My LTV:CAC is bad. Which number do you actually fix?

1. Blog
2. Conversion & Pricing

By Saroj Jha · August 2, 2026 · 9 min read

Most answers to this question are a list — cut CAC, improve retention, raise prices, upsell more. All true, all useless, because none of them tells you which lever moves your ratio, or by how much.

Three fixes, one answer

1.7–1.8:1

where all three single-lever fixes land

Cut CAC 25%, cut churn a quarter, or raise ARPA 27% — none of them reaches 3:1.

The reason a list doesn't help is that these levers don't behave the same way. Two of them compound and one doesn't. Fixing the wrong one first is how a team spends a quarter on a 0.4x improvement.

Here's the actual arithmetic, run on a set of numbers, with the output shown.

The company

A B2B SaaS business at seed stage:

• $150 average revenue per account, per month
• 75% gross margin
• 2% monthly churn
• $4,200 to acquire a customer

Nothing pathological. These are numbers a real company has.

Running them through the unit-economics engine:

Illustrative example — worked outputs, not research findings

Illustrative example — worked outputs from the AAJ unit-economics engine, not research findings.

AAJ Unit Economics — subscription
------------------------------------------------------
Gross-margin LTV        $5,625   (2.0% monthly churn (~50 mo lifetime))
Monthly gross profit    $113/customer
CAC                     $4,200
LTV : CAC               1.3:1
CAC payback             37.3 mo

✗ LTV:CAC 1.3:1 is below the 3:1 floor — acquisition is inefficient;
  lift LTV (retention, margin, ARPA) or cut CAC before scaling.
✗ CAC payback 37.3 mo exceeds the ~12-month guideline — cash is tied up
  longer; watch burn.

1.3:1. Every customer acquired returns about a third more than they cost, over a fifty-month lifetime. That's not a business, it's a slow leak.

The 50-month lifetime is worth pausing on. It looks like good news — 2% monthly churn is respectable. But it means the LTV figure assumes you keep this customer for over four years, and you are spending the $4,200 today.

Three obvious fixes, none of which work

Take the three things you'd actually try, each at a level a team could plausibly deliver in a quarter.

Cut CAC by 25%. Better targeting, kill the worst-performing channel, tighten the funnel. $4,200 → $3,150.

CAC                     $3,150
LTV : CAC               1.8:1
CAC payback             28 mo

Cut churn by a quarter. Real onboarding work, a proper success motion. 2.0% → 1.5% monthly.

LTV : CAC               1.8:1
CAC payback             37.3 mo

Raise ARPA by 27%. Repackage, push the middle tier. $150 → $190.

LTV : CAC               1.7:1
CAC payback             29.5 mo

Three different quarters of work. Three different teams. 1.8, 1.8, 1.7. Not one of them reaches the floor.

This is the part the listicles omit. When a ratio is badly broken, no single lever fixes it, and a team that picks one and executes it well still fails — which feels like a competence problem and isn't.

How far would one lever have to move?

If you insisted on fixing this with one change, here's the required size:

A 55% CAC cut is not an optimisation, it's a different go-to-market. Raising prices 127% is a different product. And churn can't do it alone at all — you can drive it to 0.9% and still sit at 2.7:1, because the customer's monthly gross profit of $113 is simply too small relative to a $4,200 acquisition cost. Longer life doesn't fix a thin customer fast enough.

What actually works

The same three moves, applied together, each at the modest level a team can genuinely hit:

AAJ Unit Economics — subscription
------------------------------------------------------
Gross-margin LTV        $9,500   (1.5% monthly churn (~66.7 mo lifetime))
Monthly gross profit    $143/customer
CAC                     $3,150
LTV : CAC               3:1
CAC payback             22.1 mo

✓ LTV:CAC 3:1 is at or above the 3:1 floor.
✗ CAC payback 22.1 mo exceeds the ~12-month guideline — cash is tied up
  longer; watch burn.

CAC down 25%, churn down a quarter, ARPA up 27%. Each individually unremarkable. Together they take the ratio from 1.3 to 3.0 — a 2.3x improvement that no single lever could produce at any plausible level of effort.

That's the mechanism worth understanding: these levers multiply. LTV is monthly gross profit divided by churn, and the ratio divides that by CAC. Improving three inputs by roughly a quarter each doesn't add up to 75% better — it compounds to more than double.

The second verdict is the one that kills you

Note what stayed failed. Payback is still 22.1 months.

The ratio says each customer is worth three times what they cost. Payback says you wait nearly two years to get the money back. Both are true, and only one of them is about survival.

A seed-stage company with 22-month payback needs to fund almost two years of acquisition spend out of a balance sheet that's typically 18 months long. You can have a healthy ratio and run out of cash before it ever pays you.

This is why LTV:CAC on its own is a dangerous metric to manage to. It says nothing about when. A business at 3:1 with 6-month payback and one at 3:1 with 22-month payback are not the same business, and the second one dies first.

The fix for payback is different from the fix for the ratio. Payback responds to CAC and to monthly gross profit — not to churn at all, because churn changes how long the customer lasts, not how fast they pay you back. If payback is your problem, retention work is the wrong quarter's work.

What to do with your own numbers

1. Calculate both, not one. A ratio without a payback period is half a picture, and it's the reassuring half.
2. Test your levers before you pick one. Run your numbers, then run them again with one input changed. If a plausible move produces a small change, that lever isn't your problem.
3. Expect to need all three. A badly broken ratio is a compounding problem and needs a compounding fix.
4. Check whether payback and ratio disagree. They often point at different work. Payback usually wins, because it's the one denominated in runway.

The same approach — run an open engine, publish the output, argue with the number rather than the anecdote — is applied to AI visibility in Why doesn't ChatGPT recommend my company?

You can run the engine above on your own numbers:

npx skills add sarojkjha/aaj-marketing-skills --skill unit-economics
node .agents/skills/unit-economics/resources/unit-economics.js --demo

Then swap the demo values for yours:

node .agents/skills/unit-economics/resources/unit-economics.js \
  '{"model":"subscription","arpaMonthly":150,"grossMargin":75,"churnMonthly":2,"cac":4200}'

It's free and MIT licensed, and it works with ecommerce and services models too — the thresholds differ by model, which is another reason a generic "3:1 is healthy" rule is worth less than the arithmetic on your actual numbers. You can also run the same math in the browser with the Unit Economics Calculator. The same method applied to quarterly marketing reporting is in How do I know if marketing is working?