Traffic arbitrage is buying attention on one platform and selling the actions it produces to someone who pays more for them. The gap is your margin. Everything else in this guide is about how big that gap really is in 2026, what you have to pay before you can even test it, and which rules will close your account if you ignore them.
It is written for someone starting from zero and outside the closed chats where most of this knowledge circulates. Every price, minimum and policy quote below is taken from the vendor’s own published page, and where a figure is billed in euros rather than dollars that is stated.
What traffic arbitrage actually is
There are two prices in every campaign: what you pay for a click or a thousand impressions, and what an advertiser pays you for a completed action such as an install, a lead, a deposit, a sale, or a click on a search feed. You sit between them.
You are not building a brand or a product. You are running an arbitrage on the price of attention against people who have bought the same placements for years and know exactly what they convert at. It is a numbers job with a creative department attached.
Three models dominate right now, and beginners confuse them constantly:
- CPA arbitrage. You send traffic to an offer inside an affiliate network and get paid per action. Payout is fixed and known in advance.
- Search arbitrage. You buy a cheap click on a native or social platform, land the user on a page carrying a search feed, and get paid when they click an ad inside that feed. Payout is variable and depends on the keyword.
- Media arbitrage. You buy traffic to your own content property and monetise it with display or affiliate placements. Slowest, most durable, least popular with beginners.
The formula that decides whether you keep going
Profit is revenue minus spend. Revenue is conversions multiplied by payout. Spend is clicks multiplied by cost per click. Rearranged, the only number that matters on day one is your break-even conversion rate: cost per click divided by payout.
Say your click costs $0.05 and the offer pays $2. Break-even is 0.05 divided by 2, which is 2.5 percent. Every hundred clicks needs at least two and a half conversions before you make a cent. If your landing page converts at 1.8 percent, no amount of budget will fix it, and you should stop buying and rewrite the page instead.
Do this arithmetic before you launch. It tells you immediately whether an offer is worth testing on a given source at all.
Search arbitrage: the model that took over
Most English-language guides describe CPA offers and stop there. The volume moved. A large share of current beginner-to-mid activity is native-to-search: you buy a cheap native click and monetise it through a search feed such as Google’s Related Search for Content, known as RSOC.
RSOC is not something you switch on in a dashboard. Google’s own documentation is explicit: to use related search on content pages you must complete contracting for AdSense for Search, share a mock-up with your account manager for review, and comply with the AFS Product-Integrated Feature policies. In practice most operators reach it through a feed partner rather than directly.
The rules on paid traffic are tighter than people expect. Google’s AFS policy states that where the traffic source is under your control, “Traffic sources must not promise products, services, or promotional offers that are unavailable or are not easily found on the destination page”, and each destination page must be substantially the same as the page an organic visitor would see. Since 1 November 2025 you are also required to pass the referrerAdCreative parameter containing the precise and complete creative text of the ad that sent the user. Your ad copy is now handed to Google alongside the traffic. Mismatched creatives are no longer invisible.
Where the payout comes from: CPA networks and their terms
On the CPA route the network holds your money between the conversion and your bank account, and its payment terms shape your cash flow more than your ROI does.
MaxBounty, one of the larger English-language networks, publishes its terms openly: the minimum balance before a payment is issued is $100, affiliates are paid monthly on net 15 to start, and only after the first successful payment are electronic payees moved to a weekly schedule. Read that again as a cash flow problem. Money you spend in the first week of January may not reach you until 15 February.
Practical consequences for a beginner:
- Budget for two payment cycles of float, not one. Your ad spend is due immediately, your revenue is not.
- Ask the affiliate manager which traffic sources are allowed before you buy. Networks reject conversions from disallowed sources after the fact.
- Check whether the payout is on a lead or on an approved lead. The difference between the two can be half your revenue.
