Cohort Analysis: Why your averages are lying

IO Advisory
Dec 17, 2025
10 min read

Updated: Jan 29

TLDR

Your supplement brand shows €127 average order value and 2.8 orders per customer. Looks healthy. Until cohort analysis reveals your Meta ads cohort from Q1 2025 has 1.4 orders per customer with 15% retention, while your organic cohort has 4.2 orders with 48% retention. You're scaling the channel that destroys value and starving the one that creates it. Averages hide which customers actually drive profit and which burn cash. Cohort analysis shows you exactly where to spend, what to charge, and which products to launch.

The Calculator: Model Your Cohort analysis

Use the calculator below to model your own cohort analysis. Input your acquisition costs, retention rates, purchase frequency, and contribution margins. See how different assumptions change payback period, lifetime value, and LTV/CAC ratios over 12 months.

The calculator shows you three critical outputs:

Payback period: How many months until cumulative revenue covers CAC? If it's longer than 12 months, you're burning cash on customer acquisition for at least a year before breaking even.
12-month LTV vs lifetime LTV: How much value does a customer generate in year one vs their entire lifetime? If 12-month LTV is only 40% of lifetime LTV, you're dependent on long-term retention to justify acquisition costs.
LTV/CAC ratio by month: How does the ratio evolve over time? If you're at 1.2x after 12 months but 3.8x after 24 months, you need cheap capital or high margins to survive the payback period.

Run your numbers. See if your unit economics actually work or if averages have been lying to you.

The €2.4M Question Nobody Asks

A supplement brand spent €2.4M on customer acquisition in 2025. Revenue grew 140%. The board celebrated. Then someone asked: "Which customers from which month are actually profitable?"

Silence.

They knew average order value (€127). They knew average customer lifetime value (€340). They knew overall CAC (€86). All averaged across 18 months of customers acquired through six different channels at different price points with different product mixes.

The cohort analysis told a different story. January 2025 customers acquired through Meta ads: €142 CAC, €198 lifetime revenue, 1.4 orders, 12-month retention 15%. May 2025 customers from organic search: €0 CAC, €487 lifetime revenue, 4.1 orders, 12-month retention 51%. They had spent €840K scaling Meta ads in H2 2025—acquiring customers who would never be profitable.

This is not about Meta ads being bad. This is about averages lying. When you blend profitable cohorts with unprofitable ones, stable metrics with deteriorating ones, you make decisions in the dark. Cohort analysis turns the lights on.

What Cohort Analysis Actually Reveals

Most DTC brands track customers as a single mass. Revenue per customer. Retention rate. Repeat purchase rate. All blended. This masks three critical patterns:

Pattern 1: Cohort deterioration. Your December 2024 cohort had 42% 6-month retention. Your June 2025 cohort has 28% 6-month retention. Your blended retention rate is stable at 35% because older, better cohorts offset newer, worse ones. You think you're fine. You're not. Every new cohort performs worse than the last.

Pattern 2: Channel economics divergence. Your Meta ads customers have €156 CAC and €203 lifetime value. Your Google organic customers have €8 CAC (attribution to content/SEO effort) and €441 lifetime value. Your blended CAC of €94 and blended LTV of €287 suggest healthy 3.1x returns. But 60% of your spend goes to Meta ads with 1.3x returns while organic, with 55x returns, gets no investment.

Pattern 3: Product-cohort mismatch. Customers who buy your protein powder first have 3.2x higher lifetime value than customers who buy your multivitamin first. Your product launch decisions don't account for this. You just launched a new multivitamin variant because multivitamins are 40% of revenue. You're scaling the entry product that predicts low lifetime value.

Cohort analysis disaggregates averaged metrics into time-based segments. Customers acquired in January 2025 are one cohort. Customers acquired in February 2025 are another. You track each cohort's behavior over time. Revenue. Orders. Retention. AOV. By month. This reveals which customer segments create value and which destroy it.

The Supplement Brand That Scaled Backwards

Consider a supplement brand with these 2025 numbers:

Total customers acquired: 28,000
Total revenue: €3.56M
Average order value: €127
Average orders per customer: 2.8
Average customer lifetime value: €356
Average CAC: €86
LTV/CAC ratio: 4.1x

Board conclusion: healthy unit economics, scale acquisition.

The cohort view told a different story. They segmented customers by acquisition month and channel:

Q1 2025 Meta Ads Cohort (3,200 customers, €455K spend):

CAC: €142
Month 1 revenue per customer: €134
Month 6 revenue per customer: €198
6-month retention: 18%
Average orders through month 6: 1.5
Projected 12-month LTV: €203
LTV/CAC: 1.43x

Q1 2025 Organic Cohort (1,100 customers, ~€0 direct CAC):

CAC: €0 (attributed acquisition cost from content)
Month 1 revenue per customer: €118
Month 6 revenue per customer: €312
6-month retention: 47%
Average orders through month 6: 2.6
Projected 12-month LTV: €487
LTV/CAC: infinite (or 60x if you attribute €8 content cost)

Q3 2025 Meta Ads Cohort (4,800 customers, €720K spend):

CAC: €150
Month 1 revenue per customer: €129
Month 3 revenue per customer: €142 (only 3 months of data)
3-month retention: 14%
Average orders through month 3: 1.1
Projected 12-month LTV: €176
LTV/CAC: 1.17x

The pattern was clear. Meta ads cohorts were deteriorating (CAC rising, LTV falling). Organic cohorts remained strong. Blended metrics masked this because strong 2024 cohorts and early 2025 organic cohorts offset poor paid cohorts. The brand was spending 75% of its budget scaling unprofitable customers.

