Written by Krisztián Stark
Key takeaways
- A CPA network sits between advertisers and publishers: it lists offers, tracks clicks and conversions, checks quality, and pays publishers for each approved action.
- Tracking runs on a click ID that travels with the user and comes back in a server-to-server postback when the action happens.
- Approval and reversals are part of the model, so a conversion is only final once the validation period is over.
What is a CPA network?
A CPA network, or cost per action network, is a company that connects advertisers who want a specific action, such as a sign-up, an install or a sale, with publishers who can send people to do it. The advertiser pays the network an agreed price for each completed action, the network tracks and checks the results, and it pays the publisher for every action that is approved. Nobody pays for impressions or clicks that lead nowhere.
CPA networks are also called affiliate networks or performance networks. The labels overlap: what defines them is that money moves only when a measurable result happens. This guide is about the network itself. For the metric and how to calculate it, read What is CPA?, and for how CPA compares with paying per view or per click, see CPM vs CPC vs CPA.
Who does what in a CPA network
| Role | What they do | What they want |
|---|---|---|
| Advertiser | Creates an offer: the action, the price, the countries, and the traffic rules. Confirms conversions from its own system. | Real customers at a known cost per result |
| Network | Lists offers, gives publishers tracking links, records clicks and conversions, checks quality, invoices advertisers and pays publishers | Volume from both sides, and enough quality that advertisers stay |
| Publisher or affiliate | Promotes offers to an audience: a website, an app, a newsletter, a community, or paid traffic it buys itself | Offers that convert well for its audience, and reliable payment |
| User | Sees the offer and completes the action, sometimes for a reward on an offerwall | Something worth their time |
The network earns the difference between what the advertiser pays and what it passes on to publishers, and in exchange it carries the work of tracking, quality control, billing and payouts that neither side wants to build alone.
How a CPA network works, step by step
- The advertiser sets up an offer. For example: “Free trial sign-up, United States and Canada, $4 per confirmed account, no email traffic.”
- A publisher picks the offer and receives a tracking link. Many offers need the network to approve the publisher first.
- A user clicks. The network records the click, creates a unique click ID, and redirects the user to the advertiser with that ID in the link.
- The user completes the action. The advertiser’s system stores the click ID with the new account or order.
- The advertiser reports the conversion. Its server calls the network’s postback URL with the click ID, and often the event name and amount.
- The network credits the publisher, usually as pending first, and confirms it after any validation period.
- Money moves on a schedule. The advertiser pays the network, and the network pays publishers in payout cycles, such as weekly or monthly.
Payout models you will meet
CPA is the umbrella term. The action it pays for gives each model its name:
- CPL, cost per lead: a form, a sign-up or a quote request.
- CPI, cost per install: an app install, usually counted on first open.
- CPS, cost per sale: a purchase, paid as a fixed amount or a percentage of the order.
- CPE, cost per engagement: an in-app goal such as finishing the tutorial or reaching a level.
- Multi-step or multi-event: several payouts for one user, for example install, registration and first purchase.
- Revenue share: a percentage of what the referred customer spends over time, instead of a fixed price per action. CPA vs revenue share compares the two.
How tracking works: click IDs and postbacks
Everything depends on one identifier. When a user clicks, the network generates a click ID (also called a transaction ID) and adds it to the advertiser’s link, for example ?click_id=abc123. The advertiser keeps that ID, and when the action happens, sends it back. That is how the network knows which publisher, and which click, earned the conversion. What are click IDs and sub IDs? covers the parameters in detail.
There are two common ways to send it back:
- Postback (server-to-server): the advertiser’s server calls a network URL directly. It does not depend on the user’s browser, cookies or ad blockers, which is why most CPA tracking now uses it. Server-to-server tracking explains the method and What is a postback URL? the format.
- Pixel (browser-based): a small script or image on the thank-you page fires when the page loads. It is easier to install but loses conversions when browsers block third-party tracking.
