The 7-Day Launch Playbook

Before day one: the prerequisite

Confirm customers already recommend you unprompted. A referral program amplifies existing satisfaction; it cannot manufacture it. If nobody is recommending you today, a reward will mostly buy signups from people who wanted the reward, and the week is better spent elsewhere.

Day 1 — Economics

Decide what you can afford before deciding what to offer. Work from gross margin on a newly acquired customer, not revenue, and settle whether the reward is one-off or repeats. Every later decision is constrained by this number, so making it first prevents rework.

• Gross margin on a first order or first contract year
• The maximum total reward across both sides, as a share of that margin
• Whether the reward repeats on renewal — a very different commitment
• A ceiling on total program spend for the first quarter

Day 2 — The qualifying event

Choose the action that earns the reward. This decides more about the program's outcome than the reward amount does. Later events cost volume and buy quality; earlier ones do the opposite. Write it precisely enough that a dispute can be settled by reading it.

1. 01List the candidate eventsSignup, first order placed, first order past the refund window, second purchase. Each is defensible for some businesses.
2. 02Pick the latest you can tolerateEvery step later reduces fraud and improves quality at the cost of volume and advocate patience.
3. 03Write it as a testable sentence"A first paid order of £25 or more that is not refunded within 30 days" is testable. "Becomes a customer" is not.

Day 3 — Reward structure

Decide the currency and the split. Two-sided is the default for consumer programs because it turns an ask into a gift. Match the currency to each side: account credit for the existing customer, something that lowers the barrier to a first purchase for the new one.

Resist tiering on day three. Tiers and milestones are for programs that already have repeat referrers, and adding them at launch is complexity you cannot yet justify or explain.

Day 4 — Tracking, and testing it properly

Set up the link and the code, then walk the whole path yourself as a customer would. Most launch failures are here, and they are silent — attribution that never fires looks exactly like customers who are not interested.

1. 01Issue both a link and a codeThey fail in opposite circumstances. A link-only program silently loses every referral that travelled as a screenshot or across devices.
2. 02Share the link the way a real customer willInto a real messaging app on a real phone. Several apps rewrite or strip the query parameter that carries the identifier.
3. 03Convert on a different deviceClick on mobile, buy on desktop. If attribution does not survive that, you have a cross-device gap to close before launch.
4. 04Decline the consent prompt and try againIf attribution depends entirely on storage the visitor can refuse, you need the code path to work independently.
5. 05Confirm the reward actually releasesRun the qualifying event and watch the payout fire — including the delay you configured.

Day 5 — Terms and eligibility

Write the terms before launch, not after the first dispute. The clauses that matter are who counts as a new customer, what voids a reward, and when payment is released. Vague eligibility is what makes self-referral arguments unwinnable later.

• New customer defined by prior account, payment instrument and delivery address
• Whether employees and their families may take part
• Specific voiding behaviours, named rather than covered by a general clause
• When the reward is released, and whether it expires
• That rewards may be taxable to the recipient

Day 6 — Placement and copy

Put the ask where satisfaction happens, which is almost always inside the product rather than in an email sent later. Write the offer as one sentence naming both sides and the condition, because that sentence is what gets forwarded.

Pre-fill the share message. A blank text box asks the customer to compose a recommendation, and most will not — the drop-off between opening the share sheet and sending is where a large share of intent is lost.

Day 7 — Launch to a subset

Do not launch to everyone. Release to a slice of customers first, watch the tracking fire on real traffic, and confirm rewards release correctly before widening. A tracking bug found on ten percent of your base is an inconvenience; on all of it, it is a trust problem.

What to ignore in week one

Tiers, gamification, contests, leaderboards and multi-currency rewards. Each is a reasonable later addition and none of them fixes a program nobody has seen. Ship the simple version, get real participation data, then decide what the data says is missing.