Syntha Lab RU

CRM analytics: RFM, cohorts, retention, LTV/CAC, loyalty

A pack of reports on the customer base: RFM segments, cohorts and retention, churn, LTV and acquisition payback (CAC), NPS, the loyalty program and triggered communications. It is used to decide how much to pay for a new customer, whom to win back first and where to invest in loyalty.

Production, logistics, channels Brand and retail Verified: 2026-11-14

Open interactively →

Why this document

Average ticket and total revenue hide that some customers return and others buy once on a discount. So the marketing director, the CRM manager and the finance lead read these reports: they split the base into groups, see which cohorts are retained and compare LTV with acquisition cost. This decides whether to keep paying for traffic or switch to winning back existing customers. Without cohorts and LTV/CAC the ad budget is moved blindly: rising revenue may be bought with customers who never pay back.

At a glance

What is this document and why is it needed?
A pack of reports on the customer base: RFM segments, cohorts and retention, churn, LTV and acquisition payback (CAC), NPS, the loyalty program and triggered communications. It is used to decide how much to pay for a new customer, whom to win back first and where to invest in loyalty.
Who owns the document and how often is it rebuilt?
Run by the head of CRM or marketing with an analyst; refreshed weekly and monthly; the result goes to the commercial director, finance receives LTV and CAC for plans.
What mistakes are the most common?
LTV on revenue rather than gross profit → the customer looks more profitable than reality → use gross profit after returns. Averaging cohorts of different length → the curve is distorted → average only the elapsed months. Measuring mailing effect on all recipients → overstated because some would buy anyway → control group and uplift.

For whom

Brand
A brand builds its customer base in its own channels (site, stores, app); often only part of the customers is visible because wholesale and marketplaces do not share data. LTV is therefore computed on own sales, and the brand’s overall effect is estimated separately.
Retail
A retailer sees all receipts and customers by card: the basis for segmentation, clienteling and personal offers in store; the key issue is the share of identified receipts, without which RFM and cohorts are incomplete.

What it consists of

1. Sources and single customer (CDP)

  • Retailer: Store receipts, online orders, loyalty program, mailings, enquiries; matching by phone, e-mail and card into a single profile; share of identified receipts as the main quality indicator.

2. RFM

  • R: recency of the last purchase; F: number of purchases in the period; M: amount spent; each scale is split into 3–5 groups and the combinations give segments (champions, loyal, sleeping, lost); thresholds come from the distribution of the base.

3. Cohorts and retention

  • A cohort: customers who made their first purchase in the same period; the share active in following months (retention); the retention curve shows how many customers remain; churn is its inverse.

4. LTV and CAC

  • Brand: LTV: accumulated gross profit from a customer over the chosen period; CAC: acquisition spend ÷ number of new customers; the LTV/CAC ratio and payback period.

5. Loyalty and NPS

  • The loyalty program: bonus cost, liability for unredeemed points, uplift of frequency and ticket among members; NPS: share of promoters minus share of detractors on the “would you recommend us” question.

6. Triggers and clienteling

  • Retailer: Trigger scenarios: welcome after the first purchase, abandoned basket, winning back sleepers, birthday, re-order; clienteling: the salesperson sees the customer’s history and recommends personally.

Loyalty and points: what each side accounts for

  • Retailer: the loyalty program operator estimates the points balance, expected redemption and expiry, defers part of revenue at the relative standalone selling price and shows the owner the program cost separately from discounts.
  • Brand: sees customers only in its own channels (site, stores, app): computes points and LTV on them and assesses the wholesale and marketplace contribution separately because customer data stay with the platform or retailer there.

Brand and retailer: what differs in practice

  • Brand: the own site and app give the brand direct customer data; in wholesale and on platforms there is no such data so cohorts are counted only for direct sales.
  • Brand: counts acquisition cost by channel (advertising, bloggers, marketplaces) and compares it with first order margin and repeats.
  • Brand: builds the loyalty program to encourage repeat rather than a discount on the first purchase: accruals reflect a liability.
  • Brand: builds mailings and personalization on consents and segments (new, active, dormant) and controls the unsubscribe share.
  • Brand: limits the program by size and season: a brand with rare purchases (outerwear) measures retention by years.
  • Brand: does not mix customer data from platforms with its own base: the rules of transfer and use are set by the platform.
  • Retailer: identifies the customer by card or phone at the till: without a share of identified receipts cohort analysis is unstable.
  • Retailer: runs the loyalty program common to stores and the site and accounts for points as a liability.
  • Retailer: segments the base by frequency and recency of purchases and runs promotions for dormant customers rather than everyone.