The Four Decisions Cohorts Actually Drive

Cohort analysis isn't academic analysis. It guides four operational decisions that determine whether you build a profitable business or burn cash at scale.

Decision 1: Where to Spend

Every acquisition channel produces customers with different economics. Meta ads vs Google ads vs organic search vs referral vs retail partnerships. Cohort analysis shows you which channels produce customers who actually pay back their acquisition cost—and how long that takes.

The supplement brand tracked CAC payback by channel cohort:

Organic search: Payback in month 2 (second order), 12-month LTV €487
Referral program: Payback in month 3, 12-month LTV €394, CAC €24
Google Ads (branded): Payback in month 3, 12-month LTV €356, CAC €34
Google Ads (non-branded): Payback in month 5, 12-month LTV €287, CAC €78
Meta Ads (prospecting): Payback in month 11, 12-month LTV €203, CAC €142
Meta Ads (retargeting): Payback in month 4, 12-month LTV €298, CAC €67

The decision became obvious. Stop scaling Meta prospecting. Double down on organic content, referral program, and Google branded. Maintain Meta retargeting. This isn't "Meta ads don't work." This is "Meta prospecting ads for this brand at this CAC acquire customers with these retention characteristics who don't pay back fast enough."

Without cohort analysis, you see blended CAC of €86 and blended LTV of €356 and think everything works. With cohort analysis, you see that two channels subsidize four others.

Decision 2: What to Charge

Pricing decisions typically use cost-plus logic or competitive benchmarking. Cohort analysis reveals what customers actually pay over time—and whether higher or lower entry prices produce better lifetime value.

The supplement brand tested two pricing strategies:

Strategy A: €89 entry offer (40% discount), regular price €149

Month 1 AOV: €89
Conversion rate: 3.2%
Month 6 retention: 22%
12-month LTV: €234

Strategy B: €119 entry offer (20% discount), regular price €149

Month 1 AOV: €119
Conversion rate: 2.1%
Month 6 retention: 41%
12-month LTV: €412

Lower entry price acquired more customers (higher conversion) but worse customers (lower retention, lower LTV). Higher entry price acquired fewer but better customers. At €86 blended CAC, Strategy A produced 2.7x LTV/CAC. Strategy B produced 4.8x LTV/CAC.

This pattern repeated across product categories. Deep discounts on first purchase attracted deal-seekers with low repurchase intent. Modest discounts attracted customers who valued the product. The brand shifted from 40% off to 20% off entry offers. Acquisition volume dropped 28%. Profit increased 47%.

Decision 3: Which Products to Launch

Most brands decide what to launch based on revenue contribution or margin. Protein powders are 35% of revenue, launch more flavors. Multivitamins have 68% margin, expand the line. Cohort analysis reveals which entry products predict high lifetime value—which tells you what to launch to acquire valuable customers.

The supplement brand tracked lifetime value by first purchase:

First purchase: Omega-3: 12-month LTV €509, 46% retention
First purchase: Protein powder: 12-month LTV €447, 41% retention
First purchase: Greens powder: 12-month LTV €398, 38% retention
First purchase: Multivitamin: 12-month LTV €198, 19% retention
First purchase: Single vitamin (D3, Magnesium): 12-month LTV €167, 15% retention

Customers who bought Omega-3 first had 3x higher LTV than customers who bought multivitamins first. Why? Omega-3 buyers were health-serious customers building a supplement stack. Multivitamin buyers were casual one-product users. Single vitamin buyers were often one-time deficiency addressers.

The brand had planned to launch four new multivitamin SKUs (women's, men's, 50+, immune). Cohort analysis said: don't scale the entry product that predicts low LTV. Instead, they launched Omega-3 variations (high-dose, triglyceride form, algae-based) and protein powder flavors. They used multivitamins as loss-leaders for retargeting but stopped feature product launches there.

This isn't "multivitamins are bad." This is "multivitamin-first customers have different behavior than Omega-3-first customers, and we should design acquisition and product strategy accordingly."

Decision 4: When to Change Course

Averaged metrics move slowly. By the time your blended LTV drops from €356 to €298, you've spent six months acquiring unprofitable customers. Cohort analysis shows deterioration in real-time—new cohorts performing worse than old cohorts—which tells you when to change course immediately.

