Compare Customer and Revenue Retention by Cohort

A business can finish a period with the same number of active customers it had at the start while losing some of the original customers and replacing them with new ones. It can also report more total revenue while revenue from the original group declines. Those statements are compatible because they describe different groups and measures. Before using a retention figure in an acquisition conversation, identify the cohort, the period, the activity rule and the amount being compared.

This guide uses original fictional amounts to separate customer-count retention, a same-cohort revenue ratio and total growth. It does not describe an actual seller, customer relationship, accounting policy, valuation or financing decision. No contract transfer, future purchase or lender approval is established by the arithmetic. The practical result is a clear explanation of what a defined comparison says and what additional evidence a real transaction would still need.

Define the starting group before reading the current total

A cohort is a group selected by a stated rule. In this exercise, it consists of the four customer identifiers with positive supplied amounts in the prior period. The group is fixed before examining the current results. A fifth identifier appears only in the current period and is recorded separately as new. These are deliberate teaching definitions, not a claim that every platform or business uses the same definition of an active or retained customer.

Write the period and selection rule alongside the result. A beginning-of-period contract population, customers with a prior-period purchase and customers whose first order occurred in a particular month can produce different groups. The label cohort does not make those groups interchangeable. For a real analysis, identify what the records support and agree the appropriate rule with the people responsible for the measure. Keep any unresolved definition visible instead of choosing a population because it produces a reassuring percentage.

Shopify's customer-report documentation illustrates that report configuration and cohort definition matter: its default customer cohort report groups customers by first-order timing, with configurable metrics and filters. Our fictional comparison does not reproduce that platform report. It selects a prior-period positive-amount group for a different teaching question. Reading the source helps explain why a familiar dashboard label should not replace an explicit definition in a separate acquisition discussion.

Keep the amount definition and customer identifiers consistent

The supplied amounts represent one fictional measure in a single reporting currency across two comparable periods. They are not bank receipts, orders, receivable balances and recognized revenue mixed together. For a real business, those categories can answer different questions and require appropriate accounting interpretation. Establish which measure an export actually contains before calling the result revenue retention. A ratio with matching arithmetic but mismatched inputs can be misleading.

Customer identifiers also need an understood basis. Separate accounts, sites, legal entities and consolidated customer groups may not represent the same unit. Preserve the source mapping used for the comparison and obtain clarification when an identifier changed or several accounts were combined. Do not count a renamed account as a new customer without evidence, or merge different relationships merely because a name looks similar. The fictional identifiers A through E avoid real identity and grouping questions so the calculation can be reproduced.

Use authorized information and keep real customer details within the appropriate process. The public CSV is wholly invented, not an anonymized client export. A real acquisition team might require controlled access, reconciliations and qualified review that this exercise does not perform. The aim of the example is to explain group membership and denominators, not to approve a data room, determine disclosure obligations or certify that an actual seller's records are complete.

Reproduce the five-row example before interpreting a rate

The four starting customers have prior amounts of one hundred for A, two hundred for B, three hundred for C and four hundred for D. Their combined prior amount is one thousand. In the current period, A has fifty, B has zero, C has four hundred fifty and D still has four hundred. The current amount from that same starting group is therefore nine hundred. Customer E has no prior amount and a current amount of three hundred.

The current total across all five identifiers is twelve hundred. That total includes nine hundred from the original group plus three hundred from the new customer. Four identifiers have positive current amounts: A, C, D and E. The count is again four, but the membership differs from the original four. Preserve B's zero row in the fixed cohort rather than dropping it and forgetting that the starting relationship is part of the denominator.

Check those totals before calculating a percentage. The prior cohort total, current cohort total and current all-customer total each have a separate role. Their values are not competing versions of the same number. The downloadable expected-results file names them separately so another reader can reproduce the original example. None of the numbers is an actual business result, a forecast or evidence about why a particular customer did or did not produce revenue.

Calculate customer-count retention using the fixed group

Under this exercise's rule, a starting customer is counted as retained when its current supplied amount is positive. A, C and D meet that rule; B does not. Three of the four starting customers are counted as retained, so the customer-count retention is three divided by four, or seventy-five percent. The new customer E is not added to the numerator because it was not a member of the starting group.

This measure counts customers rather than weighting them by amount. A retained customer with fifty contributes one count, just as a retained customer with four hundred fifty contributes one count. That is appropriate for the defined count question, but it does not describe how much amount was retained. A business could lose one small account or one large account and produce the same count percentage while having a different amount comparison.

Do not treat zero activity in one period as proof that a relationship has permanently ended. The exercise's rule identifies positive supplied activity for a defined period. A real customer might purchase irregularly, have a paused contract or have another status that requires separate evidence. If the business uses a different retained-customer definition, state it and calculate accordingly. The public example supplies no customer interviews, contract assessment or causal explanation for B's zero.

