CRM chains and the loyalty program: triggers, frequency, consents, tiers, points lifetime and payback by cohort
The document links two things: automatic message chains by customer behavior (sign-up, abandoned cart, repeat purchase, win-back, birthday) and a loyalty program with tiers and points. It helps decide which chains to switch on and how much they bring above ordinary customer behavior, and how to calculate whether the loyalty program will pay back and when.
Why this document
A repeat purchase is usually cheaper than acquisition, so CRM seems free: “we sent an email and got an order”. But customers would place some of the orders without the email, and discounts and points are real costs. An honest calculation separates the increment from habitual purchases through a control group and compares it with the costs of messages, discounts and rewards. The document is read by the CRM manager, the marketing director and the finance director; it is needed to decide what to automate, where to cut the discount and how to evaluate the points liability.
At a glance
- What is this document and why is it needed?
- The document links two things: automatic message chains by customer behavior (sign-up, abandoned cart, repeat purchase, win-back, birthday) and a loyalty program with tiers and points. It helps decide which chains to switch on and how much they bring above ordinary customer behavior, and how to calculate whether the loyalty program will pay back and when.
- Who owns the document and how often is it rebuilt?
- The CRM manager and the marketing director; chains are reviewed monthly by control group results, the loyalty program once a year; the finance director approves the points liability.
- What mistakes are the most common?
- All orders of recipients are counted as the chain’s orders → the contribution is doubled → keep a control group and count only the increment. Brand: gives a discount in every chain → customers wait for the email and the price is devalued → limit the discount and introduce non-monetary privileges. Retail: does not count the points liability → costs are hidden and surface at expiry → estimate the balance with expected expiry and put it into the budget.
For whom
- Brand
- The brand works with customers of its own site, app and own stores; purchase data of wholesale partners and marketplaces is not available, so brand CRM is built around D2C and subscription; loyalty helps move a customer from a platform to own channels.
- Retail
- The retailer sees purchases across the chain and several brands, so it has the richest customer base: a single card, tiers, cross-brand offers; a retailer’s loyalty program often works as an ecosystem with partners.
What it consists of
Segments and triggers
- Segments: new (before the first purchase), first-time buyers, repeat buyers, lapsed, VIP; a basic split by recency, frequency and spend (RFM).
- Triggers: sign-up, abandoned cart, first purchase, repeat purchase after N days, win-back, birthday, product return; each fires on an event, not on the calendar.
- Channels and cost: email, push, SMS, messengers; channels have a different message price and deliverability; an expensive channel is kept for messages with high returns (cart, win-back).
- Frequency and fatigue: a limit of messages per person a week, chain priority on overlap, a pause after a purchase; a growing share of unsubscribes is a signal to lower frequency.
- Consents: receiving advertising over telecom networks needs prior consent, processing personal data a separate one; unsubscribing works in one step.
Measurement and experiments
- Control group (holdout): part of the audience (for example 10%) does not receive the chain; the difference in conversion between recipients and control is the increment, the rest customers would do themselves.
- Chain contribution: incremental orders × order margin − message costs − the cost of discounts on all orders of the chain (including those that would have happened without it).
- A/B tests: subject, time, offer, channel; change one thing at a time and keep the sample sufficient; a conversion rise on a small sample is often noise.
- Lags: an increment from a message may come after several days; the measurement window is set in advance (for example 7 or 14 days) and is not changed after seeing the result.
Loyalty program design
- Goal: repeat purchases, a higher average basket, data collection, protection from churn to a competitor or a move from a platform to own channels; the goal determines the mechanics and the indicator.
- Mechanics: points for purchases, cashback, tiers by annual spend, closed sales and early access; non-monetary privileges are cheaper than discounts and often stronger.
- Tiers and points lifetime: status by spend over 12 months, the lifetime of points (expiry) sets the rhythm of purchases and reduces the liability; too short a term angers customers.
- Points liability: unused points are a liability of the company; it is estimated at nominal value allowing for expected expiry, and the effect on revenue and margin is accounted for under revenue accounting rules.
- Partners and ecosystem: shared points with partners extend value but complicate settlements and risks; fix settlement terms between partners.
What the brand does
- Brand: builds CRM around its own site and stores, joins receipts and orders into one base and asks for consent to messages at every touch.
- Brand: launches 4-5 basic chains (welcome, cart, repeat purchase, win-back, birthday) and immediately sets a holdout to know their real contribution.
