In this article
Cohort analysis groups patients by the month of their first visit, then tracks what share of each group comes back and how much revenue it brings as time passes. It shows whether a clinic or hospital is keeping patients better or worse over time, which monthly patient totals cannot.
Patient numbers rising every month looks like good news. But if the rise comes from buying more new patients while existing ones do not return, marketing costs climb year after year. Cohort analysis separates the two. This guide explains how to read a cohort table and which numbers matter, with a tool to try on your own data straight away.
Analyse your patient cohorts
Try the sample data to see the output, or open a CSV exported from your clinic or hospital system. The sample is made-up data for one clinic over 12 months.
Patient cohort analysis tool
Paste visit data or open a CSV with 3 columns: patient ID, visit date and amount. The tool groups patients by the month of their first visit and shows what share came back in each following month. Data is processed on your device and never uploaded.
Or paste your own data
One line per visit, separated by commas or tabs, e.g. HN0001,2025-01-15,3500 Dates can be year-month-day or day/month/year (Thai or Western years). Always use patient IDs, never names.
How to read a cohort table
A cohort table has one row per group of patients, such as those first seen in January, and one column per month since that first visit.
| Read it | Look at | Questions it answers |
|---|---|---|
| Along rows (left to right) | How fast one group stops coming back | What share of January’s new patients still return in month 6? |
| Down columns (top to bottom) | Whether newer groups do better or worse at the same month | Since we started reminders, do newer groups return more in month 1? |
| Along diagonals | Whether one calendar month affected every group | When a regular doctor left, did all groups drop at once? |
Empty cells have not been reached yet; a group that started three months ago has no month 6.
The numbers to watch
| Number | Meaning | Informs |
|---|---|---|
| Month 1 return rate | Patients back in the month after their first visit | Follow-up and aftercare |
| Month 3 and 6 return rates | Whether patients stay attached to you | Recall programmes and service quality |
| 12-month revenue per patient | What one new patient brings in their first year | The ceiling on marketing cost per new patient |
| Change between groups | Whether the latest groups beat earlier ones | The effect of changes such as new channels or new staff |
What counts as a good return rate depends on the service. Dental and aesthetic clinics have natural return cycles; one-off procedures such as LASIK naturally have low return rates. Compare yourself across groups and by service, rather than against other providers’ figures.
Putting results to work
- Set your marketing spend ceiling from 12-month revenue per new patient times gross margin and the share you will spend on acquisition; see the hospital marketing plan
- Fix where patients drop off: low month 1 points to follow-up after the first visit; low months 3–6 point to recall with a medical reason, such as repeat checks, dental cleaning or the next vaccine dose
- Compare acquisition channels by splitting groups by the channel of their first visit; a cheap channel whose patients never return may cost more than it looks
- Measure the effect of changes by checking whether groups starting after a change do better at the same month
Preparing data, and the Excel template
Export visit data from your clinic or hospital system with three columns: patient ID (such as a hospital number), visit date and amount, one line per visit. Use IDs instead of names and leave out treatment details, so the file holds no personal or health data.
If you prefer Excel, this template builds both cohort tables, return rate and cumulative revenue per patient, from raw visits using formulas. It is ready for 3,000 rows and can be extended.
Patient cohort analysis template
Frequently asked questions
What is cohort analysis?
Cohort analysis groups customers or patients by when they started, such as by month of first visit, then tracks each group’s behaviour over time, such as the share who return and cumulative revenue per person. It reveals trends that monthly totals hide.
What data do I need for patient cohort analysis?
Three columns: patient ID, visit date and amount, one line per visit. Use IDs instead of names and leave out treatment details.
What is a good patient return rate?
It depends on the service. Services with natural cycles, such as dental and aesthetics, should return more than one-off procedures such as LASIK. Compare across your own groups and by service rather than with other providers.
How does cohort analysis help set the marketing budget?
It gives 12-month revenue per new patient. Multiplied by gross margin and the share you will spend on acquisition, that is the most you can pay for a new patient, which sets budgets and shows which channels pay.
Is it safe to use the tool on this page with real patient data?
The tool processes data in your browser and sends nothing to a server. Still use patient IDs instead of names and only the data you need, in line with data protection principles.
How often should we run cohort analysis?
Monthly or quarterly, with at least 12 months of history, so you see full-year groups and can compare newer groups with older ones.
