What is Blended CAC?

A business spends A$180,000 across every channel in a quarter and gains 600 new customers, giving a blended cost of A$300. The platforms, added together, claim 810 of those customers — 35% more than the business actually acquired — because several are crediting the same people. Each platform's own figure looks better than A$300, and the A$300 is the only one that reconciles with the customer table.

Blended customer acquisition cost divides everything spent on winning customers in a period by the number of new customers the business actually gained in it. No channel gets credit for anything. The advertising, the agency, the content, the affiliate commissions and the tooling go into one numerator; the customer table supplies the denominator.

The case for a metric this blunt is that its competitor has a structural flaw. Per-channel acquisition cost is built from attribution, and attribution is a set of claims rather than a set of facts. Each platform reports the customers it believes it caused, using its own window, its own view-through rules and its own modelling of the conversions it could not observe. Those claims overlap heavily, because customers see more than one ad. Added together, the platforms frequently account for more customers than the business has.

The blended figure cannot be wrong in that way. It may be wrong about causation — it credits advertising for customers who would have arrived anyway — but its denominator is a count, and counts do not double.

Blended CAC = total acquisition spend ÷ new customers

Both terms hide a decision.

The numerator has to include everything spent with the intent of winning customers. Media is obvious. Agency and contractor fees, creative production, acquisition tooling, affiliate and partner commissions, and the loaded salaries of the people doing the work are all defensible inclusions, and the most common mistake is excluding the ones that are hard to attribute — brand advertising, content, sponsorships — precisely because they are hard to attribute. Leaving them out produces a better-looking number that the business cannot bank.

The denominator has to be new customers, from the system that records customers, over a period aligned with the spend. Not orders, which include repeat purchases. Not conversions, which include whatever the ad account was configured to count. Not leads. Where the sales cycle is long, the alignment of periods matters more than the precision of either number.

Both choices should be written down once and then left alone, because a definitional change looks exactly like a trend.

It is the control. When every channel reports an improving acquisition cost and the blended figure is flat, something in the attribution has changed rather than something in the business. That check is unavailable from any other single number.

It survives measurement loss. Consent rates, tracking restrictions, in-app browsers and modelled conversions have all made per-channel figures noisier over recent years. None of it touches total spend or the customer table.

It captures everything. Channels with no tracking at all — a podcast read, a print insert, an event — contribute customers that per-channel reporting simply loses. The blended figure counts them without being told.

It matches how budgets are set. Most businesses hold a total acquisition budget and a new-customer target. This is the ratio between them.

It will not allocate. There is no channel dimension in it, so the only decision it supports directly is whether to spend more or less in total.

It will not distinguish incremental customers from customers who would have arrived anyway. A business with strong word of mouth and a well-known brand will report an attractive blended figure while its paid activity does very little, and nothing in the number will say so.

It moves with things nobody did. A seasonal peak, a press mention, a competitor leaving the market and a strong email programme all add customers to the denominator without any acquisition spend behind them.

And it responds slowly. Aggregating a period means waiting for the period, which makes it a poor instrument for weekly decisions and a good one for quarterly ones.

The most useful thing to do with this metric is not to read it in isolation but to compare it with the per-channel figures added together.

If the channels collectively claim 35% more customers than the business gained, that over-claim is a measurable quantity and its trend is informative. A widening gap means the platforms are crediting themselves with more of the same customers, usually because attribution windows lengthened, modelling increased, or retargeting grew as a share of spend. That is a measurement change, and treating it as a performance improvement leads directly to over-funding whichever channel is best at claiming credit.

Retargeting is the standard example. It advertises to people already close to buying, so it claims customers efficiently and cheaply. Per-channel reporting rewards it; the blended figure is indifferent to it; a holdout test settles it.

It rises. This is not a defect.

The cheapest customers in any market are the ones already looking, and they are bought first. Each additional increment of budget reaches people who are further from a decision and more expensive to convince. A blended acquisition cost that climbs as spend grows is the normal shape of a scaling business, and the question is not whether it rose but whether it is still under the ceiling the contribution allows.

A figure that stays perfectly flat through a doubling of spend is worth investigating rather than celebrating. It usually means the denominator is absorbing customers the business would have won without the extra money.

Blended acquisition cost is one of three views, and each covers the others' weaknesses.

The per-channel figures allocate the budget and are collectively optimistic. The blended figure controls them and cannot allocate. Incrementality testing — a geo split, a holdout, a clean on-off — measures causation directly and is the only one of the three that does, at the cost of time and some deliberately forgone revenue.

Beneath all three sits the ceiling: what a customer contributes over the relationship, which is what decides whether any acquisition cost, blended or otherwise, is one the business can pay.

• Count new customers from the customer table, not from the ad platforms. The whole value of the metric is that its denominator is a fact rather than a claim.
• Track the gap between the blended figure and the sum of the per-channel ones. The gap is the size of the attribution over-claim, and its trend is more informative than either number alone.
• Hold the numerator's definition fixed. Whether agency fees, production and tooling are inside acquisition spend is a choice, and switching it mid-year turns a definitional change into an apparent trend.
• Watch it against the share of revenue from new customers. A blended figure improving while the new-customer share falls means the existing base is carrying the result.
• Use it as the control on channel decisions rather than as a substitute for them. It says whether the allocation worked; it cannot say what the allocation should be.