Compare revenue from the same cohort without adding new customers

The same-cohort revenue ratio in the example is current revenue from the starting group divided by that group's prior revenue. Nine hundred divided by one thousand equals ninety percent. The calculation includes expansion by retained customers, contraction and the zero amount from B. It excludes E's three hundred because E did not belong to the selected cohort. This is the stated teaching ratio; do not assume its label or method matches every company's net-retention metric.

Notice why the ninety-percent result differs from seventy-five-percent customer-count retention. The count measure gives each starting customer one place in the denominator. The amount measure weights the comparison through the amounts each customer supplied. C's increase partly offsets A's reduction and B's zero. The two percentages answer different questions, so reporting one in place of the other would conceal useful information even though each calculation is internally correct.

A reported company or platform metric needs its own definition, period and calculation method. The SEC's guidance on metrics in public-company MD&A emphasizes explanatory context such as a metric's definition, calculation and relevant assumptions. That is source context for clearer reporting, not a statement that this fictional private-acquisition worksheet is subject to that disclosure framework or satisfies it. No legal, accounting-compliance or securities-reporting conclusion is made here.

Separate contraction, lost activity and expansion in the explanation

The starting-group change can be described without collapsing it into a single rate. A falls from one hundred to fifty, a reduction of fifty. B falls from two hundred to zero, a reduction of two hundred. C rises from three hundred to four hundred fifty, an increase of one hundred fifty. D remains at four hundred. Together, negative two hundred fifty plus positive one hundred fifty gives a net reduction of one hundred, matching the movement from one thousand to nine hundred.

An additional illustrative comparison caps each current starting-customer amount at its own prior amount before adding. The capped values are fifty, zero, three hundred and four hundred, totaling seven hundred fifty. Dividing by the prior cohort total gives seventy-five percent. This chosen capped comparison removes the offset from C's expansion for the teaching exercise. It is not presented as a universal definition of gross revenue retention or a replacement for the actual business's agreed metric.

Keep the method visible if you use such a comparison. Negative amounts, refunds, currency changes, different product populations or other real-world features could require additional decisions that these nonnegative fictional rows do not address. Do not silently discard a negative amount or force it to zero to make a formula work. Ask the appropriate accounting or analytical question and document the chosen treatment before presenting a result from actual records.

Understand how total growth can coexist with a cohort decline

All-customer current revenue of twelve hundred is two hundred above the prior one thousand, giving twenty-percent total growth in the constructed example. The original cohort nevertheless declined by one hundred, and a new customer's three hundred more than offset that decline. Both statements belong in a clear explanation. Calling the twenty-percent increase retention would incorrectly attribute new-customer activity to the starting cohort.

The customer stock also reconciles separately. Four starting customers, less one with no current positive amount under the chosen rule, plus one new positive-amount customer, gives four currently active customers. An unchanged total count does not establish one hundred percent retention of the original group. It establishes that the current count equals the starting count under the stated definitions. Membership is the missing information that a simple start-versus-end comparison cannot supply.

In a real acquisition discussion, ask which result the speaker is referring to when they say the customer base grew or retention was strong. Request the definition and supporting reconciliation rather than assuming the words refer to one standard ratio. The exercise does not establish whether any of these fictional percentages is acceptable, financeable or worth a particular purchase price. A descriptive comparison can inform a question without determining a transaction decision.

Test the meaning with two explicitly separate variations

In the first original variation, change only C's current amount from four hundred fifty to seven hundred. The original cohort's current total becomes eleven hundred fifty, so its revenue ratio becomes one hundred fifteen percent. Three of the four starting customers still have positive current amounts, so count retention remains seventy-five percent. Expansion by one retained customer can make the amount ratio exceed one hundred percent while a starting customer still has zero activity under the count rule.

The capped comparison in that variation remains seven hundred fifty over one thousand because C's amount is still capped at its own prior three hundred. That invariance follows from the chosen formula, not from a business insight observed in the world. The example is deliberately constructed so the reader can inspect which part of the definition allows expansion to offset other changes and which part removes that offset.

In a second separate variation, use current amounts of one hundred, two hundred, three hundred and zero for A through D. Three starting customers again have positive amounts, but their current cohort total is six hundred, giving a sixty-percent same-cohort revenue ratio. This variation illustrates a different pattern of lost activity. It is not another period to append to the base dataset, a prediction of customer behavior or an alternative real seller outcome.

Keep missing information separate from recorded zero

A recorded zero in the teaching file is part of its deliberately complete inputs. An empty cell in a real export might instead mean missing, not applicable, delayed or unavailable information. Do not convert every blank to zero and call the result customer loss. Establish the field meaning and the completeness of the period before applying the activity rule. If the necessary information remains missing, the supported result may be an unresolved comparison rather than a rate.

