The owner of an online coffee shop in Almaty shows you a partner's report: "Twenty customers brought in another twenty. Each customer brings one more, so we won't need ads soon." There were a hundred buyers in the programme, but eighty invited nobody. Removing them from the denominator made the result five times larger before anyone discussed the money.
Suppose the shop is in the second month of a separate referral programme. The numbers in this lesson are illustrative; order economics are the same as in the influencer campaign: a kilogram costs 15,000 ₸, leaving 6,000 ₸ after beans, roasting, packaging and delivery. This time both the new buyer and the person who introduced them to the shop receive a gift.
From a recommendation to the next buyer
A referral loop has several steps. Someone enjoys their coffee, recommends it to a friend, the friend places a first order and gets a reason to invite someone else. In Reforge's description of growth loops, the output of one pass feeds the next. For our shop, that means new buyers who can recommend the coffee in turn.
There has to be a reason to recommend it first. A convenient link helps share the offer, but the customer still has to explain why their friend would want it. If delivery was disappointing, a promised gift turns the message into a favour for the shop rather than a personal recommendation. Offer the invitation after the buyer has received the order and tried the coffee.
A programme and a loop are also different things. You can award a bonus for bringing someone in once. To describe recurring growth, you need to see new buyers inviting the next group. In AARRR, this is the referral stage feeding new acquisition.
Agree on who counts first
The shop counts a new buyer's first paid order, delivered and not refunded at the reporting date. Registrations, link clicks and existing customers' orders are kept separately. A reward for a new account would be too far from a sale here: one person could create several accounts without buying anything.
The report records how a new buyer is identified, the allowed time from invitation to order and when the data is ready. In this programme, for example, the first order must occur within 30 days of the invitation. Give the last invitee the full 30 days; if an order is still being delivered or a refund question is unresolved, mark the result as incomplete. Keep the rule fixed during measurement instead of changing it to improve the number.
For a service used by teams, the unit will be different. A colleague invited into an organisation's existing paid account expands usage within that customer. If the aim is to acquire new companies, count a new organisation that completes the chosen action. Employees and organisations cannot share the same denominator.
All hundred buyers stay in the denominator
For the first calculation, take a simplified export: a hundred eligible buyers had the opportunity to take part. Twenty sent two hundred invitations. The recipients are all different, none had bought from the shop before, and each received one invitation. Every observation window has closed, and twenty recipients met the first-order condition.
The diagram calls them activated users. For our shop, activation in this calculation means a completed first order.
| Metric | Calculation | Result |
|---|---|---|
| Buyers who sent invitations | 20 / 100 | 20% |
| Invitations per sender | 200 / 20 | 10 |
| Invitations per original participant | 200 / 100 | 2 |
| Invitees who met the first-order condition | 20 / 200 | 10% |
| New buyers per original participant | 20 / 100 | 0.2 |
The final number is called the K-factor for the chosen cycle. In this simplified model, K = i × c: average invitations per original participant multiplied by the share of invitees who reached the result. That gives 2 × 0.1 = 0.2.
The partner used ten invitations per sender and got 10 × 0.1 = 1. That measures new buyers per sender. To bring the whole original group back into the calculation, multiply by the share who sent invitations: 0.2 × 10 × 0.1 = 0.2. For the owner's report, it is simpler to divide twenty new buyers directly by the original hundred.
Handle duplicates and unknown sends
In a real programme, someone can receive links from two friends, click on a phone and buy on a computer. Before launch, choose who gets credit for that order and keep the connection between invitation, buyer and order. One possible rule is the first confirmed invitation within the chosen window. This is a programme decision; no attribution rule wins in every situation.
Review existing customers, duplicate accounts, cancelled orders and self-invitations separately. Deduplication must carry through the whole calculation. After cleaning the data, you cannot divide unique buyers by raw clicks and call it the conversion rate of unique invitees.
A "Share" button does not tell you whether the person sent a message. If the channel does not return that fact, name the event opening the share dialog. The shop may reliably observe referral-link visits and orders without knowing how many invitations were delivered. In that case, leave invitation-to-order conversion blank. You can still calculate the final twenty confirmed new buyers per hundred original participants.
The buyer shares the link with their friend. This mechanic does not require the shop to collect someone else's address book for a mailing. The recipient opens the offer and decides whether to order.
The shop pays for the gift
Suppose each confirmed first order earns two sample packs: one for the newcomer and one for the referrer. Each pack costs the shop 1,000 ₸, as in the influencer example. For this calculation, assume all rewards have been fulfilled and their cost already includes delivery. Handling this batch of orders and rewards costs another 10,000 ₸.
Twenty first orders contribute 20 × 6,000 = 120,000 ₸ before programme costs. Gifts cost 20 × 2 × 1,000 = 40,000 ₸, or 50,000 ₸ including administration. That is 2,500 ₸ per buyer credited to the programme, leaving 70,000 ₸ contribution from the batch after those costs. The business's fixed costs are still outside the calculation.
This reports orders attributed to the programme. Some friends might have bought without a gift, while some buyers may forget the link. So 2,500 ₸ cannot be called the cost of an additional customer. That conclusion needs a separate effect measurement plan, accounting for friends in different groups sharing the offer and mixing the test conditions.
Points instead of gifts do not remove costs. Describe when they can be redeemed, what happens after a refund and how their cost is calculated. If points have already reduced the order payment, do not subtract that discount from contribution again; see the loyalty programme economics example.
How quickly the loop repeats
K = 0.2 over seven days and K = 0.2 over three months produce the same number of new participants per original participant at different speeds. Alongside the coefficient, the report needs the group's starting point, its window rule and the time to the chosen result.
In an ideal model of identical generations, K above one produces growth from one generation to the next. At the shop, people's audiences overlap in the same chats, gifts can lose their appeal, and new buyers may recommend less often than the first group. The next report row therefore follows the twenty new buyers: how many were offered participation, how many invited others and how many new orders they brought.
Repeat purchases by existing customers belong in retention. They do not enter an acquired generation for a second time. This keeps two distinct questions in the report: how many new people the programme brought and how many later stayed with the shop.
Exercise: prepare the calculation for your project
Choose a business whose customers have someone to recommend the product to. Build one table for a fully observed group.
A partner's table lists fifty senders out of five hundred eligible participants, bringing in a hundred new buyers. Their K = 2 describes senders: 100 / 50. For the entire group, K = 100 / 500 = 0.2. Correcting the denominator still leaves unique buyers, a closed window and completed orders to establish.
The paragraph for the owner should name the group, period and buyers per participant, then give the costs and the decision. That might be to continue a small launch to follow the next generation, or to fix duplicate-account handling first. One report like this cannot support a promise of growth without advertising.
Video: how referral mechanics work
The video shows a referral mechanic. Define the invitation's result, period and reward cost separately for your own project.