CPA Goal Bidding: Set Your Target CPA, Let the Algorithm Scale
CPA Goal is an automated bidding model. You set the price you want to pay for one conversion, and PropellerAds buys the traffic slices most likely to convert at that price. It runs on Onclick, Push (including In-Page Push), Interactive, Telegram Mini App and Paid Social campaigns, and needs S2S postback tracking.
Tired of babysitting campaigns? Most media buyers spend the day tweaking bids, killing zones, second-guessing traffic, and refreshing stats like it owes them money. CPA Goal takes that part off your hands: you give the system one number, and it spends your budget hunting conversions at that price.
In this guide you’ll learn what CPA Goal bidding actually does, how it differs from Target CPA in Google Ads, which formats it runs on, and what to check when the numbers don’t add up.
In this guide you’ll learn what CPA Goal bidding actually does, how it differs from Target CPA in Google Ads, which formats it runs on, and what to check when the numbers don’t add up.
What is CPA Goal bidding
CPA Goal is an auto-optimization tool based on machine learning. It uses big data to hunt down the traffic slices most likely to convert at your specified CPA.
Under the hood: whether you’re using Push or Interstitial ads, the system uses either CPC or CPM bidding. Then it dynamically adjusts to meet your target CPA while delivering volume and efficiency.
That means you don’t have to manually analyze and blacklist or whitelist zones, instead, CPA Goal lets you spend smarter and save time.
In other words:
- you set the price you want per conversion
- the system finds the traffic & optimizes bid prices
- then it feeds you conversions (often at scale)
Inside the campaign, the algorithm keeps sorting your traffic into three groups:
- High-converting segments take the largest share of the budget.
- Average and low-converting segments get a smaller share, and only while the price still fits your target CPA.
- Non-converting segments are frozen, so the budget stops going there.
That is the practical value of the model: machine-level optimization across all formats, instead of hours of manual zone management and bid edits. It also scales with your results. As conversions come in, the system feeds the campaign more traffic, and it keeps buying only the segments that still fit your target.
CPA Goal is part of the automation stack on our self-serve ad platform, so you can run it next to rule-based optimization and multi-format campaigns without leaving the campaign builder.
Target CPA vs CPA Goal: what is the difference
If you have run search campaigns, you have already met Target CPA in Google Ads. Both models answer the same question: what should one conversion cost, and how do I make the platform respect that number?
They answer it in different places.
| Target CPA (Google Ads) | CPA Goal (PropellerAds) |
|---|---|
| Buys Google inventory: Search, Display, YouTube, Discovery. | Buys PropellerAds inventory: Onclick, Push, Interactive, Telegram Mini App and social feed placements. |
| Bids on keywords and audiences, so the signal is search intent. | Bids on traffic slices built from zone, GEO, OS, device and creative. |
| Reads conversions from Google tags or an imported source. | Reads conversions from your tracker through an S2S postback. |
| Learns on the history sitting in your Google Ads account. | Learns on the campaign itself, usually inside the first 24 hours, and leans on patterns from similar campaigns on the network while it does. |
| Adjusts bids to hit your target CPA on average across the campaign. | Adjusts CPC or CPM bids to hit the target and freezes the segments that miss it. |
What is common: you name the price of a conversion and the algorithm looks for traffic that fits it. What is different: where that traffic comes from, and how much data the model needs before it settles.
So it is not a choice between the two. If you already buy search traffic against a target CPA and want the same logic outside search, CPA Goal is where that logic lives on PropellerAds.
And what about SmartCPM? SmartCPM is the other automated model on the platform, and it solves a different task. SmartCPM looks for the cheapest placement that still wins the auction, so it works on the price of an impression. CPA Goal works on the price of a conversion: it watches how each segment performs against your target and cuts the ones that do not reach it. Pick SmartCPM when your priority is cost-effectiveness, and CPA Goal when you know what a conversion is worth to you.
Where CPA Goal works: Onclick, Push, Interactive, Telegram and Paid Social
CPA Goal runs on all PropellerAds formats. The logic is the same everywhere, but what you hand the algorithm is different in each case.
