Your blended ROAS says 3.2x. Your dashboard is green. Your ad manager is confident. But somewhere in your business, an entire campaign is losing money on every customer it acquires—and blended metrics are hiding the evidence.
This is the single most dangerous trap in DTC marketing analytics. Blended numbers average your best campaigns with your worst, your returning customers with your new ones, your profitable channels with your money pits. The result looks healthy even when individual parts of the business are bleeding cash. Cohort analysis is how you see the truth.
What blended metrics actually show you
When you look at blended ROAS, you're dividing total store revenue by total ad spend. That sounds reasonable. It's not. Here's why:
- Returning customers inflate the number. Revenue from repeat buyers (who cost nothing to reacquire organically) gets attributed to your total revenue pool, making ad spend look more productive than it is.
- Organic traffic gets mixed in. SEO, word-of-mouth, and direct traffic all contribute revenue that has nothing to do with your paid campaigns.
- Good campaigns subsidize bad ones. A Meta campaign with a 5x ROAS and a TikTok campaign with a 1.2x ROAS might average out to a comfortable 3.2x. But the TikTok campaign is destroying value on every dollar spent.
Blended ROAS is not a performance metric. It's a comfort metric. It tells you what you want to hear, not what you need to know.
How cohort analysis works
Cohort analysis groups customers by when and how they were acquired, then tracks their behavior over time. Instead of asking "how did the store do this month?" you ask "how did the customers we acquired in March from Meta perform over 30, 60, and 90 days?"
Here's the basic approach:
- Define your cohort. Group customers by acquisition date (week or month) and by source (Meta, Google, TikTok, organic). Each group is a cohort.
- Track cohort revenue over time. For each cohort, measure the cumulative revenue they generate at day 7, day 14, day 30, day 60, and day 90.
- Compare cohort revenue to cohort cost. How much did you spend to acquire this specific group? When does their cumulative revenue (at margin) exceed that cost?
- Identify divergence. Some cohorts will pay back in days. Others will take months. Some may never pay back. This is the insight blended metrics hide.
The math: blended vs. cohorted ROAS
Let's make this concrete. Imagine a DTC brand running two campaigns in March:
Total Ad Spend: $15,000
Blended ROAS: $48,000 ÷ $15,000 = 3.2x
Now let's break it down by cohort:
Revenue from these 400 customers: $36,000
Cohorted ROAS: $36,000 ÷ $10,000 = 3.6x
Revenue from these 100 customers: $7,200
Cohorted ROAS: $7,200 ÷ $5,000 = 1.44x
The remaining $4,800 in revenue that made the blended number look good? That's from returning customers and organic traffic—revenue that had nothing to do with the $15,000 in ad spend. Campaign B is losing $15.44 per customer, and blended ROAS completely hid it.
A real-world example: the hidden money pit
We see this pattern constantly. Here's a hypothetical (but representative) DTC wellness brand:
| Cohort | Ad Spend | New Customers | CAC | 30-Day Revenue | Cohorted ROAS |
|---|---|---|---|---|---|
| Meta — Lookalike | $6,000 | 280 | $21.43 | $22,400 | 3.73x |
| Meta — Broad | $4,000 | 150 | $26.67 | $10,500 | 2.63x |
| TikTok — Influencer | $3,000 | 55 | $54.55 | $3,300 | 1.10x |
| Google — Brand | $2,000 | 120 | $16.67 | $9,600 | 4.80x |
| Blended | $15,000 | 605 | $24.79 | $45,800 | 3.05x |
Blended ROAS: 3.05x. Blended CAC: $24.79. Everything looks fine. But TikTok influencer is burning cash at a 1.1x ROAS with a $54 CAC. At a 48% contribution margin, each TikTok customer generates $38.40 in gross profit on the first 30-day window against $54.55 in acquisition cost. That's a $16.15 loss per customer, and the brand acquired 55 of them. That's $888 in losses hidden inside "healthy" blended numbers.
Kill the TikTok campaign, reallocate $3,000 to Meta Lookalike, and the same $15,000 in spend generates significantly more profit. You can't make that decision with blended data.
How CohortCredit uses cohorted repayment
This is exactly why CohortCredit's financing model is built on cohort analysis rather than blended revenue sharing. Traditional revenue-based financing takes a percentage of your total revenue—including revenue from customers you acquired months ago who have nothing to do with the funded spend.
CohortCredit's approach is different:
- We fund a specific campaign. $1,000 goes toward a defined marketing spend.
- The customers acquired form a cohort. We track that specific group.
- Repayment comes from cohort revenue. As those customers purchase, a portion flows back to cover the $1,100 repayment.
- Everything above $1,100 is yours. No claim on existing customer revenue, no percentage of total sales.
This alignment means we only get repaid if the funded spend actually works. If your payback period math is strong and your campaigns generate profitable cohorts, the advance pays for itself. That's a fundamentally different relationship than a lender skimming a percentage off your top line.
See what your cohort economics look like
Plug your channel-level CAC and margins into our calculator. → Try the CohortCredit Calculator
How to start running cohort analysis
You don't need expensive analytics software. You need discipline. Here's the minimum viable setup:
- Tag your acquisition source. Use UTM parameters religiously. Every ad, every campaign, every channel should have a unique UTM that you can trace back to a customer's first order.
- Export by cohort weekly. Pull a list of customers acquired in each week, segmented by source. Track their cumulative revenue at day 7, 14, 30, 60.
- Calculate cohorted ROAS and CAC. For each cohort, divide their revenue by the spend that acquired them. Compare this to your blended number. The gap is your blind spot.
- Kill losers fast. If a cohort's 30-day ROAS is below your breakeven threshold (typically 2x for a 50% margin business), that campaign needs to be paused or restructured.
Shopify's built-in analytics has basic cohort views, but they're limited. For serious cohort analysis, you'll want to export data to a spreadsheet or use a tool like Lifetimely, Triple Whale, or Polar Analytics.
The connection to growth capital
Here's the payoff: when you know your cohort economics, you can confidently deploy growth capital because you know exactly which campaigns produce fast-payback, profitable cohorts. You stop guessing and start compounding.
A brand with cohorted data can say "our Meta Lookalike campaigns have a 12-day payback period and a 3.6x cohorted ROAS. Fund those." A brand with only blended data says "our ROAS is 3.2x"—and has no idea if the next $1,000 in spend will generate a profitable cohort or a money-losing one.
For more on how payback period drives funding decisions, read our deep dive on the payback period math most DTC brands get wrong. And if you want to understand how revenue-based financing leverages cohorted economics, our guide to non-dilutive funding options compares the major alternatives.
The brands that win are the ones that see clearly. Cohort analysis is how you see.