+150 XP

Why cost per acquisition means something different for a citizen than a customer

NHS Blood and Transplant needs about 5,000 donations a day in England to keep hospitals stocked. Say a recruitment push brings in 4,000 first-time donors. What did each one cost? The answer depends on four decisions that were probably never made before the campaign launched: whether a booked appointment counts or only a bag of blood, whether the session nurse's time sits in the numerator, whether people who would have come anyway are stripped out, and whether an O negative donor and an A positive donor count as the same unit. Change any one of them and the number moves by a factor of three.

The formula is stable, the units are not

CPA = Total Campaign Spend / Number of Acquisitions

In a commercial funnel the denominator defines itself. Money moved, an order exists, finance and marketing agree on the count. Nothing arbitrates for you when the converted unit is a Self Assessment return, an e-Residency application or a donated pint. You define the unit, and whoever defines the unit sets the price. What that unit is worth afterwards, in dues, licence fees or avoided public cost, is the value-side lesson's problem. This one ends when you have a cost per unit that survives an audit.

Two rules do most of the work: settle the denominator before you spend, and declare where the numerator stops.

Defining the denominator: HMRC and the compelled action

Roughly twelve million people have to file a Self Assessment return in the UK each year, and the deadline is 31 January. Around a million miss it and collect an automatic £100 penalty. HMRC advertises hard through January. What is that advertising buying?

Divide January media spend by every return filed in January and you get a CPA of a few pence, which is nonsense. The obligation is statutory and most of those people would have filed with no advertising at all. The defensible denominator is the marginal on-time filer, and you only get at it by comparison: against last year's filing curve at the same date, against a region or segment held out of the media plan, or against the group that received reminder letters and nothing else.

This is where public sector CPA breaks most often, and it breaks in the flattering direction. A compelled action hands you a large denominator for free. The same trap sits under voluntary disclosure work, where the action you are advertising for (telling the tax authority about undeclared rental income, as HMRC's Let Property Campaign asked landlords to do) can also arrive through data matching and third-party reporting. A disclosure that would have been forced out by the data anyway is not something your campaign bought.

Publish two numbers: gross acquisitions and estimated incremental acquisitions, with the method for the second written beside it. A CPA with no stated counterfactual is a press release.

Defining the numerator: where the cost stops

A blood donation session carries costs the media plan never sees. The venue, the donor carers, the nurses, the appointment system, the reminder texts, the tea and biscuit afterwards. Marketing teams tend to count the ads. Operations counts everything. Both conventions are defensible, neither is comparable to the other, and switching between them across a reporting year produces a trend line that means nothing.

A workable boundary: include every cost that rises when you add one more acquisition, plus the fixed cost of the campaign itself, and exclude capacity that exists whether or not the campaign runs. Then state which convention you used, every time.

The reason to care is not tidiness. Leaving fulfilment out makes the channels that dump unqualified people onto a service look cheapest. NHSBT defers donors at the session for low haemoglobin, recent tattoos, travel to malaria risk areas, illness. A deferred donor consumes the full session cost and produces nothing usable, so a channel recruiting enthusiastically from a population with a high deferral rate can post the lowest cost per registration and the highest cost per unit collected. If nobody prices the deferrals, the service quietly funds the marketing team's good-looking number.

OrganisationThe flattering denominatorThe defensible oneRough size of the gap
HMRC Self Assessmentevery return filed during the campaignadditional on-time returns against a baseline or held-back groupan order of magnitude, sometimes more
Estonia e-Residencyapplications startedapproved e-residents who form a company and keep itaround five times
NHS Blood and Transplantregistrations on blood.co.ukunits collected and released to hospitalslarge: registrations lapse before a first appointment, and some attendances end in deferral

Worked example: pricing an Estonian e-resident

Estonia opened e-Residency in December 2014 and has since issued digital identities to more than 100,000 people from well over 100 countries. It is one of the few public programmes that charges at the point of acquisition: the applicant pays a state fee, roughly €100 to €150 depending on where the card is collected. That changes the arithmetic in a way most public sector marketers never get to enjoy.

Take an illustrative quarter: €400,000 of spend across digital media, events in target countries, translation and agency fees, producing 4,000 approved e-residents.

