Calculating lifetime value when the customer doesn't pay you directly

A National Trust member pays a subscription once a year, then walks free into houses, gardens and car parks for the next twelve months and may never send a separate donation. Some of those members also give sixty hours of their own labour to a footpath party. A small fraction leave a legacy that dwarfs everything they paid while alive, and legacy income across the Trust runs into the tens of millions of pounds a year. Ask three people what one member is worth and you get three defensible answers an order of magnitude apart. Whichever one you write down becomes the ceiling on your acquisition budget.

This lesson builds the value side of that equation: a lifetime valuelifetime valueLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → figure made out of dues, licence fees, donated hours and avoided public cost that will survive a finance director reading it.

Why the commercial formula breaks

The commercial version, average purchase value × frequency × lifespan, assumes a clean transaction stream. Public and non-profit relationships produce four different kinds of value, and only one of them behaves like a purchase:

• Membership dues arrive on a renewal cycle but are bundled with a service you have to fund (a magazine, free entry, wardening, car parking).
• Licence fees are set by policy, not willingness to pay. NHK's receiving fee applies to households with a receiver, whatever they think of the programming.
• Volunteer hours are labour, not cash, and never touch the income statement.
• Avoided public cost is money that never appears in your accounts at all, and usually lands in somebody else's budget.

The structure of the formula survives. What changes is what you put in the first term, and how honest you are about what you subtract.

Rebuilding it with the right proxy

LTV = average annual net value × retention-adjusted lifespan

Where lifespan comes from the observed renewal rate:

Average lifespan = 1 / (1 − retention rate)

Worked example: a membership body

Illustrative figures, not the Trust's actual accounts:

• Annual subscription: £90
• Renewal rate: 85% (large UK heritage memberships sit high, in the region of 80% and above)

Lifespan = 1 / 0.15 ≈ 6.7 years. Gross LTV = £90 × 6.7 ≈ £600.

Now subtract the cost to serve. If magazine, handbook, postage and the marginal cost of that member's visits come to £30 a year, net annual value is £60 and LTV falls to about £400. A third of the headline number was never yours. Membership models that fund a physical estate almost always have a cost to serve; ignoring it is the single most common way these models get inflated.

Legacies are worth modelling as an expected value rather than a hope. If roughly one member in 500 leaves a bequest and the average charitable residual legacy is in the tens of thousands of pounds, say £30,000, that is £60 of expected value per member. But it arrives twenty years out. Discounted at 4% it is worth about £27 today. Report the discounted figure, or a trustee will eventually ask why the cash never showed up.

What recurring commitment does to the arithmetic

US public radio stations have spent more than a decade converting listeners into monthly sustainers, and sustainer retention is generally reported far above one-time donor retention (estimate, it varies by station). Take 80%: lifespan = 1 / 0.20 = 5 years. At $15 a month, LTV = $180 × 5 = $900, against roughly $218 for a $120-a-year one-time giver at 45% retention.

Nothing about the gift size did that. Retention did. How you actually move someone from a single gift onto a standing order belongs to the donor journeys lesson; what matters here is that the arithmetic rewards it by a factor of four.

Volunteer hours that survive a finance review

Independent Sector, a US non-profit coalition, publishes an annual estimated dollar value of volunteer time (independentsector.org/value-of-volunteer-time); check their site for the current year's figure.

Volunteer LTV = hourly reference value × annual hours × years of service. Fifty hours a year for four years at an illustrative $33/hour is about $6,600 from someone who never gave a penny.

The failure mode: valuing hours you would never have bought. If the footpath would have stayed unrepaired without the volunteer party, the replacement-wage figure is notional and a finance director will strike it out. Count hours that displace paid work, generate measurable output, or release staff time onto something you do pay for. Everything else is a communications number, not a planning number.

Avoided public cost

The test is a counterfactual, not an aspiration: if this had not happened, who would have written the cheque, and how big was it? Published unit costs make this defensible rather than rhetorical (in the UK, the Treasury Green Book and the local-government unit cost databases give per-incident figures for things like an emergency admission or a tribunal hearing).

