Most campaigns on pop, push, native and display traffic lose money in the first days. That is normal. The first phase is not about profit; it is about buying data cheaply enough to find the sources, countries and creatives that work, and then cutting everything else. This guide walks through a practical optimization process that affiliate marketers and small advertisers can follow on any self-serve ad platform.
Before you start: set up conversion tracking
You cannot optimize what you cannot see. Before spending real money, make sure every conversion is reported back to the ad platform and tied to the click or impression that produced it.
Pixel or postback?
- Pixel (image or JavaScript): placed on the thank-you page. Easy to set up when you control the landing page, but it relies on the visitor’s browser and can be blocked or lost.
- Server-to-server postback: the offer owner or your tracker calls a URL on the ad platform with the click ID when a conversion happens. It does not depend on the browser and is the standard for affiliate offers.
How a postback works in practice: the ad platform adds a click ID macro to your destination URL, your tracker stores it, the affiliate network passes it back when someone converts, and your tracker fires the postback with that ID. Test the full chain with a test conversion before launch. A broken postback is one of the most expensive mistakes in performance marketing, because every decision you make afterwards is based on missing data.
1. Source sampling with frequency capping
When a campaign starts, a few large sources can take most of your budget before you learn anything about the rest. Sampling fixes that. Set a cap on impressions per source (domain or zone) per period. For example, a cap of 1,000 impressions per domain per day means no single domain can take more than that each day, so your budget spreads across many domains.
After a few days, open your reports and break them down by domain or zone. You now have a comparable sample from many sources and can see which ones deserve more budget and which ones should be blacklisted.
Separately, set a user frequency cap (how many times the same person sees your ad). For pop and push, showing the same ad to the same person many times a day mostly wastes money.
2. Read the placement report correctly
The placement, zone or source report is where most optimization happens. For each source, look at:
- Spend and impressions or clicks
- CTR (for push, native and display)
- Conversions and cost per conversion
- Landing page CTR to offer, if you use a tracker
Sort by spend, not by cost per conversion. A source with one lucky conversion on a tiny spend looks great but proves nothing. Focus first on the sources that took the most money.
| What you see | Likely meaning | Action |
|---|---|---|
| High spend, no conversions | Source does not fit the offer | Blacklist |
| Conversions at or below target cost | Profitable source | Whitelist, test a higher bid |
| Conversions above target, but close | Possibly fixable | Lower the bid for this source |
| Very high CTR, no conversions | Possible accidental clicks or poor traffic | Check carefully, usually blacklist |
| Low spend | Not enough data | Keep running |
3. Automate with optimization rules
Checking reports by hand every hour is not practical. Rules do it for you. A rule says: if a source meets a condition within a time period, take an action.
Example: your CPM is $1 and you do not want to pay more than $10 for a conversion. That budget buys 10,000 impressions. A matching rule is: “For each domain, if impressions are greater than 10,000 and conversions are less than 1 today, blacklist the domain.” The rule then checks your sources regularly and removes those that spent your target conversion cost without converting.
You can build similar rules on clicks instead of impressions (useful for CPC push campaigns), or on cost per conversion once you have enough volume. Start with loose thresholds and tighten them as you learn. Rules that are too strict will cut sources before they have had a fair test.
4. Blacklists and whitelists
A blacklist removes sources you have proven do not work. A whitelist restricts a campaign to sources you have proven do work. A common structure:
- Test campaign: broad targeting, source sampling, rules that blacklist poor performers.
- Whitelist campaign: only the proven sources, with higher bids to win more of their traffic.
- Move new winners from the test campaign into the whitelist every few days.
Keep your blacklist per offer or vertical. A source that fails for a dating offer may work for a utility app.
5. Bid adjustments
Bidding too low usually means you only win leftover inventory, which often converts poorly. Bidding too high on everything wastes money. The practical approach is to start near the market price for your country and format, then adjust per source.
- Raise bids on sources that convert below your target cost so you win more of their traffic.
- Lower bids on sources that are close to target instead of cutting them.
- Change bids in small steps and wait for new data before the next change.
To choose a starting bid, it helps to know what others are paying. Our pop and push CPM benchmarks by country show average prices by country and device from recent advertiser data.
6. Split campaigns by country, device and OS
A campaign that targets several countries and all devices produces blended numbers that hide what is really happening. Prices and conversion rates differ a lot by country and device. Split at least by country and by mobile versus desktop. For app offers, split by operating system too.
7. Test creatives and landing pages
For push, native and display, the creative decides your CTR and your CTR affects how much traffic you win. Run three to five creatives per campaign, pause the weakest once they have enough impressions, and add new ones regularly, because push and native creatives lose effectiveness over time.
For every format, the landing page decides your conversion rate. Keep it fast, mobile-friendly and in the visitor’s language. Test a pre-lander against a direct link to the offer.
8. Dayparting and budget control
Look at conversions by hour and day of week. If certain hours never convert, exclude them or lower bids during them. Set daily and total budget limits so a test cannot overspend while you are away.
Common mistakes
- Launching without tested tracking. Every later decision depends on it.
- Optimizing on too little data. Decide in advance how much spend a source gets before you judge it.
- Changing many things at once. If you change bids, creatives and targeting on the same day, you will not know what worked.
- Chasing cheap traffic. A low CPM is only good if the cost per conversion is good.
- Never refreshing the whitelist. Sources change over time. Keep testing new ones.
- Ignoring device and country splits. Blended data leads to wrong decisions.
Optimization checklist
- Postback or pixel tested with a real test conversion.
- Campaigns split by country and device.
- Source frequency cap set for sampling; user frequency cap set.
- Target cost per conversion defined.
- Blacklist rule based on that target in place.
- Three to five creatives running.
- Daily budget limit set.
- Review the placement report daily, sorted by spend.
On PPCmate you can buy on CPM or CPC and use conversion tracking with pixel and postback, frequency capping, dayparting and daily and total budget limits, as described on the pricing page.
FAQ
How long should I test before optimizing?
Base it on spend, not time. A useful rule of thumb is to let each source spend at least your target cost per conversion before judging it, and let the whole campaign collect enough conversions to compare sources.
Should I use CPM or CPC?
CPM is common for pop traffic, where you pay for each view of the page. CPC is common for push, where you pay only when someone clicks. Choose the model that matches how the format is usually sold and track cost per conversion either way.
What is the difference between a blacklist and a whitelist?
A blacklist excludes specific sources from a campaign. A whitelist allows only specific sources. Test with blacklists, scale with whitelists.
Do automatic rules replace manual review?
No. Rules handle the routine work of cutting obvious losers. You still need to review reports to spot patterns such as a country, device or hour that performs differently.