Gross CPA = €400,000 / 4,000 = €100 per e-resident

Net of the state fee, the marketing cost per approved e-resident is close to zero. Stop there and you have a triumph. Now change the unit. Roughly one in five e-residents goes on to found an Estonian company, and the company is what generates state revenue and the political argument for the programme:

Cost per company-forming e-resident = €400,000 / 800 = €500

Same campaign, same quarter, two numbers five times apart, both arithmetically correct. Choosing between them is not a technical decision. It decides which channels look efficient, which country markets get next year's budget, and whether the programme is judged on cards issued or on economic activity created. A market that produces plenty of curious applicants and few founders wins on the first number and loses badly on the second.

Not every acquisition is the same unit

A blended CPA assumes the units are interchangeable. In blood they are not. Around 8% of people in the UK are O negative, while NHSBT has put hospital demand at closer to 13% of orders, because O negative can be given in an emergency before a patient's group is known. The Ro subtype, which many sickle cell patients need for regular transfusions, is about ten times more common among Black donors than among white donors, and NHSBT runs standing appeals for that reason.

So a campaign recruiting 10,000 general donors at £25 each and a campaign recruiting 1,200 Ro donors at £90 each cannot be compared on CPA, and the expensive one may be the better buy. Report them blended and the cheap channel wins the budget every year while the shortage the service actually has gets worse. The second-order effect is nastier: optimising for the lowest cost per donor pushes you towards the groups already over-represented in the donor base, because they are the cheapest to reach and the easiest to convert.

The same logic applies anywhere the population you need is not the population that is easy to acquire: e-Residency applicants from countries where banking relationships actually work, or Self Assessment filers in the segments with the highest error rates.

Knowledge check

1. According to the lesson, what is the core distinction between how CPA is judged in commercial marketing versus public sector marketing?

2. Why is a raw CPA figure like '$15 per vaccinated person' described as 'meaningless in isolation'?

3. A city government wants to evaluate whether its $180,000 CPA for a job-training enrollment campaign was a good investment. Which approach best reflects the lesson's framework for public sector CPA analysis?

MULTIPLE CHOICE

4. Select ALL correct answers about how 'acquisition' and 'success' can differ between a commercial campaign and a public sector campaign.

Select all the correct answers.

MULTIPLE CHOICE

5. Select ALL correct answers about why comparing CPA figures across sectors (e.g., a nonprofit's donor CPA vs. a health department's vaccination CPA) without context is misleading.

Select all the correct answers.

Amortising acquisition cost across repeat units

In England men can give blood every 12 weeks and women every 16. A donor who stays active for five years at two or three donations a year delivers something like a dozen units, so a £60 acquisition cost lands near £5 per unit. A donor who gives once and never returns cost £60 per unit. Identical CPA at the point of acquisition, twelve times the difference in cost per unit delivered. NHSBT charges hospitals a bit over £100 for a unit of red cells, which gives you a sense of the room available.

That is a cost calculation and it belongs here. Which early behaviours tell you who will come back is the retention-signals lesson's territory, and what a returning donor is worth in avoided cost belongs to the value one. What you owe your finance director is the amortised figure printed next to the headline one. Approving a CPA target without a paired repeat-rate assumption approves a number that can be hit by buying one-off actions cheaply.

🎬 [VIDEO: "Nonprofit Fundraising Metrics Explained" - youtube.com/results?search_query=nonprofit+fundraising+metrics+donor+retention - search for current nonprofit sector channels (e.g. NonprofitReady, Network for Good) covering donor CPA and retention benchmarking with worked examples]

Making the number auditable

Before a CPA leaves your team, write five things next to it:

  • the unit, in one sentence precise enough that operations would count it the same way
  • the counterfactual method used and the incremental figure it produced
  • the cost boundary (media only, media plus fulfilment, fully loaded) and what was excluded
  • CPA per segment wherever the segments are not interchangeable
  • the date range, and any fee income netted off

Comparing the result against a peer set is the benchmarking lesson's job. Producing a number that is comparable at all is yours.

Key takeaways

  • The formula never moves; the unit does. Decide what an acquisition is (registration, disclosure, collected unit, active company) before you spend, because that choice can swing the reported cost by five or ten times.
  • Statutory and semi-compelled actions inflate the denominator for free. Report gross and incremental acquisitions separately, with the method attached.
  • State where the numerator stops. Media-only costing hides deferrals, drop-outs and everything the delivery service absorbs after your ad did its work.
  • Fee income at the point of acquisition, as with the e-Residency state fee, can be netted off gross CPA, but only against the same unit you are claiming.
  • Blended CPA rewards the cheapest population, not the one you are short of. Split by segment where a Ro donor and an average donor are plainly not the same purchase.