SaaS Unit Economics: The Cohort Decay Problem

## SaaS Unit Economics: The Cohort Decay Problem

When we audit a new SaaS client, the conversation often starts the same way.

"Our LTV-to-CAC ratio looks great," the founder says. "We're at 3.2x, which is above benchmarks."

Then we ask one question: "What's your LTV by cohort?"

Long pause. Spreadsheet fumbling. Usually: "We haven't really broken it down that way."

This is the cohort decay problem—and it's silently destroying unit economics at companies we work with every day.

Your blended LTV isn't real. It's an average of healthy early cohorts masking deteriorating recent ones. When we model forward, when we plan hiring, when we forecast Series B metrics—we're building on sand.

Let's fix that.

## What Cohort Decay Actually Is (And Why It Matters)

### The Basic Problem

Cohort decay is the pattern where customers acquired in recent months generate materially lower lifetime value than customers from earlier cohorts, even after controlling for product improvements and pricing changes.

It happens for reasons founders rarely track:

- **Product-market fit degradation**: Early customers had stronger product-market fit because you were solving their problem harder. Recent customers are buying into a more generalized product.
- **Competitive positioning shift**: Your first 50 customers were desperate for a solution. Your last 50 are comparing you against 10 alternatives.
- **Sales quality compression**: Early customers were founder-sold or hand-picked. Recent customers came through PPC or cold outreach where unit CAC is cheaper but unit quality is different.
- **Packaging dilution**: You've added lower-tier plans to expand TAM. These expand revenue but shrink per-customer LTV.
- **Market saturation in segment**: You're penetrating the easy part of your market first. Recent cohorts come from harder-to-retain segments.

In our experience, cohort decay is the single largest driver of the gap between founder LTV predictions and actual realized LTV. It's also invisible until you deliberately measure it.

### Why Your Blended Metrics Hide It

Let's use real numbers. Say you have:

- **Cohort 1 (6 months old)**: 200 customers, $45 blended monthly retention, $150/MRR avg = $9,000 LTV
- **Cohort 2 (4 months old)**: 350 customers, $42 blended monthly retention, $140/MRR avg = $7,980 LTV
- **Cohort 3 (2 months old)**: 600 customers, $38 blended monthly retention, $125/MRR avg = $5,750 LTV

Your blended LTV across all three cohorts?

**$6,847**

But here's what that average conceals: Cohort 3 isn't broken yet—it's just on a different trajectory. If it follows current decay patterns, month 6 LTV will be $4,200, not $6,847. Your forward models that assume blended LTV apply uniformly are off by 38%.

When we present this to founders, the reaction is consistent: "But we've improved the product. Why wouldn't newer cohorts be better?"

Great question. Let's talk about that.

## The Counterintuitive Part: Why Better Product Often Means Lower LTV

This trips up most founders.

You've improved your onboarding (good). You've added features (good). You've optimized conversion (good). So recent cohorts *should* have higher LTV.

Instead, they have lower LTV. Why?

**You've optimized for customer acquisition volume, not customer quality.**

When you were smaller, you had three ways to grow:

1. Founder sales (highest quality)
2. Word-of-mouth (high quality, limited volume)
3. Manual outreach (medium quality, medium volume)

As you scaled, you added:

4. Self-serve / PPC (lowest quality, unlimited volume)
5. Partnerships / affiliates (variable quality)
6. Sales hiring (dilutes founder quality control)

You didn't replace channels 1-3. You added to them. So your CAC decreased, your volume increased, and your blended metrics looked better. But you were drawing from an increasingly broad part of the market.

In SaaS terms: your early cohorts were the "must-have" segment. Your recent cohorts are the "nice-to-have" segment. Of course they have lower LTV. Of course they churn faster.

This isn't a product problem. It's a market segmentation problem. And it won't fix itself with better onboarding.

## How to Measure Cohort Decay Properly

### The Cohort Retention Matrix You Need

Stop looking at blended retention. Build a cohort retention table that shows:

- **Rows**: Acquisition month (Cohort 1, Cohort 2, etc.)
- **Columns**: Months post-acquisition (Month 1, Month 2, Month 3, etc.)
- **Cells**: % customers retained OR MRR retained

Example:

```
Cohort M1 M2 M3 M4 M5 M6
Jan 100% 87% 74% 62% 51% 41%
Feb 100% 85% 69% 55% 44% 35%
Mar 100% 82% 63% 48% 37% 28%
Apr 100% 78% 58% 42% 31% 23%
May 100% 74% 51% 36% 25% 17%
```

Notice the diagonal pattern? Month-2 retention for Jan cohort (87%) vs. May cohort (74%). This is decay.

### Calculate True LTV by Cohort

For each cohort, calculate:

**LTV = (Average MRR at acquisition) × (Gross Margin %) × (Sum of monthly retention rates)**

If Jan cohort started at $150 MRR with 80% gross margin and retained [100, 87, 74, 62, 51, 41] over 6 months:

LTV = $150 × 80% × (1.00 + 0.87 + 0.74 + 0.62 + 0.51 + 0.41) = $150 × 0.80 × 4.15 = $498

Do this for all cohorts. Plot them over time. The slope tells you everything.

### The Decay Rate Metric

Calculate month-over-month decay:

**Decay Rate = (Current Cohort LTV - Prior Cohort LTV) / Prior Cohort LTV**

If Jan cohort LTV is $6,200 and Feb cohort LTV is $5,750:

Decay Rate = ($5,750 - $6,200) / $6,200 = -7.3% per cohort

At -7.3% monthly decay, your LTV is halving every 10 months. That's a problem.

