The media buyer at an online coffee shop wants to raise bids on Instagram. A first order from there currently costs 5,000 ₸ in ads, and he is ready to pay up to 10,000 ₸: "Customers come back afterwards, don't they?" The owner asks you whether he's right. The answer depends on how much a customer brings the shop after the first order, and working that out is the job of CVM. We will use the same shop as in the RFM analysis; the numbers are made up.
Where CVM sits in the ecosystem
CVM (Customer Value Management) connects the two halves of marketing. Paid media buys first orders, analytics shows where they came from, and the CRM holds every customer's history. Salesforce, for one, describes a CRM as software that helps a company build customer relationships.
Neither the ad account nor the CRM will answer three questions, and CVM exists to answer them:
- what a customer is worth, as opposed to a single order;
- how much you can pay for a first order;
- which groups of customers deserve money and time after the purchase, and which don't.
What a customer brings in
You start from the same export of a year's paid orders that RFM uses. What you count is contribution, what's left of an order after beans, roasting, packaging and delivery: for this shop, 6,000 ₸ out of a 15,000 ₸ order, as in unit economics. Revenue won't do, because the shop pays for ads and discounts out of contribution.
| Group | Customers | Orders a year | Contribution a year | Contribution per customer |
|---|---|---|---|---|
| Ordered once | 1,560 | 1,560 | 9,360,000 ₸ | 6,000 ₸ |
| Ordered more than once | 840 | 5,640 | 33,840,000 ₸ | 40,286 ₸ |
| All | 2,400 | 7,200 | 43,200,000 ₸ | 18,000 ₸ |
35% of customers bring about 78% of the contribution. The average of 18,000 ₸ per customer hides the gap between groups: some bring 6,000 ₸, others around 40,000 ₸ on average. That's why value is counted by group.
What a new customer is worth
The media buyer is acquiring a newcomer, and nobody knows in advance which group that person will join. Over the summer roughly a quarter of new customers placed a second order within 60 days, which is how the shop measures retention. For a rough annual forecast, suppose newcomers who return within those 60 days place as many orders in their first year as today's repeat buyers placed over the past 12 months:
- first order: 6,000 ₸ from everyone;
- if the newcomer returns and orders as assumed, that is about 5.7 more orders, or 34,286 ₸ of contribution;
- with a 25% return rate, the forecast allocates 34,286 × 0.25 ≈ 8,600 ₸ of repeat-order contribution to each new customer.
That is about 14,600 ₸ in the first year. This is a rough forecast: the current base includes people who have bought from the shop for years and may describe a newcomer's first year poorly. Three newcomers out of four made no second order within 60 days; that rate does not tell you what they buy later. Nor can an annual total tell you when 10,000 ₸ of ads pays back: that needs a schedule of repeat orders. So compare the forecast with cumulative contribution from cohorts over the same number of months after the first purchase.
The answer to the media buyer is to keep the ad cost of a first order at or below 6,000 ₸ for now, so its contribution covers acquisition. That ceiling leaves no margin for fixed costs. Raise it once mature cohorts show both the contribution from newcomers and when it arrives. Paying 10,000 ₸ today is a bet that the forecast will hold.
Retention changes the price of a first order
For a monthly cohort of 150 newcomers, take the target from the SMM strategy: 50 second-time buyers instead of 38. That is 12 extra buyers. Their second orders contribute 12 × 6,000 = 72,000 ₸.
If each of these 12 places as many repeat orders in their first year as today's returning customers place in a year, the forecast extra contribution is 12 × 34,286 ≈ 410,000 ₸. That is about 2,740 ₸ per newcomer across the cohort of 150. The first amount, 72,000 ₸, covers only second orders within 60 days. The annual forecast includes those orders and later ones, so the two amounts must not be added together.
Once mature cohorts confirm both the increase and the later purchases, the shop can revisit its acquisition ceiling. The extra 2,740 ₸ is a guide only with the same annual payback horizon and after accounting for retention costs.
Segments: who to spend on
Who gets which message is decided by the RFM breakdown as of 1 October. CVM adds money to its segments: what is known about their value and how much it makes sense to spend.
| Segment | Customers | What we know about the money | How much to spend |
|---|---|---|---|
| Regulars | 300 | Order at their own pace | No discount without a test: 10% off an order is 1,500 ₸ out of 6,000 ₸ contribution, including on orders that would happen anyway |
| Newcomers | 220 | The second order decides whether a person brings 6,000 ₸ or around 40,000 ₸ | A guide and one reminder with consent; cost the sending and the team's time |
| Due a reminder | 320 | Missed one order | One message, no discount |
| Valuable but slipping | 80 | 520 orders a year, 3,120,000 ₸ contribution | The owner's own time |
| Lapsed | 1,480 | Mostly one order each | A message through a permitted channel if the expected contribution pays for preparing and sending it, or nothing |
The "valuable but slipping" brought 39,000 ₸ of contribution each over the year on average (3,120,000 / 80). If three return and order as much in the following year, that is about 117,000 ₸ of contribution, not yet counting the owner's time: 80 personal conversations with preparation and replies are work too. Lapsed customers who agreed to emails can be sent one about a new lot, but it still has to be prepared, sent and answered when people reply.
One person can fall under several rules: a newcomer with an open complaint, a regular who unsubscribed from the newsletter. The order is written down in advance: service first, meaning the reply to the complaint and the order status, then help such as a grind guide, and only then sales. In this example the shop capped promotional messages at one per person per week, whoever sends them: the manager, the marketer or an automated flow. Before sending, it checks consent and the current state: the segments were calculated on 1 October, and by mid-month some customers due a reminder have already ordered.
The picture shows the same principle on an online course with assignments: the next step depends on the person's state.
Discounts: count them in contribution
A discount for someone who would have bought anyway spends contribution without adding an order. In early October the owner wanted to send a 15% promo code to the whole base. 15% of a 15,000 ₸ order is 2,250 ₸, and an order with the code leaves 3,750 ₸ of contribution instead of 6,000 ₸.
If the 300 regulars order as usual but now use the code, the shop gives away 300 × 2,250 = 675,000 ₸ on orders that would have happened anyway. To earn that back, the code has to bring 675,000 / 3,750 = 180 orders from people who wouldn't have ordered without it. For comparison, the "due a reminder" and "valuable but slipping" segments together hold 400 people. How many orders the code really added can only be shown by a group without the code, picked at random before the send.
Exercise: customer value in your own business
You need an export of paid orders for 12 months: customer ID, date and amount. A client's business, your own or a friend's will do.
You will most likely get the contribution per order from the owner or the accountant: the CRM holds revenue. If the owner doesn't know it, work it out together using unit economics. Otherwise every figure in the exercise ends up in revenue and is overstated by as many times as revenue exceeds contribution, two and a half for the coffee shop.
If ads already pay more for a first purchase than it contributes, the business lives on repeat purchases, and that assumption should be checked before anything else. To do that, value is calculated on a closed cohort rather than the whole base: the base mixes newcomers with customers who have been buying for years, and those tell you nothing about a newcomer.
In a subscription business where everyone pays the same, the "once" and "more than once" groups are replaced by months of payment. That's how the cumulative contribution per salon is counted at the online booking service: by cohort of first payment, including those who left.
Video: CRM marketing
A video from the RetailCRM channel. As you watch, note where the examples count customer value and where they only count messages sent.