Two rules keep it credible. Only claim avoided cost that a named body would otherwise have incurred. And never add it into the same total as income without labelling it, because it is not money you can spend on salaries.

Where the model breaks

Retention close to 1. NHK has over 40 million receiving-fee contracts, and payment is a legal obligation, upheld by Japan's Supreme Court in 2017, rather than a renewal decision. Put a 97% retention rate into 1 / (1 − r) and the model returns a 33-year lifespan and a lifetime value bigger than the household. Above roughly 90%, cap the horizon (ten years is a common convention) and discount it, because the real limit is how long the household exists at that address, not how loyal it feels.

Policy resets your whole book. NHK cut its receiving fee in 2023. Every modelled lifetime value in its files dropped that day, with no marketing failure attached. If your unit price is set by a board, a ministry or a regulator, run the model at two price points and know what a 10% cut does to your acquisition ceiling before someone else discovers it.

Double counting through a federated structure. A listener's donation goes to their local NPR member station; NPR itself is funded substantially by station dues, programming fees and sponsorship. That single dollar is real once. If the station books the full lifetime value and the network books its share of the same dollar, and then the joint fundraising case adds both, the sector has invented money. Decide which entity owns the LTV figure and let the other one hold a fraction.

Blended cohorts. One retention rate across one-time donors, sustainers and members produces a number that describes nobody. Model each separately or the average will hide a channel that loses money.

Setting LTV against channel cost

The figure only earns its keep next to what acquisition costs by channel, defined the way the CPACPACost Per Acquisition: the total cost to generate one customer or conversion, computed by dividing total spend by the number of acquisitions.View full definition → lesson defines it for a non-purchase conversion:

The commercial 3:1 target is directional at best here, and the benchmarking lesson explains why importing a SaaS ratio is a category error. What is not negotiable: a channel sitting below 1:1 is draining programme money.

A second-order consequence worth naming. A high LTV model licenses higher acquisition spend, and higher spend chases the audiences that model best, which in a membership charity means older, wealthier, more retentive households. Left unchecked for three years, an LTV-optimised acquisition plan will quietly narrow who your organisation serves.

For European benchmarks, the UK's Fundraising Regulator and Chartered Institute of Fundraising reports are the place to check retention assumptions rather than importing US figures; the Fundraising Effectiveness Project publishes the US ones (data.givingtuesday.org/fundraising-effectiveness-project).

A retention cohort snippet

# Simple year-over-year donor retention calculation
donors_year1 = 1000
donors_returned_year2 = 450

retention_rate = donors_returned_year2 / donors_year1
avg_lifespan = 1 / (1 - retention_rate)

print(f"Retention rate: {retention_rate:.0%}")
print(f"Estimated avg donor lifespan: {avg_lifespan:.1f} years")

Run it by acquisition cohort, not across the whole file, and cap the lifespan output as above. This measures retention that has already happened; forecasting it from early behaviour is the engagement signals lesson's territory.

🎬 [VIDEO: "Donor Retention: Why It Matters More Than Acquisition" - youtube.com/results?search_query=donor+retention+nonprofit+fundraising - search this term for current, sector-specific breakdowns of retention economics from fundraising conferences and nonprofit consultancies]

Key takeaways

• Swap "purchase" for the right proxy: net annual dues for members, reference hourly value for volunteers, attributable counterfactual cost for public savings.
• Subtract cost to serve. A £90 subscription with £30 of member benefits is a £60 asset, and the gap widens every year you ignore it.
• Retention drives the result, but 1 / (1 − r) becomes unstable above about 90%. Cap the horizon and discount anything more than a few years out, legacies especially.
• In federated or licence-funded structures, agree which entity owns the number before anyone builds a budget on it.
• Treat every benchmark as an estimate with a country and a year attached, and re-run the model whenever a policy sets your price for you.