The supplement brand saw this in Q3 2025:

April 2025 cohort: Month 1 retention 34%, Month 3 retention 26%
May 2025 cohort: Month 1 retention 31%, Month 3 retention 23%
June 2025 cohort: Month 1 retention 28%, Month 3 retention 19%
July 2025 cohort: Month 1 retention 25%, Month 3 retention 16%

Each new cohort performed worse than the previous one. This wasn't seasonal. This was systematic deterioration. They investigated and found three causes: (1) Meta ads scaled to lower-quality audiences, (2) creative fatigue reduced message resonance, (3) longer shipping times reduced satisfaction.

They caught this in August—after three months of declining cohorts but before it showed in blended metrics. They paused Meta scaling, refreshed creative, and fixed fulfillment. September and October cohorts recovered to April levels. Without cohort tracking, they wouldn't have seen the problem until Q4 when blended LTV finally dropped—after spending another €600K on deteriorating cohorts.

How to Actually Do This

Cohort analysis sounds complex. It's not. You need three things: a way to tag customers by acquisition date, a way to track their revenue over time, and a way to segment by acquisition channel or product.

Step 1: Tag every customer with acquisition date and source. Your e-commerce platform does this automatically. Shopify, WooCommerce, whatever you use. Every customer has a "created at" timestamp and a UTM source if they came from paid channels. That's your cohort identifier.

Step 2: Pull monthly cohort data. Export customer data with acquisition month, revenue by month, orders by month. If you have 10,000 customers acquired over 12 months, you have 12 cohorts. Track each cohort's cumulative revenue and orders over time.

Step 3: Calculate cohort metrics. For each cohort, calculate:

Month 0 (acquisition month): customers acquired, revenue, AOV
Month 1: retained customers, revenue from that cohort, cumulative revenue
Month 2: retained customers, revenue from that cohort, cumulative revenue
Month N: retained customers, revenue from that cohort, cumulative revenue

Then calculate retention rate (Month N customers / Month 0 customers), cumulative LTV (sum of all revenue through Month N divided by Month 0 customers), and payback period (month when cumulative LTV exceeds CAC).

Step 4: Segment by channel and product. Repeat the above for customers acquired by channel (Meta ads, Google ads, organic, referral) and by first product purchased. Now you see which channels and products produce valuable customers.

You don't need fancy software. A spreadsheet works. If you have more than 50,000 customers, use your analytics platform (Google Analytics 4 has cohort reports built in, so does Shopify Analytics). The point is to stop looking at averaged metrics and start looking at time-based segments.

What to Track

Once you build cohort tracking, four metrics tell you everything:

1. Cohort LTV by channel at months 3, 6, and 12. This shows you which channels produce customers who actually pay back and how long that takes. If your Meta ads cohort has €156 LTV at month 12 and €142 CAC, you know payback happens between months 11-12. If your organic cohort has €312 LTV at month 6, you know they pay back much faster.

2. Cohort retention at months 3, 6, and 12. Retention is the leading indicator of LTV. If your new cohorts show 18% month-6 retention vs 35% for older cohorts, you know LTV will fall before it shows in revenue. This gives you time to fix the problem.

3. Month-over-month cohort performance. Is each new monthly cohort performing better or worse than the previous one? If April, May, June cohorts show declining month-3 retention, you have systematic deterioration. If they're stable or improving, you're on track.

4. Entry product to LTV correlation. Which products, when purchased first, predict high lifetime value? These are the products you should feature in acquisition campaigns and expand with new SKUs. Which products predict low LTV? These might still be profitable but shouldn't drive acquisition strategy.

These four metrics replace 30+ vanity metrics in your dashboard. They tell you where to spend, what to charge, what to launch, and when to change course.

Red Flags Hidden in Aggregated Data

Cohort analysis exposes problems before they become visible in standard metrics:

Red Flag 1: Improving Gross Margin with Declining Contribution Profit

Gross margin increased from 64% to 71% after a formulation change that reduced COGS (positive signal in the aggregated P&L). Cohort analysis showed Meta-acquired customers’ average orders fell from 1.5 to 1.1 and projected 12-month LTV declined from €203 to €176. The margin gain was cosmetic: lower costs degraded product experience, reduced repeat purchases, and destroyed total customer profitability.

Red Flag 2: Stable AOV with Collapsing Purchase Frequency

Average order value was €127 across the business (stable in aggregated reporting). Cohort analysis revealed that Meta ads cohorts averaged 1.4–1.5 orders per customer, while organic cohorts averaged 4.1–4.2 orders. Stable AOV masked the real problem: customers were not coming back. Total customer value differed by more than 2x, with no warning signal in AOV trends.

Red Flag 3: Growth Masking Cohort Decay

Total customers acquired grew sharply in 2025 and revenue increased 140% (positive in aggregated metrics). Cohort analysis showed systematic deterioration: Meta ads cohorts declined from 18% to 14% 3-month retention and projected LTV fell from €203 to €176 within two quarters. Growth in customer count obscured that each new paid cohort was worth less than the last. The business was scaling acquisition while destroying marginal value.

Disclaimer

All data and calculations in this article are simplified for illustration purposes. Actual results depend on each company’s product mix, margins, service levels, and supply chain structure.

🛠 Curious how to apply this to your business? We guide you step by step - our services. Book a chat here.

🌐Fuel your strategy addiction. Follow us on LinkedIn for more content.