A Shopify cohort analysis groups customers by the month they were acquired, then tracks their behavior — repeat purchases, total revenue, churn — over time. It's the most useful retention tool most small Shopify merchants don't use, because the default analytics dashboards don't surface it. This article walks through how to compute cohorts manually (no data team required), the three patterns to look for, and why cohort-aware merchants out-decide cohort-blind ones over a 12-month horizon.
What a cohort analysis actually shows
A standard cohort table has:
- Rows: cohort (acquisition month — January, February, etc.)
- Columns: months since acquisition (M0, M1, M2, ..., M12)
- Cells: a metric for that cohort at that month — typically retention rate, cumulative revenue per customer, or order count.
A revenue-cumulative cohort table shows how much each acquisition cohort has spent in total by each month after acquisition. It's the cleanest view of LTV trends.
A retention cohort table shows what percent of each acquisition cohort placed an order in each subsequent month. Best for detecting churn patterns.
Most Shopify stores benefit most from the revenue-cumulative version because it answers "are my LTV economics improving or deteriorating over time?"
How to compute cohorts in Shopify (manually)
You don't need a data team. Three options:
Option 1: Shopify's "Customer cohort analysis" report (Plus)
Shopify Plus has a built-in cohort report under Analytics → Reports → Customer cohort analysis. Limited cuts but adequate for most cohort questions.
Option 2: Export to spreadsheet
For non-Plus merchants:
- Export all customers + their order history (Customers report → Export).
- In a sheet, group customers by month of first order.
- For each cohort, calculate cumulative revenue or repeat-purchase rate at month-1, month-2, etc.
The pivot table operation takes 1–2 hours the first time, ~15 minutes after that.
Option 3: Cohort-analysis tools
Lifetimely, Triple Whale, Polar Analytics — all show cohort tables natively. $29–$200/month. Worth it at $50K+/month or if you do cohort analysis monthly.
The three patterns to look for
1. Stable cohorts (good)
Each new cohort tracks roughly the same revenue trajectory as previous cohorts:
- January cohort: $42 cumulative revenue at month-1, $68 at month-3, $95 at month-6.
- February cohort: $44 at month-1, $70 at month-3, $96 at month-6.
- March cohort: $43 at month-1, $69 at month-3.
Each new cohort behaves like the last. The store's underlying economics are consistent.
2. Improving cohorts (excellent)
Newer cohorts track higher revenue trajectories:
- Q1 cohorts: $42 at month-1, $95 at month-6.
- Q2 cohorts: $48 at month-1, $115 at month-6.
- Q3 cohorts: $54 at month-1.
Could be: better acquisition (you're acquiring better customers), better product (existing customers are buying more), better email/retention.
3. Declining cohorts (concerning)
Newer cohorts spend less than older ones:
- Q1 cohorts: $48 at month-1, $115 at month-6.
- Q2 cohorts: $44 at month-1, $98 at month-6.
- Q3 cohorts: $40 at month-1, $85 at month-3.
Possible causes:
- Acquisition channel quality dropped (paid traffic now going to less-qualified buyers).
- Product quality dropped (returns up, repeat down).
- Competitive pressure (alternatives are taking market share).
- Discount-heavy acquisition (Q3 cohorts came in via heavy discounts and stayed bargain-conscious).
The diagnostic is to find which channel's cohorts declined. Often the issue is isolated to one source.
Reading a cohort table
A real example. Consider this cumulative revenue per customer:
| Cohort | M0 | M1 | M3 | M6 | M9 | M12 |
|---|---|---|---|---|---|---|
| 2025-04 | $45 | $52 | $74 | $115 | $138 | $156 |
| 2025-07 | $44 | $50 | $71 | $108 | $128 | — |
| 2025-10 | $46 | $53 | $76 | $118 | — | — |
| 2026-01 | $48 | $55 | $79 | — | — | — |
| 2026-04 | $50 | — | — | — | — | — |
What we see:
- Slight dip in 2025-07 cohort (could be seasonal, a one-off campaign, or supplier issue that month).
- Recovery in 2025-10 onward.
- 2026-01 cohort is performing best yet — indicates whatever changed Q4 2025 is paying off.
The reading: portfolio is slowly improving. Quarterly review confirms the trend. New cohorts are coming in better.