Traffic sources that accept arbitrage, and what an account costs to open
This is where translated guides mislead an international reader. Assume nothing about which networks are open to you. Current published entry costs, from each network’s own pages:
| Source | Entry cost | Notes |
|---|---|---|
| PropellerAds | Minimum deposit from $100 | Stated in the advertiser terms as starting from $100, depending on the payment method |
| Adsterra | $100 by card, Paxum or PayPal | $200 via Inxy digital wallets, $1,000 by wire transfer |
| RichAds | Minimum deposit $150 | Minimum CPC from $0.003 on push and in-page; popunder CPM from $0.5 on Tier 3 geos and $1 on Tier 1 |
| Outbrain | Minimum $20 per day | Monthly budgets must be at least $600, which is $20 per day across the month |
| Taboola (Realize) | Recommended daily budget of 10x your CPA goal, or at least $50 | The stated aim is roughly 10 conversions per day so the algorithm has data to work with |
| TikTok Ads | Campaign budget must exceed $50, ad group budget must exceed $20 per day | A 31 day lifetime ad group therefore needs at least $620 |
Two things fall out of that table. A serious first test on native starts in the hundreds of dollars per source, not the tens. And push and popunder networks are genuinely cheap to enter, which is why every beginner guide points there and why competition on the good placements is brutal.
What the big platforms will not tolerate
The fastest way to lose an arbitrage business is a policy ban, not a bad campaign. Two rules cause most of the damage.
Bridge pages and thin destinations. Google’s destination requirements policy prohibits “Destination content that is replicated from another source without adding value in the form of original content or additional functionality” and “Destinations that are solely designed to send users elsewhere”. A one-screen page whose only job is to bounce the visitor to an offer is exactly that description.
Cloaking. Meta’s Spam policy defines it plainly: “Cloaking is any attempt to circumvent our content policies by intentionally presenting different off-platform content, such as URLs or applications, to our integrity systems versus what is shown to users.” The same policy separately bans misleading links, deceptive redirects where the domain changes after a captcha or a forced action, and landing pages that generate accidental traffic through pop-ups or clickjacking.
Beginners often treat these as risks to be managed with more accounts. That is the expensive path. It costs money, it costs time, and it caps how big you can get. If you are going to run multiple accounts for legitimate reasons, read our guide on how to protect ad accounts from bans first, because the infrastructure mistakes are more predictable than the policy ones.
A tracker is the first thing you buy
Without attribution you cannot tell which creative, placement, geo or hour produced a conversion, which means you cannot cut anything with confidence. Trackers are a real line item. Current published prices:
- Keitaro: Starter at €40 per month, billed in euros, 1 user and 1 domain with SSL. Self-hosted.
- RedTrack: Builder at $69 per month with 2 million tracking events; Solo at $141 with 5 million.
- Voluum: Profit at $119 per month billed annually with 1 million events, and $0.06 per additional 1,000 events.
- Binom: v2 at $149 per month, or $104 per month billed yearly, self-hosted, with unlimited domains and users and a 30 day trial.
Two pricing philosophies. Voluum and RedTrack charge by event volume, so tracking cost scales with traffic. Keitaro and Binom charge a flat licence on your own server, so cost per click falls as you scale but the server problems are yours. Pick based on where you expect to be in six months.
The learning phase is a line in your budget
On algorithmic platforms your money buys two things: traffic, and the data the algorithm needs to target it. The second one is not optional and it is not free.
Meta states the threshold directly. An ad set becomes “Learning limited” when it is unlikely to receive around 50 optimisation events in the week after your last significant edit, and it moves back to active once it has received enough of them since that edit. If your offer pays out rarely, you either raise budget or optimise toward a more frequent event.
Taboola frames the same problem as a daily budget: 10 times your CPA goal, or at least $50, aiming at roughly 10 conversions per day. Every edit you make resets this. The most common beginner mistake is not underfunding a campaign, it is editing a funded campaign every few hours and never letting it leave learning.
Creatives and landers are where the margin is won
Click prices across a given source and geo converge fast. What does not converge is how well your ad and your page match each other and match the offer. That is the only place a beginner has a real edge, because it is the one part of the operation that does not require capital.
Work in batches. Change one variable at a time: the hook, the first frame, the headline, the call to action. Kill losers on a rule set in advance, not on a feeling. The buyers who scale simply run more tests per week and read the data honestly.
Volume is the constraint. Spy tools show what is already running: AdPlexity publishes Native at $249 per month, Desktop, Mobile and Adult at $199, and Push and Social at $149. Useful for direction, but it shows you what is already saturated. Twenty original variants a week is the harder and better habit.
Running out of creatives before you run out of budget?
