.
Vietnam landed on the other side: less money came back than went out. The table shows profit and loss together, because that’s what keeps the math honest.
| Spend | Return | Offer | Country | Source |
|---|---|---|---|---|
| $59 | $92 | Rocket Play casino | Australia | CPA Lenta |
| $105 | $390 | men’s health | Poland | CPA Mafia |
| $184 | $357 | Leadbit offer | Greece | CPA.rip |
| $308 | $694 | N1bet betting | Nigeria | Traffnews |
| $124 | $1,126 | health and beauty | not disclosed | Cpagram |
| $203 | $1,720 | not disclosed | not disclosed | CPA.rip |
| $106 | $77 | phone carrier | Vietnam | AddSet |
| $50 | $28 | miracle product | not disclosed | HotCPA |
The offer promising a miracle, for example, gave back little more than half the spend. A single test can turn a profit, but it’s the sequence of tests that decides whether the operation lives.
Reading the full report is worth more than looking at the profit screenshot. Whoever only looks at the screenshot sees the pretty part of the sequence.
The approval rate is the share of orders the advertiser accepts, and it cuts the return before the money arrives. On a supplement campaign in Spain, that share sat at 31.5% (Cpagram, 2021).
How to start media buying on a small budget
Teams driving traffic to model profiles pay close to $0.70 per free subscriber, and no video carries a direct link, because the platform takes down profiles that point outward (Conversion, 2026).
That’s content traffic, which the reports call nearly free. The cost sits in the antidetect browser that keeps the profiles apart, the proxy, and the content production tools.
- Videos of 6 to 15 seconds with the promo code on screen, and no link in the description
- 2 to 3 videos a day per channel, with the first ones posted with no offer at all
- News-style articles on content platforms, with the offer at the end
- GoLogin‘s free plan holding up to 10 profiles for the first channels
- TikTok’s creative center as a free ad reference
This route’s job is teaching you to operate and saving up the money for the paid route. That’s how one team built its own working capital before buying ads.
Mistakes that double the cost of starting in CPA
The script repeats across the reports, always in the same order. The beginner spends everything on tests, finds the good campaign, and watches it die with no cash, because the hit came too late.
The most expensive mistakes come before the first click: spending everything on tests, paying for tools too early, and polishing the ad before testing the offer. They all come out of the same bill.
Budgeting the test and forgetting the working capital
The network pays weeks late, and the card charges for ads every day. Without a reserve, the good campaign dies before the payout arrives.
- Working capital holds half the total money or more
- A profitable campaign on pause loses the audience the platform learned
- The reserve for rejected leads goes into the math before the first test
Paying for tools before having an offer
The full casino infrastructure runs up to $500 before the first ad. Proxy and antidetect together pass $50 a month, and neither buys a single click without an offer.
- The cloaker, the filter that hides the page from moderation, runs up to $150
- Each ad account costs up to $100 in the most trusted versions
- The offer picked with the manager costs nothing and decides more than every tool combined
Testing the ad when the question is the offer
If you start by swapping ads, you burn the whole test limit without knowing whether the offer responds. The first test measures the offer with one page and one ad.
- The 10-lead-value limit applies per offer, not counting ad swaps
- The cheap cut at 3 or 4 lead values weeds out the offer that doesn’t respond
- Polishing the page and the ad comes after the first sign of life
Media buying cost FAQ
How much does it cost to start media buying?
Between $500 and $2,000 in a cheap country (CPA.live, 2026). Casino on Facebook runs up to $5,000 for the full package, from the accounts to the money sitting idle waiting on the network.
Can you start with $100?
You can learn and even close one test in profit, like the $59 case in Australia. But the money runs out before the operation can survive the normal string of failed tests.
What counts as a beginner’s infrastructure?
The ad accounts, the proxy, the antidetect browser, the virtual card, and the domains. The set comes out to $245 when everything with a free tier stays on the free tier.
Why plan for more than one test?
Because for every test that works, another 4 or 5 fail first. With money for a single test, the first bad result ends the operation.
How much does testing an offer cost?
It depends on the lead’s value: the teams’ rule is to spend up to 10 times that value on the test. With a $50 lead the ceiling sits at $500, and on a cheap casino offer the test closes near $70.
What is working capital in media buying?
It’s the money that pays for ads while the network hasn’t paid you yet, that is, during the hold. The teams’ recommendation is to keep at least half the total budget in it.
Why does a profitable affiliate go broke?
Because the profit is on the dashboard and the money is locked in the network’s window. The card charges for ads every day, and the casino hold keeps the money for 2 weeks.
What’s the cheapest way to start?
Content traffic, with short videos and articles, starts under $500. It earns less, and its job is teaching you to operate and building the cash for the paid route.
Do I need to pay for a tracker and a spy tool at the start?
No, the tracker and the spy tool can wait. The network’s dashboard covers the first month, and Keitaro starts at $25 once the operation grows.
When do you shut off a campaign in testing?
When the loss passes half the spend and optimization changes nothing. The teams’ rule shuts off anything losing more than 70%, because the recovery almost never pays off.
How much it costs to start in CPA: the bottom line
The starting math closes when infrastructure, test sequence, and working capital close together. On the classic route, infrastructure takes $245 and the test sequence eats close to $2,500.
- Separate infrastructure, tests, and working capital before spending the first dollar
- Build the infrastructure on the free tier of everything that has one
- Budget the whole test sequence, because the hit comes after several misses
- Keep at least half the total budget in working capital
- Shut off, without mercy, any campaign losing more than half with no reaction
The classic total starts at $5,000, and content traffic starts under $500. Whoever survives the first month is usually whoever saved money for the second.
The other guides in this series show how to choose the offer and how to scale the campaign once the test responds.
How we sourced this post
The numbers come from public offers and case reports in the niche, checked in August 2026. Each offer’s terms come from the networks’ own materials.
Commission, hold, and approval rates change without notice, and the catalog turns over every week. Check the offer in the dashboard before putting money on media.
Our fact-checking process is on the About CPA RAW page, and the other guides in this series follow the same method.
Spotted an outdated number? Let us know through the contact page.
This post is informational and not financial advice. Media buying carries risk: you pay for the traffic before you know whether it converts.
Leave a Comment