-
Onclick ads open your page in a new tab on a user action. You are billed per impression, and you don’t upload creatives at all: the landing page does the work. Here CPA Goal does the narrowing for you, staying in the zones where an impression turns into a conversion at your price and pulling out of the ones where it doesn’t.
-
Push ads deliver browser notifications after a user opts in. They then reach mobile or desktop users even when they’re not actively browsing. In-Page Push looks the same to the user but runs inside the page, which is how it reaches iOS and macOS. Both are billed per click.
The algorithm automatically evaluates which audiences and segments (GEO × OS × zone × creative) convert best and pushes traffic accordingly. High-converting segments get more traffic, while the poor ones get frozen.
-
Interactive Ads put a short mini-flow in front of the offer: a tap, a swipe, a quiz. The user is warmed up before they ever reach your page. You don’t need creatives to launch, and a landing page is optional: you can run on the offer alone or add your own page. Billing is per click.
-
Telegram Mini App Ads sit inside Telegram Mini Apps, where people are already in the middle of doing something rather than browsing the open web. Billing is per click. You can upload up to eight creatives with separate mobile and desktop images, or switch on auto-creatives and let the system build them from your landing page.
-
Paid Social Ads run in social feeds through our exclusive supply partners, so you catch people mid-scroll instead of mid-browse. You don’t build creatives here either: you hand over a landing page, and the system generates native ads from it.
| Format | Billed | What you supply | Where it fits |
|---|---|---|---|
| Onclick | Per impression | Landing page URL | Volume and fast tests, no creative production |
| Push and In-Page Push | Per click | Title, text, icon and image, four creatives or more, badge, auto-creatives | Offers that need a message, not just a page |
| Interactive Ads | Per click | Nothing beyond the offer or your landing page URL | When you want the user engaged before the landing page |
| Telegram Mini App Ads | Per click | Up to eight creatives, or auto-creatives | In-app audiences inside Telegram |
| Paid Social Ads | Per click | Landing page URL, ads are generated from it | Feed audiences on partner social placements |
Launch more than one format in one go
You don’t have to build a separate campaign for each format. Create a Push campaign, choose CPA Goal as the pricing model, and further down the form pick the additional formats you want the system to create alongside it. Onclick is on by default. Your Push creatives are adapted to the extra campaigns automatically, so you don’t redo them format by format.
You keep the controls that matter:
- Budget is either shared across the campaigns or set separately for each one.
- Targeting stays per format, so you can send each one to the audience that suits it.
- Tracking takes the macros you need. Add {adformat} so your tracker tells you which format brought the conversion.
In statistics the campaigns share a name, and you filter them by ad format. It is the fastest way to find out which format suits an offer, because they run at once instead of one after another.
How to launch a CPA Goal campaign
Here’s a streamlined “recipe” to launch CPA Goal campaigns that deliver:
Basic Recipe
1. Log in to your PropellerAds account.
2. Click “Create Campaign” → choose your ad format: Onclick, Push, Interactive Ads, Telegram Mini App Ads or Paid Social Ads.
3. Select CPA Goal as your pricing model.
4. Add your Target URL and the creatives your format needs. Push takes titles, icons and images, Telegram Mini App Ads takes up to eight creative sets or auto-creatives, and Onclick, Interactive Ads and Paid Social run on the landing page alone. Make sure your Target URL carries the ${SUBID} macro so every click gets a unique ID your tracker can send back in the postback. No SUBID, no conversion matching, and the algorithm has nothing to learn from. Then choose your GEO(s), OS, device, zones, or leave it broad for testing.
5. Set your Target CPA (desired cost per conversion), and daily/total budgets.
6. Ensure S2S (postback) tracking is set up: without tracking, the algorithm can’t optimize based on conversions
7. Launch the campaign and leave it alone. The test period takes up to 24 hours, longer when the target is tight, the budget is thin, or the funnel is long. Plan for 3 to 5 days of uninterrupted work before you judge the numbers, and don’t pause or restart it in that time.
