Blended CAC calculator

Total marketing spend is everything spent to acquire customers over the period: media across every platform, agency fees, the acquisition share of salaries, creative production, affiliate and referral payouts, tooling. The value of this figure is that it is complete; an incomplete numerator produces a number that is wrong in whichever direction somebody preferred.

New customers is the count of people who bought for the first time, from the business's own records rather than from any platform.

Customers a platform claimed is optional and is the count some platform attributed to itself, added up across platforms. Because platforms can each claim the same order, this total can exceed the real one — and the third output says so rather than hiding it.

Blended is what the business paid per customer. One numerator, one denominator, both from its own books, every customer counted once whoever claims them. It cannot double count and it cannot be improved by a reporting change.

Claimed is what it paid per customer that some platform took credit for. It is more conservative — the same spend over fewer customers — and it is the figure a campaign decision is made on, because a campaign can only be judged on outcomes something connects to it.

At 60,000 of spend, 400 new customers and 240 claimed, the two are 150 and 250. Neither is wrong. The business paid 150 a customer; the tracked half of the business paid 250 each.

The truth about the other 160 customers is almost always mixed. Some were influenced by advertising that failed to track them — an untagged link, a blocked cookie, a person who saw an ad on a phone and bought on a laptop a week later. Some genuinely arrived by word of mouth, search for the brand name, or a friend. The gap is not an error to eliminate; it is a quantity to understand.

The unattributed share is the number to watch over time, because it moves for reasons that change what the other two mean.

A rising share with flat spend usually means tracking degraded: a consent banner changed, a tag broke, a platform shortened a window. Nothing about the business improved, and the claimed cost will be rising in step while the blended one holds.

A rising share with rising spend can mean advertising is producing demand that arrives through channels nobody credits — brand searches, direct visits, referrals from customers who were themselves acquired. This is the case where attributed reporting most understates advertising, and it is why a business running only on platform figures tends to under-invest.

A falling share usually means either better tracking or a growing dependence on paid acquisition. The two are opposite situations with the same signature, and only the total customer count tells them apart.

A negative share — more claimed customers than real ones — is double counting, which is the ordinary behaviour of attributed figures rather than a fault. It is also the clearest possible demonstration of why the blended number exists.

It cannot tell you which campaign to change. It is an average over every channel, every creative and every audience, so it says whether the whole machine is working and nothing about which part.

It also moves for reasons outside marketing. A quarter with strong repeat purchase, a press mention, a seasonal peak or a product that found its audience lowers the blended cost without any campaign improving. Crediting the campaigns for that is a mistake that gets made confidently, usually in a good quarter.

And it says nothing about incrementality. A business could stop advertising entirely and keep some of those customers; the blended figure has no way to know how many.

The last limitation is timing. Marketing spent this month produces customers over the following weeks, and a blended figure computed month by month charges each month's customers against a different month's spend. In a business with a considered purchase that mismatch is large enough to invent trends: a month that cut spend looks efficient because it is still harvesting the previous month's work, and the month that restores spend looks expensive for the same reason in reverse. Reading the figure as a rolling quarter rather than a month removes most of it.

Hold the blended figure as the constraint. Set the maximum the business can pay per customer from lifetime gross profit and the payback the cash position allows, and check the blended number against it every period. That is the question of whether acquisition is affordable.

Run the campaigns on their own attributed numbers, knowing those numbers are conservative and knowing by roughly how much — the gap on this page is that measure.

When the two disagree persistently and materially, the answer is a test rather than another report: move spend somewhere, hold everything else still, and watch the total. The blended figure is the only one of the three that will answer honestly.