How to read the result and what decisions follow

  • A falling retention curve after the first month is normal; a bad sign is when new cohorts retain worse than old ones: check acquisition quality and range.
  • If the first purchase does not pay back acquisition, pay for the channel only while repeats actually come; otherwise stop the channel.
  • Decision: the acquisition budget by channel, a retention program for dormant customers and the value-to-acquisition threshold for stopping a channel.

Control checks

  • The cohort size equals the number of unique customers with a first purchase in the month and no customer is in two cohorts.
  • Revenue from identified receipts plus unidentified ones equals total revenue for the period.

Checklist before launching a loyalty program and mailings (unverified: a reference, not legal advice)

  • Consent to personal data processing is obtained separately and stored; purposes and term are stated (Law 152-FZ, unverified against the text).
  • Advertising and mailings meet the advertising law requirements: consent to mailings, an option to opt out (unverified).
  • The loyalty program rules are published and the points liability is booked in accounting (FSBU 9/2025, paragraphs not checked, unverified).

Where the data comes from

Data sourceFields
Tills and online ordersreceipt, date, amount, styles, returns, customer ID, store
Loyalty programcard, points, tiers, consents
Marketingspend by channel, mailings, promotions, traffic sources
SurveysNPS scores, reasons, enquiries

How to build it step by step

  1. Retailer: Step 1: Export 12–24 months of receipts with the customer ID; compute the share of identified receipts.
  2. Step 2: Merge duplicate customers (phone, e-mail, card) and clean returns; build the customer profile.
  3. Step 3: Build RFM: set thresholds by base percentiles, assign scores and form segments; look at the share of customers and revenue.
  4. Step 4: Split customers into cohorts by first-purchase month; compute the share active in each subsequent month.
  5. Step 5: Compute LTV on gross profit after returns; then CAC by channel and the LTV to CAC ratio.
  6. Retailer: Step 6: Set triggers for segments: welcome, win-back of sleepers, retention of “champions”; define a control group.
  7. Step 7: Measure the effect with a control group; compute the uplift, not the mean revenue of mailing recipients.
  8. Step 8: Refresh curves, segments and the finance report (LTV, CAC, points liability) monthly.

Formulas

MeasureFormulasNote
Retention in month kRetention(k), % = cohort customers active in month k ÷ cohort size × 100For incomplete cohorts use only elapsed months and do not average with empty ones.
LTV (accumulated gross profit)LTV = first-order margin + Σ over k of Retention(k) × orders per active × average ticket × gross marginFor a period shorter than the full lifetime name the horizon (4-month LTV).
CACCAC = acquisition spend in the period ÷ number of new customers in the periodInclude marketing payroll and partner commissions; exclude the first-order discount only if it is already in the margin.
NPSNPS = % promoters (9–10) − % detractors (0–6)Compute from a survey, not a blast to everyone; compare with the previous period.
Retailer: Points liability (deferred revenue)Points share of price = Sale price × SSP of points ÷ (SSP of goods + SSP of points); SSP of points = points issued × expected redemption rate × standalone selling price per pointExample: a sale of RUB 10,000, 500 points of RUB 1 issued, 80% redemption: SSP of points RUB 400, 10,000 × 400 ÷ 10,400 = RUB 384.62 is allocated to points (goods revenue RUB 9,615.38); deferred revenue is recognized on redemption or expiry of points. The logic of FSBU 9/2025 and IFRS 15 (paragraphs not checked against the text); the bonusliab calculation.

Formulas in notation

Cohort retention
r_k = A_k / N
A_k customers who bought again in month k, N cohort size. Average by month: 28.7%, 21.4%, 17.6%.
Customer value over the horizon
LTV = m_1 × (1 + Σ_k r_k × n)
m_1 margin of the first order, n orders per active customer. Example: RUB 4,807 over four months.
Value to acquisition ratio
LTV / CAC
Example: 3.3. This is an illustrative metric; set target values for your own business.
Payback of acquisition by the first purchase
m_1 ≥ CAC
First order margin RUB 2,704 against acquisition RUB 1,450: pays back at once.

Worked example

First-purchase cohorts: retention by month

CohortCustomersMonth 1Month 2Month 3
January1,00028.0%20.5%17.6%
February90029.0%21.0%17.0%
March1,10027.0%21.0%18.0%
April1,00031.0%23.0%—
Cohort average28.7%21.4%17.6%

The average share of active customers is 28.7%, 21.4% and 17.6% in months 1, 2 and 3. At a ticket of RUB 5,200, a gross margin of 52% and 1.15 orders per active customer the 4-month LTV is RUB 4,807 (first-order margin RUB 2,704). With a CAC of RUB 1,450 the LTV to CAC ratio is 3.32; the first order pays back acquisition immediately. Cohort retention here means “active in the month”, not “not unsubscribed”: define it in advance.

