LTV by channel: why your blended CAC is lying to you

LTV by channel means knowing what users from each acquisition channel are worth, not just what they cost. Your blended CAC cannot tell you that. It averages a cheap channel, a mid channel and an expensive channel into one comfortable number, and that number hides the only decision that matters: which channel to cut and which to feed.

I run paid user acquisition for a GCC fintech app, and the most useful habit I keep is refusing to read the blended number first. Blended CAC is the figure everyone reports and almost no one should act on. Here is why it misleads, and how to replace it with a by-channel view you can actually run budget against.

Why is blended CAC a vanity metric?

Because an average is where differences go to disappear. Any blended CAC is consistent with one channel at half that cost, one near it, and one well above it. The blend can look healthy while a meaningful slice of your budget buys users at a price you would never approve if you saw it on its own line.

This is not hypothetical. On the GCC fintech app I run, breaking cost per registration out by channel showed the most expensive channel paying roughly 3x what the cheapest one paid, in the same market, in the same period. The blended figure sat comfortably in the middle and reported that everything was fine.

Here is the shape of the problem, with illustrative numbers indexed to the blend:

The blend reads 100 and everyone relaxes. Channel C is quietly buying registrations at 170. Nobody approved 170. The average approved it on their behalf.

Objective mix hides in the blend too. On the same account, a purchase-optimised objective delivered registrations roughly a third cheaper than an install-optimised objective on the same audience. Fold both into one average and you lose that finding entirely. I wrote that split up in app install vs event-optimised campaigns.

What does LTV by channel add that CAC cannot?

Cost tells you what you paid. LTV by channel tells you whether it was worth paying. The two questions have different answers surprisingly often: the channel with the cheapest registrations is frequently not the channel with the most valuable users, and you cannot see that until value is broken out by source.

You do not need a finance team to start. You need your MMP (Adjust, AppsFlyer or similar) reporting one downstream value event by channel, on cohorts, in one place. The one-place part matters. Comparing Meta's opinion of Meta against Google's opinion of Google is comparing two self-reports. Pick one source of truth for cross-channel decisions and let each ad platform remain its own opinion of its own work.

Why does adding spend raise your blended CAC?

Because new spend buys at the margin, and arithmetic is undefeated. If extra budget lands anywhere above your current blended average, the blend rises. By definition. This is why "we will scale our way back to target" so rarely survives contact with the invoice: the target was an average, and the new spend is not average.

The lever that actually moves a blend is reallocation. Cut the channel above the average, feed the channel below it, and the blend falls without any new budget at all. On the account I run, this reframing changed more decisions than any tool did. When the monthly number needed defending, the answer was mix, not money.

What is marginal cost, and why is it 3-4x your average?

A channel at its efficient frontier sells you the next conversion for far more than its average suggests. On a Google App campaign (still widely called UAC) on the account I run, the bid simulator priced marginal registrations at 3-4x the campaign's own average cost. The average said this channel is a bargain, scale it. The margin said the bargain is fully bought.

The tell is a campaign under-spending its budget while showing "Eligible" rather than "Limited by budget". That campaign is bid-capped, not budget-capped or creative-capped. Raising the budget does nothing, and raising the target cost buys those expensive marginal conversions.

Averages describe the past. Margins price the future, and only one of them should set your next budget move.

How do downstream events expose which channels deserve budget?

A registration is not value. It is a promise of value, and downstream events tell you which channels keep the promise. On the fintech app I run, adds of payment details became the quality proxy: an ad with a registration cost roughly two thirds higher than the account's cheapest ads, but steady adds-of-payment volume behind it, was the better buy than a cheap ad producing nothing after the registration.

That finding repeats at channel level. Rank channels on registration cost alone and you will fund the ones that are good at producing registrations. Rank them on cost per downstream event and you fund the ones producing customers. Cheap registrations are easy. Users who go on to fund an account are the point.

How do I build an LTV by channel view this week?

You can stand up a working version in an afternoon with tools you already have:

1. Pick one source of truth. Your MMP, not the ad platforms. Platforms grade their own homework.
2. Break spend and cost per registration out by channel, with iOS and Android on separate lines. If your iOS numbers look untrustworthy, they probably are: see SKAdNetwork explained.
3. Add one downstream event column per channel. Add of payment details, first deposit, first order, subscription. Whatever your app's first honest signal of value is.
4. Compute each channel's downstream-event rate. What share of registrations go on to the money event (an add-of-payment, a funded account)? This is your early LTV proxy, readable within weeks instead of quarters.
5. Tag every channel above or below the blended average, then cross-reference with the downstream column. Above-blend cost with no downstream signal is your cut. Below-blend cost with strong downstream signal is your feed.
6. Before scaling the winner, check its marginal cost. If the bid simulator prices the next conversion at multiples of the average, expand creative and audience instead of budget.

FAQ

What is the difference between blended CAC and CAC by channel?

Blended CAC divides total acquisition spend by total new users across every channel at once. CAC by channel does the same division one channel at a time. The blend is a reporting convenience; the by-channel view is the decision tool, because budgets move between channels, never against an average.

Is blended CAC ever useful?

Yes, as a trendline. It is a fair board-level summary of whether acquisition efficiency is drifting up or down across quarters. It fails the moment it is used to decide anything, because every decision it could inform, cut, scale or reallocate, happens at the channel level the blend erased.

How do I measure LTV by channel without years of revenue data?

Use an early proxy event that correlates with value: add of payment details, first deposit, first order. Measured per channel in your MMP on cohorts, it separates high-quality channels from merely cheap ones within weeks. Refine toward true revenue LTV later; in my experience the ranking rarely flips.

Why does my CAC rise every time I increase budget?

Because you buy conversions at the margin, not at the average. Each budget increase pushes the channel into more expensive inventory, so marginal cost climbs above the average your dashboard quoted. At the efficient frontier that gap can reach 3-4x. Check marginal pricing before scaling, not after.

Do I need an MMP to do this?

For cross-channel truth, yes. Adjust, AppsFlyer and their peers exist precisely so every channel is measured by the same referee. Without one you are comparing each platform's self-report, and platforms systematically flatter themselves. On iOS the self-reports can sit multiples apart from the MMP.

What if one channel looks expensive but drives my best users?

Then it may be your best channel. Judge it on cost per downstream event, not cost per registration, and keep it funded as long as the downstream maths works. An expensive registration that becomes a funded account beats a cheap one that never opens the app again.

Not sure which of your channels is quietly overpaying? I run a free App Growth Review: a short, no-pitch look at your live campaigns, your attribution setup, and where the budget is leaking.

Get a free App Growth Review →

Related reading: app install vs event-optimised campaigns, scaling Google App campaigns in the GCC and why GCC apps waste paid media budget.