Publishers can add their own sub IDs to a tracking link, such as the page, the ad or the user it came from, so they can see which source converts. Networks then send their own postback on to the publisher, so a rewards site can credit its user automatically. Each of these calls should be signed or checked, as Postback security describes.
Approval, validation and reversals
A CPA network has three layers of approval:
- Publisher approval: the network checks a new publisher’s site, audience and traffic sources before giving access.
- Offer approval: some advertisers approve each publisher individually, or only allow certain traffic types, such as no incentivized or no email traffic.
- Conversion validation: conversions may be held as pending while the advertiser checks them. Leads with fake details, refunded sales or installs that never open can be rejected, and an approved conversion can be reversed later, which takes the payout back.
For publishers, this means earnings are not final the moment a conversion appears. For advertisers, it means the terms of the offer, including what counts as a valid action and how long they have to dispute it, should be written down before traffic starts.
Fraud in CPA networks
Paying per action removes some fraud, such as fake impressions, but creates incentives for others. Common patterns are:
- Bots and device farms that fill forms or install apps.
- Fake leads with made-up or stolen contact details.
- Click spamming and click injection, which claim credit for conversions that would have happened anyway.
- Sending traffic the offer does not allow, such as rewarded users on an offer that forbids incentives.
- Multiple accounts and VPNs used to complete the same offer many times.
Networks fight this with publisher vetting, IP and device checks, conversion time analysis, and by sharing rejections back to the source. Ad fraud and bot traffic lists the warning signs advertisers can check themselves.
Pros and cons of CPA networks
| Pros | Cons | |
|---|---|---|
| For advertisers | Pay only for results; reach many publishers through one contract; the cost per result is known in advance | Less control over where the ad appears; fraud needs active checking; the price must be attractive enough for publishers to choose the offer |
| For publishers | Access to many offers without negotiating each one; tracking and payments handled for them | Carry the risk that visitors do not convert; earnings can be held or reversed; offers can pause without warning |
How pay-per-result platforms like Sharklio differ
Sharklio works on the same principle, payment per result, but with two kinds of campaigns. Multi-step offer campaigns work much like a classic CPA offer: the advertiser’s tracking link receives a {click_id}, and the advertiser’s server reports each step with a postback within 30 days of the click, as the offer postback docs describe. They are available by arrangement with our team.
Task campaigns work differently. There is no postback to set up. Users complete the task and send proof, such as a screenshot, a text answer or both, and the advertiser reviews each completion and approves or rejects it within an approval window. Users can dispute a rejection they believe is wrong. The traffic is openly incentivized: it comes from offerwalls on publishers’ websites, apps and communities, where users earn a reward. Pay only for approved results explains the review model.
Pay per action without building tracking first
If you want CPA pricing but have no postback setup yet, a Sharklio task campaign lets you pay for actions you check yourself, with bids per country from $0.01. Every campaign is reviewed by our team before it goes live. You can create a free account and your first campaign today.
Frequently asked questions
What does CPA stand for in a CPA network?
Cost per action. The advertiser pays for a completed action, such as a sign-up, an install or a sale, rather than for views or clicks.
Is a CPA network the same as an affiliate network?
Largely, yes. Both connect advertisers with publishers and pay for results. Affiliate network is the older term and is often used for sales and percentage commissions, while CPA network is common for leads, installs and fixed payouts.
How do CPA networks track conversions?
With a click ID created when the user clicks the tracking link. The advertiser stores it and sends it back to the network in a postback, or through a pixel, when the action happens.
Why was my CPA conversion reversed?
Usually because the advertiser found the action invalid after checking it, for example a fake lead, a refunded order or traffic the offer did not allow. Reversals are normally possible during a validation period set in the offer terms.
Do CPA networks allow incentivized traffic?
It depends on the offer. Each offer lists the traffic types it accepts, and many exclude incentivized traffic. Offerwall offers are built for it, and say so in their terms.
