CAC calculator

Media spend is what the ad platforms invoiced for the period, across every channel, in one currency. This is the firmest number in the calculation and it is usually the smaller half of the answer.

Other acquisition cost is everything else you spent to win customers: agency fees, the acquisition share of salaries, creative production, affiliate and partner commissions, referral rewards, the tools the acquisition team runs on. It is optional here because plenty of people arrive with a media number and nothing else, and a media-only figure is still worth having as long as it is labelled honestly. What does not belong in this box is the cost of delivering the product once it is sold — that belongs in the margin, and counting it here would charge the same money twice.

New customers is the count of people who bought for the first time in the period. Two decisions hide in that sentence: what counts as a customer, and what counts as new. Write both down, apply them to the whole period, and do not change them quietly.

A single blended figure is where most acquisition arguments go to die, because two very different situations produce the same number.

The media-only figure moves with the auction, the creative and the targeting. It is the number a media decision changes this week.

The share that is not media is a cost structure. It moves when you hire, when you sign an agency, when you change the commission on an affiliate programme. It is nearly immovable on a weekly horizon and entirely movable on a yearly one.

Watching both at once answers the question that matters when the number rises: is this a media problem or an overhead problem. An account whose media cost per customer is flat while its loaded cost climbs has not been outbid by anybody. It has added cost above the media line, and no amount of creative testing will touch it.

It has exactly one job, and it is not to be minimised. Acquisition cost is only meaningful beside what a customer is worth: the gross profit they produce over their life with the business. A low acquisition cost in a business with thin margins and no repeat purchase can be unaffordable; a high one in a business with a high margin and years of repeat can be the best purchase available.

That is why the two most useful things to compute next are the ratio between lifetime value and this figure, and the time it takes for a customer's gross profit to pay this figure back. The first says whether the unit economics work at all. The second says whether they work quickly enough for the cash the business actually has, which is the constraint that kills companies with healthy ratios.

The figure is also a planning input rather than a target. If a customer is worth a known amount of gross profit and the business wants a known multiple of that back, the acquisition budget is not a guess: it is the number of customers wanted, multiplied by the cost the arithmetic allows.

The mix, more than the media. Acquisition starts with the cheapest audiences and works outward. A campaign that has not changed at all will acquire its next thousand customers more expensively than its last thousand, because the least expensive people to reach have already been reached. A rising blended cost with flat campaign-level costs is the ordinary shape of growth, not a fault.

The denominator's honesty. A promotion that pulls purchases forward from existing customers raises orders without raising customers. A reporting change that starts counting trial signups as customers halves the figure overnight. Both look like performance and neither is.

Seasonality on both sides. The auction is more expensive in retail peaks, and conversion rates are higher in the same weeks. The two partly cancel, and comparing a peak month to a quiet one compares neither.

Channel introduction. Adding a channel that reaches people who would have bought anyway lowers the reported cost per customer on that channel and raises the blended figure, because the spend is real and the incremental customers are not.

Work in the order of what you can move.

Start with the definitions. A figure assembled from an inconsistent customer count is not a number to act on, and this is the cheapest problem on the list to fix.

Then the conversion path. Acquisition cost is the cost of a click divided by the rate at which clicks become customers, so every improvement to the landing page, the checkout and the offer lowers it without touching the auction at all.

Then the media quality. Excluding the placements, audiences and search terms that have carried spend and produced no customers lowers the numerator without lowering reach.

Then the other half. If the share that is not media is large, the conversation belongs with the agency arrangement, the tooling and the team rather than with the campaigns, and it is usually the conversation nobody schedules.