Investors will see it immediately. Model it forward, and you can see exactly when unit economics break.

## The CAC Problem Hiding in Cohort Decay

Cohort decay gets worse when combined with rising CAC.

We worked with a B2B SaaS company that had:

- **Jan cohort**: $3,200 CAC, $6,200 LTV, 1.94x ratio
- **May cohort**: $2,100 CAC, $4,100 LTV, 1.95x ratio

Their ratio stayed flat. Founder was happy. But the unit was dying:

- **Jan cohort payback period**: 11 months
- **May cohort payback period**: 18 months

Why? Because CAC is a point-in-time cost. It doesn't scale. But the cash impact of longer payback periods scales directly into your runway math.

We cover this deeper in our article on [CAC Attribution vs. Reality: Why Your Marketing Math Doesn't Match Cash Flow](/blog/cac-attribution-vs-reality-why-your-marketing-math-doesnt-match-cash-flow/), but the tldr is: cohort decay destroys payback period benchmarks you thought were solid.

## Why Gross Margin Recursion Amplifies Decay

Here's a mechanic most founders miss.

As cohorts decay and churn accelerates, your support costs per retained dollar increase. You're spreading fixed support costs across smaller customer bases.

This is the [gross margin recursion problem](/blog/saas-unit-economics-the-gross-margin-recursion-problem/) we've written about separately, but the interaction with cohort decay is critical:

- Early cohorts (higher LTV): Support cost per retained customer = $12
- Recent cohorts (lower LTV): Support cost per retained customer = $18

Your gross margin narrows. Your LTV calculations that use blended gross margin become increasingly wrong. You're feeding future cohorts into a degrading unit.

This is the decay spiral.

## The Fix: Segmentation by Quality

### Step 1: Identify Your Quality Cohorts

Don't optimize for acquisition volume. Segment by customer quality:

- **Tier 1**: Customers acquired through founder sales, warm intros, or high-fit segment (CAC $1,500-$3,500)
- **Tier 2**: Customers from optimized sales or high-quality partnerships (CAC $2,000-$4,000)
- **Tier 3**: Customers from self-serve, PPC, or broad channels (CAC $500-$2,000)

Measure retention and LTV by tier. You'll likely see:
- Tier 1: 85%+ month-2 retention, $8,000+ LTV
- Tier 2: 75%+ month-2 retention, $4,500-$6,000 LTV
- Tier 3: 60%+ month-2 retention, $2,000-$3,500 LTV

### Step 2: Cap Your Tier 3 Mix

Here's the hard choice: your Tier 3 customers might be profitable, but they're also limiting your LTV ceiling.

Instead of optimizing CAC down to $500, optimize Tier 1 + Tier 2 to 60-70% of new ARR. Let Tier 3 fill the remainder, not drive strategy.

Your blended LTV will drop. Your efficiency will improve.

### Step 3: Redesign Your Product Tiers Around Segments

Often, cohort decay accelerates because you've flattened your pricing. Earlier cohorts bought "Pro" ($199/mo). Recent cohorts buy "Starter" ($49/mo) because you broadened distribution.

Cut the Starter tier, or re-architect it:

- Make it genuinely different (lower limits, longer support latency)
- Price it to reflect support cost, not CAC math
- Use it as a trial tier, not a sustainable segment

This [payback period](/blog/saas-unit-economics-the-recursion-timing-problem-founders-ignore/) pressure is why SaaS companies eventually segment ruthlessly.

## Forecasting Forward: Using Cohort Decay to Model Reality

Once you understand your decay rate, you can project forward.

If your decay rate is -6% per cohort month:

- **Month 1 LTV**: $6,200 (baseline)
- **Month 2 LTV**: $5,828
- **Month 3 LTV**: $5,478
- **Month 4 LTV**: $5,149
- **Month 12 LTV**: $3,058

Now model against your CAC trend. If you're holding CAC flat while LTV decays, your payback period extends. Your [burn rate runway](/blog/the-burn-rate-runway-equation-what-your-financial-model-isnt-telling-you/) shrinks.

This is the math investors see. And it's why founders who don't track cohort decay can't explain why their financial model is drifting from reality.

## SaaS Unit Economics: Beyond the Blended Magic Number

The magic number (net new ARR divided by prior quarter sales spend) is useful. But it masks cohort decay.

A 0.95 magic number could mean:
- All cohorts declining uniformly (declining efficiency)
- Early cohorts strong, recent cohorts dying (unsustainable growth)
- Tier 1 declining, Tier 3 growing (shifting market mix)

They tell completely different stories for your future.

Investors dig into cohort decay because it's the early warning signal. Track it obsessively, and you'll stay ahead of the problem.

## The Bottom Line

Blended SaaS unit economics metrics—LTV, CAC ratio, magic number—are useful dashboards. But they're not true.

Cohort decay is the reason.

Your real LTV by cohort is probably 15-30% lower than your blended average. Your recent payback period is probably 6-12 months longer than you think. Your growth curve is probably less sustainable than it appears.

Start measuring cohort decay this week. Plot it. Project it forward. Share it with your board.

Then optimize around it—not the blended number.

In our work with Series A and Series B SaaS companies, the teams that move fastest are the ones who stop optimizing for "magic number" and start optimizing for "cohort quality." It feels slower at first. It compounds faster.

If you'd like a deeper look at your cohort decay patterns and what they mean for your financial model, [book a free financial audit with Inflection CFO](/). We specialize in the gap between what your metrics say and what your cash actually does—and cohort decay is usually hiding right there.