The Annual Affiliate Program Review: A One-Evening Audit That Sets Up Next Year
Once a year, look at the whole program instead of the month. The six questions to answer with data, the partners to thank or retire, and the one change worth making per year.
Monthly numbers tell you what moved; they're terrible at telling you what the program is. Once a year — the quiet week after the holiday rush is ideal — sit down with twelve months of data and answer six questions. The whole exercise fits in an evening, and it routinely finds the one change that defines the next year.
The six questions
1. What did the channel really cost? Total commissions paid plus reversals absorbed plus your hours, against affiliate-attributed revenue. This effective percentage is the number to compare with your ads CAC — the comparison that justifies (or questions) the whole program. The KPI definitions apply; the annual view just smooths the noise out of them.
2. Where did revenue concentrate? Rank partners by attributed revenue and look at the top three's share. Above two-thirds and you have a dependency, not a portfolio — the concentration problem — and next year's recruiting priority writes itself.
3. Who activated, who evaporated? Of the partners approved this year, how many produced a single referral? A weak activation share points at onboarding, not at the partners. And scan for the sadder list: previously active partners who went quiet. A personal note to five of them in January revives a surprising fraction.
4. What did fraud actually cost? Count the year's rejected referrals and reversed commissions, and — more useful — reread why. The fraud patterns that recur point at a rule worth tightening; a near-zero rejection year in a growing program can mean review is too loose, worth a spot check.
5. Did the structure earn its complexity? Every tier, bonus, and custom rate gets one honest look: did it change behavior, or just arithmetic? Structures accrete; the annual review is when they're allowed to die.
6. What did partners say all year? Reread the year's partner emails. Repeated questions are documentation gaps, repeated complaints are policy gaps, and the requests you deflected twice ("do you have a comparison table I can use?") are next year's creative kit.
The people moves
Data first, then three lists. Thank the top ten personally — a real email naming their number and what it meant, plus whatever your budget allows: a negotiated VIP rate, an annual bonus, early access to next year's launches. Retention of proven partners is the cheapest growth you'll ever buy. Retire the dead weight: applications approved years ago, zero referrals, still receiving your newsletter. A gentle "still interested?" email with a 30-day sunset keeps the roster honest and your numbers meaningful. Write down the bench: the five prospects you'd most like to recruit next year, while the year's context is fresh.
One change, announced well
The review's output should be a single structural change — a rate adjustment, a window improvement, a payout-rhythm upgrade, one new seasonal push — not five. Programs absorb one change a year gracefully; partners experience three simultaneous changes as instability, even when each is an improvement.
Announce it in January's partner newsletter alongside the year's headline numbers. Partners talk about programs that share real numbers — total paid out, top-partner earnings, average time to payout — because so few programs do. If your software keeps the year queryable — every referral, reversal, and payout in one ledger with its history — the whole evening is export-and-read. If assembling the picture takes longer than analyzing it, that's a finding too.