Same CPA, Different Economics — What Reports Hide

Same CPA, same offer, same period, and one campaign is quietly bleeding budget while the other prints. Your report won't tell you which is which, because a blended CPA is the last line of a story it never shows you: how the user actually reached your offer. This piece breaks down why identical acquisition costs hide completely different economics, what channel-level reporting exposes that aggregated numbers structurally can't, and the single question that separates a network tracking traffic by channel from one handing you a number and hoping you don't ask.

Quick definition up front: blended reporting is a single performance figure (a blended CPA, a blended conversion rate) that folds every traffic channel behind a campaign into one number. Convenient to read. Useless for deciding where the money actually worked.

Why Two Campaigns at the Same CPA Aren't Equal

CPA tells you what a conversion cost. It says nothing about the road the user took to get there, and that road is where the economics live.

‍

Picture two campaigns closing at the same cost per acquisition. One runs on push: a notification the user opted into, then clicked because something in it matched what they wanted. The other runs on popunder: a tab that loaded under their active window, an interruption on attention they never handed over. Both produce a conversion. Both hit the same CPA line. But the user who chose the click arrives warmer, and warmth doesn't stay a soft metric for long. It turns into retention rate, deposit rate, second-week activity, chargeback frequency. The opt-in user and the interrupted user behave differently after conversion, and your CPA column can't see any of it.

‍

This is why "our CPA is $14 across the board" is a comforting sentence that means almost nothing on its own. A campaign can win on registrations and lose on deposits. It can look cheap at acquisition and expensive across the lifetime of the account. The final number averages all of that into silence.

‍Takeaway: compare campaigns by the channel that produced them, not by the CPA they landed on.

What Blended Reporting Structurally Cannot Show You

A blended report isn't lying to you. It's just built in a way that erases the one thing that decides your economics, and it does that by design, not by accident.

‍

When a network sums push, popunder, in-page, and native into a single conversion rate, the channels don't sit side by side in the output. They collapse into one weighted average. A blended CR of, say, 1% could be one strong channel carrying a weak one, or a flat middle across all of them, and the report reads identically in both cases. You're looking at the arithmetic mean of things that should never have been averaged, because each one carries a different downstream cost.

‍

Here's the practical consequence. If a channel that converts poorly but scales cheaply is inflating your volume, the blend hides it inside an acceptable-looking headline number. You keep paying for it. You optimise toward the average instead of toward the channel that actually works, because the average is the only thing you can see. Aggregation isn't neutral; it quietly protects the weakest part of your mix from scrutiny.

Takeaway: if a network can't or won't hand you the channel split, you're optimising with the one variable that matters removed from the table.

The Channel Spread, in Numbers

In Software and Utility offers over the first half of 2026, one traffic channel out-converted the other by close to 3x. Same offers, same window. That gap isn't a quirk of one network; it's the industry pattern, and it's the reason channel mix decides economics before your creative ever does.

‍

Channel	How the user arrives	Typical strength
Push (opt-in)	Chose to subscribe, then clicked	Warmer intent, higher registration CVR
Popunder (interrupt)	Tab loaded under active window	Cheaper volume, faster scale

‍

The spread holds up outside our data. Push registration CVR runs roughly 2-4x higher than popunder, while popunder still wins on raw scale (Remoby, 2026). One is a click the user decided to make; the other rides on attention they never chose to give. Our close-to-3x sits squarely inside that market range, which is the point: this is a known channel property, not a lucky month.

‍

None of this means push always wins the account. A format can beat on registrations and lose on deposits, which is exactly why you read the channel and the payable event together, not one in isolation. The value isn't "push good, pop bad." It's knowing the spread exists in your vertical before you fund the experiment yourself.

Takeaway: in these verticals one channel reliably beats the other, and that's knowledge you can price in, not luck you hope repeats.

What This Buys an Advertiser: Not Guessing on Your Own Budget

The practical edge is measured in the weeks and the cap you don't spend learning something a network could have told you on day one.