Figures

Cohort retention by month after the first purchase
Heat map: 3×3; min 17, max 29Month 1Month 1Month 2Month 2Month 3Month 3JanuaryJanuary28January / Month 1: 2820.5January / Month 2: 20.517.6January / Month 3: 17.6FebruaryFebruary29February / Month 1: 2921February / Month 2: 2117February / Month 3: 17MarchMarch27March / Month 1: 2721March / Month 2: 2118March / Month 3: 18
  • 17 … 29 %
Data as a table
Month 1Month 2Month 3
January2820.517.6
February292117
March272118

Cohort retention by month after the first purchase. Rows are the month of the first purchase, columns are months after it; each cell is the share of the cohort that bought again. Cohorts with an incomplete period are not shown.

What builds customer value over four months
Waterfall: First purchase +2,704, Repeat, month 1 +892, Repeat, month 2 +665, Repeat, month 3 +546, Acquisition cost −1,450, Net of acquisition 3,35701,0002,0003,0004,0005,000First purchaseFirst purchaseFirst purchase: +2,704+2,704Repeat, month 1Repeat, month 1Repeat, month 1: +892+892Repeat, month 2Repeat, month 2Repeat, month 2: +665+665Repeat, month 3Repeat, month 3Repeat, month 3: +546+546Acquisition costAcquisition costAcquisition cost: −1,450−1,450Net of acquisitionNet of acquisitionNet of acquisition: 3,3573,357RUB
  • positive
  • negative
  • total
Data as a table
StepChangeLevel
First purchase+2,7042,704
Repeat, month 1+8923,596
Repeat, month 2+6654,261
Repeat, month 3+5464,807
Acquisition cost−1,4503,357
Net of acquisition—3,357

What builds customer value over four months. Margin of the first purchase plus margin of repeat purchases by month minus acquisition cost. Customer value RUB 4,807 against acquisition cost RUB 1,450 gives a ratio of 3.3. Illustrative figures.

Average share of returning customers by month
Line chart: Active share, %: 28.7 → 17.6161820222426283011.522.53Active share, %, 1: 28.7Active share, %, 2: 21.4Active share, %, 3: 17.6%
Data as a table
MeasureActive share, %
128.7
221.4
317.6

Average share of returning customers by month. The retention curve: after the first month the share falls quickly, then levels off.

Specifics by category

Apparel
Apparel: purchase frequency 2–4 times a year, seasonality; cohorts and RFM are computed by season rather than month; the key is the second purchase.
Footwear
Footwear: less frequent purchases and a longer cycle (1–2 times a year), higher ticket; seasonal change reminders and children’s sizes matter; LTV is computed over 12–24 months.
Accessories
Accessories: gift purchases, pre-holiday peaks; customers often buy for others, so “return” may not mean loyalty; personal recommendations matter.
Kids
Kids: the buyer is an adult, the wearer a child; segment by the child’s age and height; the purchase cycle is short (children grow) but the customer life is limited; parental consent for the child’s data is needed.
Men
Menswear: less frequent purchases but higher brand loyalty and easier re-orders of core items; size-repeat triggers work well.
Women
Womenswear: high frequency and a wide range, strong response to new arrivals and personal recommendations; bonuses and promotions are used more, so watch acquisition cost and the share of promotional orders.

Common mistakes

  • LTV on revenue rather than gross profit → the customer looks more profitable than reality → use gross profit after returns.
  • Averaging cohorts of different length → the curve is distorted → average only the elapsed months.
  • Measuring mailing effect on all recipients → overstated because some would buy anyway → control group and uplift.
  • A low share of identified receipts → conclusions on the base are biased → raise identification at the till and show the share in the report.
  • Ignoring points liability → bonuses look “free” in the report → show them as a deferred discount.

Russian specifics

  • The personal data law (152-FZ): consent to processing and mailings, purpose, retention period, customer rights; for the kids’ line children’s data only with parental consent. Rule: Law 152-FZ of 27.07.2006: Federal Law of 27.07.2006 No. 152-FZ on personal data.
  • Mailings and advertising to customers: the Advertising Law (38-FZ), subscriber consent to mailings is required; online advertising is labeled. Rule: Law 38-FZ of 13.03.2006: Federal Law of 13.03.2006 No. 38-FZ on advertising.
  • Bonus points: under FSBU 9/2025 (order 56n, mandatory from 2027) points per unchecked data are a separate performance obligation: part of the sale price is allocated to the points at the relative standalone selling price and revenue is recognized on redemption or expiry; the “points × cost × (1 − breakage)” formula is simplified; the standard paragraphs were not checked in Lab, agree with the accountant. 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).

Open interactively →