Likewise, avoid changing the cohort after looking at the current results. Dropping B because it has zero would leave three retained customers out of three and erase the loss that the fixed starting group was intended to reveal. Adding E to the starting group would answer another question. Keep the selection rule stable, and explain any genuine correction to the source mapping or definition so a later reviewer can reproduce the revised comparison.

Comparable periods matter as well. A full prior year and an incomplete current year do not become equivalent because both columns are labeled revenue. Seasonality, reporting cutoff and accounting basis may require appropriate treatment. The exercise assumes comparable periods and nonnegative amounts solely to make its arithmetic transparent. It does not supply a general adjustment method for an incomplete export or establish the correct reporting basis for a particular business.

Review summaries against the groups they actually describe

A deliberately authored weak summary of the base case says customer retention was one hundred percent because four customers were active in each period. That sentence confuses the stock count with retention of the starting members. A supported summary says that three of four starting customers had positive current amounts, while one current customer was new. It then reports the original cohort's nine hundred against its prior one thousand and the separate all-customer current total of twelve hundred.

These are original editorial examples, not observed AI outputs or a measured error rate. If you use AI to help draft a real summary, compare each numerical statement and label with the authorized source and the agreed definitions. A fluent paragraph can still switch populations between sentences or call total growth retention. The separate Elixir summary-review guide addresses meaning preservation; this article supplies the specific cohort distinctions a reviewer would need for this exercise.

Prepare a short handoff that includes the selection rule, period, amount basis, identifiers, totals, formulas and unresolved questions. The recipient should be able to distinguish count retention, the same-cohort amount ratio and all-customer growth without guessing. Keep the actual decision with the responsible people and appropriate qualified advice. Reproducible arithmetic improves the explanation of a defined measure; it does not prove future retention, contract continuity, a suitable valuation or an acquisition result.

Label averages that condition on a different group

The prior amount per starting customer is one thousand divided by four, or two hundred fifty. The current amount per currently active customer across the whole fictional file is twelve hundred divided by four, or three hundred. That increase does not establish retention of the original group. It combines amounts and membership from the current population, including E. Give the average its own name rather than using it as a substitute for the fixed-cohort count or amount ratio.

The retained starting customers also happen to have a current average of three hundred because their nine hundred is divided by three. That equality is a feature of these invented numbers, not evidence that the populations are the same. Excluding B can be appropriate when explicitly asking about the average among retained customers, but not when silently changing the denominator of the retention calculation. A clear question determines the denominator; an appealing average does not determine the question.

Frequently asked questions

Does the same customer count at the start and end mean full retention?

No. In the fictional base case, four starting customers become three retained starting customers plus one new customer, leaving the current count at four. Retention of the fixed starting group is three out of four under the chosen positive-amount rule. A start-versus-end stock comparison cannot reveal membership by itself. Keep additions and lost activity visible rather than calling the unchanged total one hundred percent retention.

Why are the count and revenue percentages different?

They use different numerators and denominators. Count retention measures three qualifying starting customers out of four. The same-cohort revenue ratio compares nine hundred of current starting-group amounts with one thousand of prior starting-group amounts. Expansion and contraction affect the amount comparison differently from a one-customer-one-count measure. Both can be calculated correctly while answering different questions; label the result and retain the supporting definitions.

Can a revenue ratio exceed one hundred percent while a customer is lost?

Yes under the defined example. When only C's current amount rises to seven hundred, the same cohort produces eleven hundred fifty against a prior one thousand, while B still has zero. The amount ratio is one hundred fifteen percent and count retention remains seventy-five percent. This is a constructed illustration of the formula, not a prediction, a real company result or evidence that the loss of a particular customer is unimportant.

Is the capped comparison a universal gross-retention formula?

No. It is an explicitly chosen teaching calculation that limits each current starting-customer amount to its own prior amount before summing. A real company or platform may use a different population, period, amount basis or metric definition. Read and agree those details before comparing results. This guide does not determine reporting compliance or prescribe the treatment of refunds, negative amounts, currencies or other features absent from the fictional file.

Should new-customer revenue be included in the same-cohort ratio?

Not under this exercise's fixed starting-cohort definition. E's three hundred belongs in the current all-customer total and in the explanation of growth, but not in the current numerator for the original A-through-D cohort. Including it would change the question and produce a different measure. If another report defines its metric differently, describe that definition accurately rather than assuming it is interchangeable with this example.

Do these calculations show whether an acquisition is a good purchase?

No. The amounts are fictional and establish only the described arithmetic and group distinctions. A real transaction requires other evidence about accounting, customers, contracts, operations, price, financing and the buyer's circumstances, with appropriate qualified review. No valuation, lending, tax or legal conclusion, customer continuation forecast, book quotation or professional endorsement is claimed. The guide helps frame a clearer question and explanation rather than approving a purchase.