Every advertiser who launches cold pays a discovery tax. You put an offer live, split budget across channels, and spend the first stretch, plus a slice of your cap, finding out which traffic type works in your category. That's normal when nobody has the data. It stops being normal when the network already does. If channel-level performance in your vertical is already mapped, the discovery tax is optional, and paying it anyway is just funding an answer that exists.

‍

That's what launching your offer with a network that tracks by channel actually gives you: it starts on the traffic that already performs in your category, drawn from data the network has paid to collect, instead of the version you test your way toward on your own spend. You still optimise. You still watch the payable event. You just don't start from zero on a question that's already been answered.

Takeaway: the question to bring to any network isn't "how much traffic do you have." It's "show me the channel split in my vertical before I launch."

How to Read a Network's Reporting Before You Commit

You can tell a lot about a network's reporting from one request, made before you spend anything. Here's a short filter worth running.

‍

Channel-level breakdown, not just blended. Can they show performance by traffic type (push, pop, in-page, native) rather than one merged CPA? If the only view is the blend, the channel signal isn't reaching you.

Vertical-specific data before launch. Do they know how your category behaves by channel, or will you be the one discovering it? Category history is the difference between a mapped launch and a cold one.

Conversion detail below the headline. Time, device, referrer, and the payable event, not just the count. Granular reporting is what lets you catch a channel that's inflating volume while underperforming on value.

A straight answer on what they can't show. A network that names its reporting limits is easier to trust than one that answers every question with a confident average.

‍

Run that filter and the honest question answers itself: does the network actually know the traffic split behind your CPA, or just the final number the dashboard hands over? One of those is reporting. The other is decoration.

Takeaway: one filter question, "will you show me the channel split in my vertical before I launch?", tells you whether a network's reporting is real or cosmetic.

FAQ: Same CPA, Different Economics

Why do two campaigns with the same CPA perform differently?

Because CPA is the final cost line, not a description of how the user reached the offer. Two campaigns can land on an identical CPA while one runs on opt-in clicks the user chose to make and the other rides on interrupt traffic they never selected. Downstream behaviour (retention, deposit rate, chargebacks) diverges from there, even though the acquisition cost looks the same on the dashboard.

‍

What is blended reporting and why is it a problem for advertisers?

Blended reporting is a single aggregated number that combines every traffic channel behind a campaign into one figure, like a blended CPA or blended CR. It's a problem because the aggregation erases the channel mix, and the channel mix is what decides the economics. If a network only shows you the blended result, you can't tell which traffic type drove your conversions or whether you're overpaying for clicks that barely reach the offer.

‍

Does push traffic convert better than popunder?

For registration and opt-in conversions, usually yes. Push registration CVR runs roughly 2-4x higher than popunder because push reaches opted-in users, while popunder is an interrupt on attention the user never gave (Remoby, 2026). Popunder still wins on raw scale and cheaper volume. In CIPIAI's Software and Utility offers over H1 2026, one channel out-converted the other by close to 3x, inside that same range.

‍

How can an advertiser see channel-level performance before launching?

Ask the network for the channel split in your vertical before you commit budget. A network that tracks traffic by channel before it reaches the report can tell you which channel already performs in your category. If the answer is only a blended CPA, the network either doesn't track at channel level or won't share it, and both are worth a second question before you fund the launch.

‍

Does CIPIAI report traffic by channel for advertisers?

Yes. CIPIAI tracks traffic by channel before it hits a blended report, so you can see the channel split in your vertical ahead of launch instead of reverse-engineering it from a final CPA. That's the difference between launching on traffic that already performs in your category and testing your way toward it on your own budget.

‍

You already know your CPA. The question is whether you know the traffic split sitting behind it, or just the number your dashboard hands you. If you're running tech, software, or utility offers and you're tired of blended numbers that hide more than they show, see the channel split before you launch with CIPIAI →.