If 2026-01 had been worse than 2025-04 — same store, but newer cohorts under-performing earlier ones — that's a flag for product or channel decay.
What to do with cohort findings
Three patterns of action:
Improving cohorts → invest more in acquisition
If newer cohorts are worth more, your CAC absorption capacity has grown. You can spend more per acquired customer profitably. Time to scale ad spend or test new channels.
Stable cohorts → focus on conversion + retention
Stable cohorts mean the engine is calibrated. Lift comes from improving the engine: PDP CRO, email flows, win-back. See the PDP CRO guide, win-back guide, email ROI guide.
Declining cohorts → diagnose first, then act
Don't double down on acquisition while cohorts are declining. You'll just acquire more customers worth less. Diagnose the cause:
- Check return rates by cohort (rising returns = product issue).
- Check repeat-purchase rate by acquisition channel (if Meta cohorts are declining but organic stable, channel-specific issue).
- Check NPS or review sentiment (declining customer satisfaction).
Fix the underlying problem, then resume acquisition spending.
Channel-specific cohort analysis
A more powerful version: separate cohorts by acquisition channel.
Compute LTV per cohort × channel:
- Meta-acquired customers, January cohort
- Google-acquired customers, January cohort
- Organic-acquired customers, January cohort
- TikTok-acquired customers, January cohort
The differences are usually large. Typical pattern:
- Organic customers have ~1.5× higher LTV than paid (they self-selected, lower price sensitivity).
- Meta customers in the middle.
- TikTok customers often lowest LTV initially (impulse-buy heavy) but improving with retargeting.
This per-channel view often reveals that "Meta is unprofitable" is wrong — Meta is unprofitable for first-time customers in month 1, but fine over 6 months. CAC payback math should use channel-cohort LTV, not blended LTV.
Common cohort analysis mistakes
Reading early data as signal
A 1-month-old cohort tells you almost nothing about LTV. Don't react to month-1 metrics on a 12-month horizon.
Mixing organic and paid
A blended cohort hides channel-specific issues. Always cut by acquisition channel.
Ignoring seasonality
December cohorts (gift buyers) behave differently than May cohorts (regular self-purchasers). Compare like-to-like seasons when available.
Defining cohorts by purchase month vs marketing-touch month
For paid acquisition, the right cohort is "first paid touch" not "first purchase." Different signal. Most stores use first purchase because it's available; the more rigorous approach uses first-touch attribution.
Updating analyses too rarely
Quarterly cohort reviews are the floor. Monthly is better at scale.
Frequently asked questions
How is cohort LTV different from regular LTV?
Regular LTV averages across all customers — old and new mixed. Cohort LTV isolates each acquisition month, so you can see if economics are improving or deteriorating over time. Cohort is more useful for trend detection; regular is fine for a snapshot.
How often should I run cohort analysis?
Quarterly minimum. Monthly if you're at scale or making active changes (channel mix, pricing, product). Less often than quarterly, you miss declining-cohort issues until they've compounded for months.
What if I don't have 12 months of data?
Use whatever horizon you have. 3-month cohort analysis on a 6-month-old store is still useful — you compare month-1 trajectories across cohorts. Wait to compute month-12 LTV until you have 12 months of customer history.
Should I segment cohorts by AOV bracket too?
Yes, eventually. A high-AOV cohort behaves differently than low-AOV. Once you have enough data (12+ months, 1000+ customers), segment by both acquisition channel and initial AOV bracket.
Does DropifyXL show cohort data?
DropifyXL surfaces operational signals (which customers to win back, which products to restock) — not cohort dashboards. For cohort analysis, use Shopify Plus's native report or a dedicated tool like Lifetimely. They complement.
Key takeaways
- Cohort analysis groups customers by acquisition month and tracks their behavior over time. The most useful retention tool small Shopify merchants underuse.
- Three patterns: stable (consistent economics), improving (newer cohorts spend more), declining (concerning).
- Always cut cohorts by acquisition channel — blended numbers hide channel-specific issues.
- Don't react to month-1 data; cohort signal lives at month-3 and beyond.
- Use Shopify Plus's report, a spreadsheet export, or a dedicated tool like Lifetimely.
Cohort analysis is the difference between "we have customers" and "we know how customers behave over time." The latter unlocks decisions the former can't make.