- Brand: uses loyalty to move marketplace customers into own channels: a registration bonus, closed sales, early access.
- Brand: limits the depth of CRM discounts so as not to devalue the main price and not to teach customers to wait for a discount email.
What the retailer does
- Retail: unites the card and data across all stores and channels, segments by behavior and counts the share of receipts tied to a customer (CRM coverage).
- Retail: builds the loyalty program as a system of tiers with chain-wide points, counts the liability and expected expiry and puts it into the budget.
- Retail: personalizes offers by behavior (category, brand, size) and does not send the same to everyone; lowers frequency for those who unsubscribe.
- Retail: agrees with brands on joint promotions for card holders and on cost sharing; fixes terms in the contract.
Where the data comes from
| Data source | Fields |
|---|---|
| Receipts and orders tied to a customer | customer, date, amount, margin, channel, points and discounts used |
| Mailing and messenger platform | sent, delivered, opened, click, unsubscribe, message cost |
| Control group | a list of customers without the chain and their orders in the same window |
| Points ledger | accrued, redeemed, expired, balance at nominal value and terms |
How to build it step by step
- Step 1: Formulate the goal of CRM and the loyalty program with one indicator (repeat purchases, basket, share of customers with a card).
- Step 2: Retail: build a single customer base and count CRM coverage: what share of receipts is tied to a customer and what share of customers gave consent.
- Step 3: Brand: choose 4-5 chains and for each set the trigger, channel, number of messages, offer and control group.
- Step 4: Launch the chains and measure the increment against the control over a window of 7-14 days (see the example), count the contribution with costs and discounts.
- Step 5: Brand: stop or rebuild chains with a negative contribution, strengthen those where the increment exceeds costs.
- Step 6: Design the loyalty program: mechanics, tiers, points lifetime, rewards; count the reward cost per rouble of purchase.
- Step 7: Model the payback by cohort: start-up costs, the increment of purchases by month, rewards and fixed costs (see the figure), set a pessimistic scenario.
- Step 8: Retail: count the liability for unused points and the expected expiry; agree the accounting with the accountants.
- Step 9: Once a quarter review chains and tiers by the holdout and cohorts.
Formulas
| Measure | Formulas | Note |
|---|---|---|
| Chain contribution | Contribution = Incremental orders × Basket × Margin − Audience × Messages × Message price − All orders × Basket × Discount | Incremental orders = all chain orders × (1 − the share of orders that would have happened without it); the share is taken from the control group. |
| Cumulative contribution of the loyalty program | Cumulative(t) = −Setup + Σ(Members × Incremental orders per month × Basket × (Margin − Reward)) − Fixed × t | The payback is the month when the cumulative contribution becomes positive; count an optimistic and a pessimistic scenario of the increment. |
| Points liability | Liability = Points balance × Nominal value × (1 − expected expiry) | Expected expiry is taken from closed cohorts, not from hopes; a new program is assessed cautiously. |
Formulas in notation
- Cumulative program contribution
W_t = −S + Σ_{i≤t} N · u_i · P · (m − r) − F · t
S: setup, N: members, u: incremental orders per member a month, P: basket, m: margin, r: reward, F: fixed costs a month.
Worked example
A teaching monthly contribution of five chains: audience, conversion, share of orders without the chain, costs and discounts
| Chain | Audience | Orders | Incremental orders | Increment margin, ₽ | Message costs, ₽ | Discounts, ₽ | Contribution, ₽ |
|---|---|---|---|---|---|---|---|
| Welcome after sign-up | 2,000 | 120.0 | 84.0 | 254,100 | 7,200 | 33,000 | 213,900 |
| Abandoned cart | 1,500 | 165.0 | 90.8 | 249,563 | 9,000 | 0 | 240,563 |
| Repeat purchase (after 30 days) | 3,000 | 135.0 | 67.5 | 193,050 | 12,000 | 0 | 181,050 |
| Win-back (90+ days) | 5,000 | 90.0 | 72.0 | 190,080 | 30,000 | 43,200 | 116,880 |
| Birthday | 600 | 54.0 | 35.1 | 115,830 | 3,600 | 32,400 | 79,830 |
| Total | 12,100 | 564.0 | 349.4 | 1,002,623 | 61,800 | 108,600 | 832,223 |
The numbers are illustrative. The “Win-back” chain gives a high increment (the control group hardly buys), but the discount and costs eat part of the contribution; the “Repeat purchase” chain counts half of the orders as its own though customers would have bought anyway: without a holdout the contribution is doubled.