8. Check how long your conversion actually takes to fire. On offers with a long conversion window, ecommerce, antivirus with a purchase step, paid trials, the first conversion can land 3 to 4 days after the click. The algorithm only sees what the postback delivers, so during those first days the campaign looks worse than it is. Two ways to handle it: track an earlier event in the funnel (registration, trial start, add to cart) as the optimization goal and keep the sale as a secondary metric, or keep the final event and give the campaign a test window at least as long as your conversion lag before you touch settings or turn it off
Don’t touch the campaign for the first few days. The test period is up to 24 hours, but that is when the algorithm finishes its first pass, not when the campaign is ready to be judged. Give it 3 to 5 days before you change anything, and don’t pause or restart it in that window. A restart wipes the learning you already paid for and sends the campaign back to exploration at full price. If day one looks bad, change the target CPA or the budget. Stopping the campaign is the one move that costs you the data.
How to pick the number. Start from your payout. A target CPA in the range of 70 to 90% of it is the usual starting point, and up to 100% when your priority is volume. Set it far below that and there is simply nothing for the algorithm to buy at that price.
How to budget the test. Start from your target cost per conversion, not from a round number. The working rule is 10 to 20 conversions a day: at a $10 payout that means $100 to $200 a day, at $20 it means $200 to $400. Anything below that gives the algorithm too few data points to learn from, and the test drags on without ever leaving the learning phase.
The longer your funnel, the more you need on top of that. A registration converts often enough to feed the algorithm quickly; a deposit or a KYC step does not, so plan for the full range rather than the lower end. Cheap tier-3 conversions let you run the same 10 to 20 conversions for less money, but the multiplier does not change, only the arithmetic does.
During this phase the cost per conversion can run above your target, and that is normal. It settles once the algorithm knows where your conversions come from.
Two habits that pay off on Push. Give the campaign at least four creatives and refresh them as the CTR flattens. And put the macros in your links from the start: {BannerID}, {OS}, {Country}, {ZoneID}, {Zone_Type}, {Cost}, {ClickID}. Without them, you’ll see that the campaign converts, but not what inside it does.
Pro Rule-Set Tweaks (Where Rule-Based Optimization still applies)
While CPA Goal is mostly auto, some advertisers combine it with rule-based controls (especially for CPC models), to cap spend, pause zones, or refine CPA/ROI thresholds.
Use rule sets when:
- You want to pause zones that haven’t converted after a certain spend threshold
- You want to limit daily spend per GEO/zone to avoid overspending during testing or scale-up phases
- You want to ensure ROI or CR thresholds before scaling
The hybrid approach gives you a safety net. But arguably the beauty of CPA Goal is that you shouldn’t need it for basic CPA offers.
Troubleshooting: What to Do When It Doesn’t Work
Even autopilot can hit turbulence. Almost every underperforming CPA Goal campaign fails for one of 9 reasons, and most of them are visible in the first 72 hours.
1. You paused and restarted it. The most expensive mistake, and the most common. Every restart throws away the learning the campaign already paid for and starts exploration from zero. The first 3 to 5 days are the window where this hurts most. If day one looks worse than your target, change the target or the budget; don’t stop the campaign.
2. S2S tracking isn’t actually working. In our experience, around a quarter of stalled campaigns are tracking failures, not creative or rate problems — and without conversions, the algorithm optimises blind. Fire one test conversion before judging anything else: it lands on our side within seconds, while real conversions can take up to 24 hours to show. Usual culprits: ${SUBID} missing or URL-encoded into %7B…%7D, a hardcoded visitor_id, a parameter name that changes on the prelander, or a redirect firing before the click ID is written.
3. Your target CPA is too tight. Set below what the traffic can deliver, there is simply nothing to buy and the campaign barely spends. Raise it to 70–90% of what a conversion costs you elsewhere, let it re-learn, then tighten from the data. Keep your min and max goals within 2× of each other — a wide spread slows learning down.
4. The budget is too thin for the target. CPA Goal learns from conversions, not from uptime: a $50 daily budget against a $5 target gives about ten a day, the bottom of the range, and leaves no room if the conversion price moves. Size the budget against the target: the higher your target CPA, the more budget the test needs to teach the algorithm the same amount.