We build AI content pipelines that turn one tested angle into dozens of ad variants, landing page versions and video cuts, so the testing loop is limited by your media budget rather than by production time. Service pages are in Russian; we answer in English.
Geos, tiers and keeping the first test narrow
Traffic is priced by tier. Tier 1 means markets like the United States, the United Kingdom, Germany and Australia, where clicks are expensive and payouts are high. Tier 3 means markets where both are low. RichAds publishes the spread plainly: popunder CPM starts at $0.5 for Tier 3 geos and $1 for Tier 1, a two times difference on the same format.
Starting on Tier 3 because clicks are cheap is a reasonable way to learn a tracker and a network without burning money. It is a poor way to learn what converts, because audiences, devices and payment habits differ. Decide which you are doing before you spend.
Whichever tier you pick, run one geo, one format, one offer and one angle at a time. A test with four moving parts produces a result you cannot attribute to anything.
A realistic first 30 days, with numbers
Assume the cheapest credible setup: Keitaro Starter at €40 for the tracker, a domain and a small server, and one push network at its published minimum deposit of $100 to $150. That is somewhere between $150 and $200 before a single click is bought, and the deposit is the only part that turns into traffic.
At RichAds’ $0.003 minimum CPC, $150 buys tens of thousands of clicks on paper, but that is the floor bid on the least competitive placements. Assume the first month buys a working tracker, three or four offers eliminated, and one angle with a pulse. Treat the first deposit as tuition.
If you want to test native or TikTok instead, the arithmetic is different and larger. TikTok will not let an ad group run below $20 a day, and Outbrain will not let a monthly budget sit below $600. A month of honest testing on either one is a four-figure commitment, not a hobby budget.
Mistakes that drain beginner budgets
- Scaling on two good days. Meta needs roughly 50 optimisation events per week before its delivery is stable. Two days of profit is not a signal, it is noise.
- Editing during the learning phase. Every significant edit restarts the clock and re-spends the learning budget.
- Ignoring the payment cycle. Net 15 monthly terms with a $100 threshold mean you can be profitable on paper and out of cash at the same time.
- Cloaking to get around a rejection. Meta’s Spam policy names it explicitly, and enforcement is at the asset level, not the ad level.
- Running without a tracker to save the licence fee. A month of Keitaro Starter costs €40. You will spend more than that within two days of not knowing which placement converted.
- Copying a spy tool ad verbatim. If AdPlexity can see it, so can everyone else paying $149 to $249 a month, and the angle is already tired.
FAQ
How much money do I need to start traffic arbitrage?
The published floor is around $150 to $200: a tracker such as Keitaro Starter at €40 per month plus a push network minimum deposit of $100 at PropellerAds or Adsterra, or $150 at RichAds, with a domain and a small server on top if you self-host. Native and social start much higher, since Outbrain requires at least $20 per day and TikTok requires a campaign budget above $50 with ad groups above $20 per day.
Is traffic arbitrage legal?
Buying and reselling traffic is legal in itself. What gets penalised is how it is done. Meta’s Spam policy prohibits cloaking, misleading links and deceptive redirects, and Google’s destination requirements prohibit pages that are “solely designed to send users elsewhere”. Stay inside those and you are running an advertising business, not a scheme.
What is the difference between CPA arbitrage and search arbitrage?
In CPA arbitrage the payout per action is fixed by the offer. In search arbitrage you are paid on clicks inside a search feed such as Google’s Related Search for Content, so revenue varies by keyword. RSOC also requires AdSense for Search contracting and a mock-up approved by an account manager, plus the referrerAdCreative parameter on paid traffic since 1 November 2025.
Which tracker should a beginner buy?
At low volume, an event-priced cloud tracker is cheapest: RedTrack Builder at $69 per month covers 2 million events, Voluum Profit at $119 billed annually covers 1 million. At higher volume a flat licence wins: Keitaro Starter at €40 or Binom v2 at $104 per month billed yearly, both self-hosted with no per-event charge.
How long before a campaign shows whether it works?
Long enough to clear the platform’s learning threshold without editing it. On Meta that is around 50 optimisation events in a week after your last significant change. On Taboola the stated target is roughly 10 conversions per day at a daily budget of 10 times your CPA goal or at least $50. Below those volumes you are reading noise.
