When a new customer signs up and redeems your welcome offer, it feels great. But that first thrill can hide a costly truth: is that discount or free gift actually bringing long-term value, or simply accelerating churn and masking acquisition inefficiency? I use cohort analysis as my go-to way to answer this question. In this post I'll walk you through the practical steps I use with SMEs to prove — or disprove — whether a welcome offer is worth its cost.
Why cohort analysis (not “average” metrics) is essential
Average metrics lie. A single average retention rate or average order value can hide divergent behaviours from profitable, engaged customers and a cohort of bargain-hunters who never return. Cohort analysis slices customers by the date (or source) of first purchase and then tracks how each group behaves over time. That lets you see whether the welcome offer improves early conversion at the expense of longer-term value.
Put simply: cohort analysis answers the question I get asked most often — “did the welcome offer buy us loyal customers, or just a spike?” — with evidence rather than intuition.
What to measure: the KPI set I always recommend
Start with a compact set of metrics you can compute from order and customer data. I focus on the ones that map directly to business outcomes:
With those numbers you can answer: did the welcome offer increase acquisition? Did it reduce AOV for the cohort? How long before the cohort becomes profitable?
Designing the cohort experiment
There are two practical approaches, depending on your systems and traffic:
I always recommend A/B where possible. For many SMEs, a pragmatic compromise is to run staggered rollouts by region or traffic source to create quasi-experimental cohorts.
Step-by-step: building the cohort table
Here’s the minimal cohort table I build in a spreadsheet or BI tool. Columns are time buckets (week or month since acquisition) and rows are cohorts by acquisition week/month. Cells contain the metric (e.g. retention rate or cumulative revenue per user).
| Acq month | Week 0 rev/user | Week 4 rev/user | Week 12 rev/user | Week 24 rev/user |
|---|---|---|---|---|
| Jan | £8.50 | £1.00 | £3.20 | £4.80 |
| Feb (with offer) | £5.00 | £0.80 | £2.50 | £4.10 |
| Mar (with offer) | £5.20 | £0.90 | £2.70 | £4.30 |
In this simplified example, you can see the first-order revenue per user is lower for the "with offer" cohorts (because of the discount), but cumulative revenue narrows over time. The key questions are: when does cumulative revenue cross the cost line (welcome offer + CAC) and what happens to repeat behaviour?
Calculating the real cost of the welcome offer
Many teams make the mistake of only counting the face value of the discount. The true cost includes:
When I model ROI I compute an adjusted CAC = marketing CAC + allocated offer cost. Then I calculate payback period = months until cumulative gross margin per user > adjusted CAC. If the payback is within your acceptable window (often within 6–12 months for SMEs), the offer can be justified; if it’s beyond that, you’re subsidising customers for too long.
What good and bad cohort patterns look like
Over the years I’ve come to recognise a few archetypes:
Examples from the field
I worked with a D2C skincare brand that ran 20% off for new subscribers. Initial conversion doubled, but first-order gross margin per customer halved. Cohort analysis showed repeat rate at 6 and 12 months was the same as pre-offer customers, so cumulative LTV recovered by month 9 — after which cohorts became profitable. For that business the welcome discount made sense because they had low CAC and a product with sticky repurchase behaviour.
Contrast that with a niche fashion retailer where a seasonal 30% new-customer code brought lots of returns and low repeat purchase. Cohort LTV never recovered past the acquisition cost. The fix there was to switch to a non-discount welcome (e.g. free shipping threshold or a low-cost gift) and couple the incentive with education content to encourage fitting and reduce returns.
Practical tips to improve your cohort outcomes
Common pitfalls and how to avoid them
Cohort analysis turns the welcome offer debate from an opinion to a measureable business decision. It forces you to account for hidden costs, to measure behaviour over time, and to design incentives that align with long-term value, not just short-term signups. If you want, I can outline a template you can plug your data into — or review a cohort report you already have and point out the leverage points. Drop a note via the contact page on Zynrewards Co or ping me on LinkedIn and we can dig into your numbers together.
