How to calculate the value of a referral?
The value of a referral is not the price of an invite nor a universal commission percentage. It results from the relationship between the cost to acquire customers through the program and the margin those customers generate over time. To evaluate the channel, calculate the referral CAC, estimate or observe the LTV based on contribution margin, track payback, and measure ROI with incremental results.
Invite, lead and paying customer are different stages. The calculation is only reliable when costs and conversions belong to the same period or cohort and when attribution avoids duplication with other channels.
A referral has value, but it also has cost
The referral marketing leverages existing relationships between customers and their contacts, but it is not free acquisition. Even if there is no payment per click or impression, the company may assume rewards, discounts, cashback, technology, communications, service, validation and commercial effort.
Before calculating, differentiate funnel events:
- Referral created: the participant generates a code, link or record.
- Referral shared: the referral is sent or used in an invite.
- Referred lead: the contact is identified and enters the company’s funnel.
- Valid conversion: the referred party completes the qualifying action and meets eligibility rules.
- Acquired paying customer: the conversion generates recognized revenue and remains valid after cancellations, duplications and defined criteria.
These numbers answer different questions. Dividing cost by referrals created produces cost per referral created, not CAC. For CAC, the denominator must represent new customers acquired according to a stable definition.
How to calculate CAC for referrals
The referral CAC shows how much the program spent, on average, to acquire a new valid customer.
Referral CAC = total program cost in the period ÷ new valid customers acquired through referrals.
Definitions:
- Total program cost: sum of all costs attributable to the channel in the period.
- New valid customers: referred customers who met the condition, became paying customers and were not nullified by program rules.
If the company spent in January but some referrals only converted in February, a simple monthly division can distort the result. There are two possible approaches:
- By period: account for costs and recognized customers within the same interval, noting the lag.
- By cohort: group referrals by origin date and track their costs and conversions until a defined maturation window.
Cohort analysis is usually more suitable when the sales cycle is long. Whatever the choice, use the same method when comparing periods and channels.
Which costs enter the calculation?
Considering only the referral commission makes CAC appear artificially low. Total cost can include:
- reward paid to referrers;
- benefit granted to the referred party;
- discounts, cashback, credits, points or products;
- technology licensing, implementation and integrations;
- email, WhatsApp, media, materials and other communications;
- service, marketing, sales, finance and operations time;
- validation, reconciliation and exception handling;
- fraud, undue rewards and unrecovered losses;
- payment method and transfer costs, when applicable;
- taxes and obligations recognized by accounting and legal responsibles.
Shared costs require a consistent allocation criterion. For example, if a platform serves referrals and loyalty, the company should document which portion it attributes to the channel. There is no single universal allocation rule, but changing the criterion to improve the result prevents reliable comparisons.
How to calculate LTV of the referred customer
LTV, or lifetime value of the customer, estimates the economic value generated during the relationship. To compare with CAC, it is better to use contribution margin, not just revenue.
A simplified formula for recurring businesses is:
Contribution LTV = average revenue per customer per period × contribution margin percentage × average relationship duration.
Or, when margin is already expressed in currency:
Contribution LTV = average contribution margin per period × average relationship duration.
Definitions:
- Average revenue per period: average revenue per customer in a month, quarter or other consistent interval.
- Contribution margin: revenue minus variable costs and expenses associated with the sale and service.
- Average relationship duration: average number of periods the customer remains economically active.
This is an approximation. More complete models can consider expansion, contraction, churn by segment, cost-to-serve, survival probability and the time value of money. The longer and more uncertain the projection, the more caution is required.
Separate:
- Observed LTV: margin actually recorded to date.
- Projected LTV: estimate of future margin based on assumptions.
Do not present projected LTV as realized value. Record the model, window and assumptions so the estimate can be reviewed.
Payback and ROI of the referral program
How to calculate payback
Payback shows how long the margin generated takes to recover CAC.
Estimated payback = referral CAC ÷ average monthly contribution margin of the referred customer.
The result is expressed in months when the margin used is monthly. The simple formula assumes relatively stable margin. If revenue varies greatly or there is a large onboarding cost, use a cumulative monthly flow and identify when cumulative margin exceeds CAC.
How to calculate ROI
ROI compares the gain attributable to the program with the investment made:
ROI = (incremental margin attributable to the program − total program cost) ÷ total program cost × 100.
Using margin instead of revenue reduces the risk of calling revenue a return. Incremental margin should represent the result that would not occur without the program, within the analyzed window. When incrementality cannot be proven, present the calculation as attributed ROI or an estimated scenario, not as causal effect.
For new programs, track at least two readings:
- Observed ROI: uses margin actually recorded to date.
- Projected ROI: uses LTV or expected future margin, identified as an assumption.
How to set a sustainable commission, discount or cashback
There is no one reward value suitable for all businesses. The ceiling depends on margin, operational cost, risk and the minimum desired return.
One way to structure the decision is:
Economic ceiling of the reward = expected contribution margin in the chosen window − other acquisition and operating costs − reserve for losses − minimum desired return.
The result is an economic limit, not an automatic payment recommendation. The effective reward also needs to be attractive to the participant, simple to communicate and compatible with legal, tax and accounting matters.
When defining the incentive:
- use a window compatible with cash flow and the business cycle;
- do not finance immediate reward only with distant and uncertain LTV;
- sum benefits for referrer and referred party;
- include caps on percentage rewards;
- consider refunds, delinquency and fraud;
- simulate conservative, base and favorable scenarios;
- test at controlled scale before scaling up.
If the incentive is cashback or usage credit, consult the article on referral program with cashback.