Figures
- Increment margin
- Costs and discounts
- Contribution
Data as a table
| Measure | Increment margin | Costs and discounts | Contribution |
|---|---|---|---|
| Welcome after sign-up | 254k | 40.2k | 214k |
| Abandoned cart | 250k | 9,000 | 241k |
| Repeat purchase (after 30 days) | 193k | 12k | 181k |
| Win-back (90+ days) | 190k | 73.2k | 117k |
| Birthday | 116k | 36k | 79.8k |
Chain contribution a month. Increment margin, message costs and contribution by chain: chains with a low increment have a contribution closer to zero.
- Base scenario
- Half the increment
Data as a table
| Measure | Base scenario | Half the increment |
|---|---|---|
| 0 | −600 | −600 |
| 1 | −410 | −525 |
| 2 | −243 | −461 |
| 3 | −99 | −409 |
| 4 | 22 | −369 |
| 5 | 132 | −334 |
| 6 | 230 | −305 |
| 7 | 316 | −282 |
| 8 | 391 | −264 |
| 9 | 455 | −253 |
| 10 | 518 | −241 |
| 11 | 570 | −235 |
| 12 | 622 | −229 |
Loyalty program payback by cohort. The cumulative contribution of a cohort of 5,000 members in the base scenario and at half the increment; the zero point is the payback time.
Map of trigger chains. From sign-up to the first and repeat purchase and to win-back: each event starts its own chain.
- Sign-up → Welcome; Abandoned cart
- Welcome → First purchase
- Abandoned cart → First purchase
- First purchase → Repeat purchase; Lapsed (90+ days)
- Repeat purchase
- Lapsed (90+ days) → Win-back
- Win-back
Specifics by category
- Apparel
- Apparel: seasonal triggers (a change of collection, school, holidays), a “complete the look” chain and a size reminder; size returns reduce the contribution, so account for them in the order margin.
- Footwear
- Footwear: a longer purchase cycle tied to the season; chains about a change of season (autumn, spring), a reminder about size and care; returns by size and last reduce the contribution.
- Accessories
- Accessories: gift occasions and birthdays, a high margin and add-on sales to the main purchase; a “goes with your purchase” chain works better than win-back.
- Kids
- Kids: the buyer is an adult and the child’s size grows; a “time for the next size” chain by age and season; the parent’s consent for the child’s data, not the child’s own.
- Men
- Men’s: rare purposeful purchases and a high share of basics; chains are rarer but more precise (repeat of basics, change of season); discounts are less effective than a reminder and convenience.
- Women
- Women’s: higher purchase frequency, more new arrivals and looks; personal selections, early access and closed sales are strong; frequency control matters so as not to tire the audience.
Common mistakes
- All orders of recipients are counted as the chain’s orders → the contribution is doubled → keep a control group and count only the increment.
- Brand: gives a discount in every chain → customers wait for the email and the price is devalued → limit the discount and introduce non-monetary privileges.
- Retail: does not count the points liability → costs are hidden and surface at expiry → estimate the balance with expected expiry and put it into the budget.
- A mailing is sent without consent or an unsubscribe → a fine and sender blocking → keep proof of consent and make unsubscribing one step.
- Loyalty is launched without a pessimistic scenario → payback is promised but does not come → count at half the increment and set a stopping point.
Russian specifics
- Advertising over telecom networks (SMS, messengers, email) is allowed only with the recipient’s prior consent (the advertising law, art. 18); the burden of proving consent lies on the distributor. Rule: Law 38-FZ of 13.03.2006: Federal Law of 13.03.2006 No. 38-FZ on advertising.
- The collection and storage of customers’ personal data in a loyalty card and CRM are subject to the personal data law: purpose, consent, storage term, subject rights; data of Russian citizens is recorded and stored in databases located in Russia. Rule: Law 152-FZ of 27.07.2006: Federal Law of 27.07.2006 No. 152-FZ on personal data.
- Brand: accounting for the loyalty program (revenue recognition when points are accrued, the liability) follows the revenue accounting standard; check the approach with the accountant and the auditor. Rule: Order 56n of 16.05.2025: FSBU 9/2025 Income (Ministry of Finance order of 16.05.2025 No. 56n) (rule check not completed).
- Retail: taxation of points and discounts, partner rewards and gifts to customers (VAT, corporate income tax, tax for the customer) depends on the scheme; unverified, consult a tax specialist.