5. Your conversion happens too late. CPA Goal needs a short feedback loop — a signup, an install, a deposit within days. If the event you count lands weeks after the click, run SmartCPM or SmartCPC and optimise on an earlier event instead, measuring the late one in your own tracker.
6. The campaign structure is fighting the algorithm. One campaign should cover one GEO and one OS, not “all of LATAM, all devices”. Don’t apply a whitelist on day one either — broad targeting is what collects the data the algorithm then narrows for you. Wide targeting plus a high target both stretch the learning phase.
7. It isn’t spending at all. Different symptom, different checklist: confirm the campaign is live and not stuck in draft after moderation, the balance cleared, and the targeting still leaves inventory. If all of that is fine, it’s the rate — an uncompetitive bid is the number-one cause of low volume, and a ~10% lift usually regains position.
8. It worked, then quietly stopped. Two innocent explanations first: long-running campaigns periodically re-explore as audiences shift, and Onclick needs 1–3 days to settle after any change (on Push, wait 24-36 hours before reacting). If it is real decay, blacklist zones spending without converting, refresh creatives that have flattened, and test a second prelander.
9. It’s the page, not the traffic. Clicks arriving and conversions not is a landing-page problem. Confirm the click ID survives the hop to your offer, then check the obvious: a form that’s too long, no licence number or address on the page, and a language mismatch — Brazil needs Brazilian Portuguese, not generic Portuguese.
CPA Goal in numbers
The figures below describe how the model behaves, not the result of any single campaign.
| Parameter | Value |
|---|---|
| Formats supported | Onclick, Push, Interactive Ads, Telegram Mini App Ads, Paid Social Ads |
| How you are billed | CPM on Onclick, CPC on Push, CPM on Interactive and Telegram Mini App Ads, and Paid Social |
| Test period | Up to 24 hours, longer on tight targets and long funnels |
| Recommended test budget | Enough for 10 to 20 conversions a day. On tier-1 and tier-2, where a conversion runs $5 to $10, that is about $100 to $200 a day; on tier-3 it is closer to $50. |
| Minimum conversions to train the algorithm | No fixed minimum, but more is always better. Campaigns compete for traffic by eCPM, and eCPM is built from two things: your rate and your CR. Conversions are what move the CR side, so the more of them the algorithm sees, the more accurately it can place your campaign in the rotation. |
| Tracking required | S2S postback |
Two campaign write-ups show the model at work on real budgets: a finance survey case study and an Audiences 2.0 campaign.
Key takeaways
If you’re serious about performance, you should treat CPA Goal as a must-use tool with PropellerAds. And it’s quite simple: set your target, launch properly (with S2S tracking and budget), and let the algorithm do the heavy lifting. After which, as a good advertiser, you monitor results, refresh creatives when needed, and scale up or adjust.
Stop wasting time chasing zones. Let CPA Goal find the converting ones for you.
Join our Telegram for more insights and share your ideas with fellow-affiliates!
Frequently asked questions
CPA Goal is an automated bidding model on PropellerAds. You set the price you want to pay for one conversion, and the algorithm buys the traffic slices most likely to convert at that price. It works on all formats, and needs S2S postback tracking.
Target CPA is Google’s name for the same idea inside Google Ads, where the bidding happens on search and display inventory. CPA Goal is the PropellerAds version and runs on Onclick, Push, Interactive, Telegram Mini App and Paid Social traffic. The section Target CPA bidding vs CPA Goal on this page compares the two side by side.
CPA stands for cost per action: the amount you pay when a user completes the action your offer counts as a conversion. We break the term down in full in what CPA means.
The test period takes up to 24 hours. That is when the algorithm finishes its first pass, not when the campaign is ready to be judged. Give it 3 to 5 days of uninterrupted work before you change anything, and longer if your target CPA is tight, the budget is thin or the conversion fires late.
The algorithm won’t find enough traffic at that price, so the campaign crawls or barely spends. Raise the target to 70 to 90% of your payout, let the test period finish, and bring it down later once you can see which segments convert.
Trends
View more posts