Hypothetical calculation example
Attention: all values and rates in this section are hypothetical. The simulation is only intended to demonstrate the formulas and does not represent Smartbis data, a benchmark or a budget recommendation.
1. Simulation funnel
| Stage | Hypothetical quantity | Use in analysis |
|---|---|---|
| Referrals created | 500 | Measures referral generation |
| Referrals shared | 300 | Measures actual use |
| Referred leads identified | 120 | Measures funnel entry |
| Conversions initially recorded | 50 | Still subject to validation |
| New paying and valid customers | 40 | Denominator for CAC |
2. Hypothetical period costs
| Component | Hypothetical value |
|---|---|
| Referrer rewards | R$ 6,000 |
| Benefits for referred parties | R$ 3,000 |
| Technology | R$ 2,000 |
| Communications | R$ 1,000 |
| Team time | R$ 4,000 |
| Validation, fraud and losses | R$ 1,000 |
| Total cost | R$ 17,000 |
Referral CAC = R$ 17,000 ÷ 40 = R$ 425 per valid customer.
Dividing by 120 leads would produce a cost per referred lead of approximately R$ 141.67, not CAC. Dividing by 500 referrals created would produce R$ 34 per referral created.
3. Hypothetical LTV and payback
Cohort assumptions:
- average monthly revenue per customer: R$ 300;
- contribution margin: 60%;
- monthly contribution margin: R$ 180;
- projected average relationship duration: 10 months.
Projected contribution LTV = R$ 300 × 60% × 10 = R$ 1,800.
Estimated payback = R$ 425 ÷ R$ 180 = approximately 2.36 months.
The LTV of R$ 1,800 is a projection. If the average duration has not yet been observed for this cohort, the company should review the estimate as real data emerges.
4. Hypothetical ROI in a six-month window
Assuming, only for the simulation, that the 40 customers remain active for six months and generate R$ 180 monthly margin:
Attributed margin = 40 × R$ 180 × 6 = R$ 43,200.
Attributed ROI = (R$ 43,200 − R$ 17,000) ÷ R$ 17,000 × 100 = approximately 154.12%.
This result does not prove incremental impact and would change if there were churn, delay, additional cost or customers who would have purchased without the program. The real analysis should replace assumptions with observed data.
How to compare referred customers and other channels
Do not assume referred customers have lower CAC, higher LTV or superior retention. Compare equivalent cohorts.
A proper comparison keeps constant:
- acquisition period;
- observation time;
- product, plan or category;
- region and customer profile;
- definition of paying customer;
- margin and allocation criteria;
- treatment of cancellations and delinquency.
Compare CAC, conversion, average ticket, margin, retention, payback and observed LTV. A cohort acquired twelve months ago should not be directly compared with one acquired two months ago, because the former had more time to generate revenue and cancel.
The best channel is not necessarily the one with the lowest CAC. A more expensive channel can generate higher margin and acceptable payback; a cheap channel can bring low-quality customers. The decision should consider unit economics and scalability.
Measurement and attribution errors
- Using invites as customers: artificially reduces CAC.
- Counting only the reward: omits technology, people and operations.
- Using revenue in ROI: ignores variable costs and overestimates economic return.
- Comparing projected LTV with observed CAC without caveat: mixes fact and assumption.
- Ignoring lag: assigns cost from one month to customers of another without method.
- Counting the same conversion in two channels: happens when referral and media both receive full credit.
- Changing window or formula: prevents historical comparison.
- Ignoring acquisitions that would have happened anyway: treats attribution as incrementality.
- Excluding fraud and cancellations: keeps invalid customers and rewards in the account.
- Comparing different cohorts: confuses channel with product, region, period or profile.
How to avoid double counting
Define an attribution rule before analysis. Examples include first touch, last touch, code used or a split between channels. No rule represents perfect causality; it organizes accounting.
Store referral, customer and transaction identifiers, as well as contact dates. Reports should distinguish:
- attributed conversion: received credit according to the rule;
- assisted conversion: participated in the journey without receiving full credit;
- incremental conversion: estimate of outcome that would not have occurred without the program, ideally supported by a proper test design.
Spreadsheet and variable checklist
A minimal spreadsheet can have the following tabs or blocks:
| Block | Variables | Verification |
|---|---|---|
| Funnel | Created, shared, leads, conversions and paying customers | Do definitions overlap? |
| Costs | Rewards, benefits, technology, communications, teams and losses | Is there a documented allocation criterion? |
| Revenue and margin | Revenue, variable costs and contribution margin | Does the analysis use margin, not just revenue? |
| Time | Referral, conversion, payment, cancellation dates and window | Do costs and customers belong to the same cohort? |
| LTV | Margin per period, retention, churn and projections | Are observed and projected separated? |
| Attribution | Source, assisted channel, rule and identifiers | Was the same acquisition counted once? |
| Quality | Repeat purchase, retention, delinquency and support | Do cohorts have equivalent periods? |
| Scenarios | Conservative, base and favorable | Are assumptions explicit? |
Before closing the report, confirm:
- the definition of acquired customer;
- the analysis window;
- the costs included and allocation criteria;
- the margin used;
- the data source;
- the LTV and churn assumptions;
- the attribution rule;
- the treatment of chargebacks and fraud;
- the comparison with equivalent cohorts;
- the separation between observed and projected results.
The value of a referral depends on the program’s economics
A referral is worth what its conversion generates in margin, net of the costs and risks required to acquire it. CAC, LTV, payback and ROI help turn an intuitive reward into an auditable economic decision.
To define audience, rules and launch, see how to plan and implement a referral program. If you need to centralize invites, registrations, benefits and tracking, learn about the solution to manage a referral program from Smartbis. The platform supports operations; sustainability must be validated with the business’s own data and criteria.