PLG Startups Are Measuring CAC Payback Wrong (Here's the Number That Actually Matters)

Why Sales-Led CAC Payback Math Breaks for PLG

The standard CAC payback formula is: take your fully-loaded cost to acquire a customer, divide it by the gross margin you collect from that customer each month, and you get the number of months before you break even. Clean, fast, widely used.

It is also wrong for product-led growth, and not just a little wrong.

In a sales-led motion, virtually every dollar you spend to acquire a customer is conversion spend: sales rep salaries, commissions, demos, outbound sequences, paid ads targeting decision-makers. Almost nothing is spent on users who never buy. The math maps neatly onto the formula because the formula was built for exactly that structure.

PLG has a fundamentally different cost structure. You spend money on two populations: the people who convert to paid, and the much larger group who use your product for free and never hand you a dollar. Those two populations share infrastructure costs, support costs, and often some marketing costs. If you blend them into a single CAC figure, you are systematically overstating what it actually costs you to acquire a paying customer.

The practical result: founders running PLG see a CAC payback number that looks worse than their sales-led competitors and panic, or they see a number that looks suspiciously clean and miss that a cost bucket is hiding inside their cost of goods.

The unit that matters in PLG is not the visitor, not the free signup, and not the MQL. It is the product-qualified lead, the user who has hit the activation threshold your product team defined (completed a key workflow, invited a second user, exported a result, whatever your PQL definition is). Everything before that threshold is a product cost. Everything after it is a sales and marketing cost. The payback formula should only touch the second bucket.

The Correct PLG CAC Payback Formula

Calculating this correctly takes three steps.

Step 1: Separate your free-user costs from your conversion costs.

Pull all costs that exist solely because you have free users: the marginal infrastructure cost per free account, the support tickets from free users, the in-app onboarding tooling (Appcues, Intercom sequences, whatever you're running) firing for users who will never convert. These costs belong on your product P&L, not in your CAC numerator.

What stays in the CAC numerator: sales assist touches on PQLs (including the prorated salary cost of whoever does those touches), any conversion-specific paid campaigns (retargeting against PQL cohorts, for example), and the CRM tooling used only for tracking conversion. That's it.

Step 2: Use net new MRR from converted accounts in the denominator, not total new MRR.

If your sales-assist team closes 40 accounts in a month and 15 of those would have converted self-serve without any touch, attribute those 15 to self-serve CAC (which is very low) and only the 25 to your sales-assist CAC calculation. Blending them rewards your conversion team for work the product did.

Step 3: Divide.

PLG Sales-Assist CAC Payback = (Conversion-Touch Costs) / (Net New MRR from Sales-Assist Converted Accounts × Gross Margin %)

Your self-serve payback is calculated separately:

Self-Serve CAC Payback = (Self-Serve Acquisition Costs) / (Net New MRR from Self-Serve Converts × Gross Margin %)

Run both numbers. They will tell you different things. A 6-month self-serve payback with a 22-month sales-assist payback tells you your sales overlay is too expensive relative to the ACV it's closing. A 4-month self-serve payback across tiny accounts tells you you have a pricing problem, not a growth problem.

For a deeper look at how sales model structure affects payback math more broadly, this breakdown of CAC payback period by sales model covers the mechanics across different GTM structures.

The 18-Month Benchmark and What It Actually Means

The "12-month payback = healthy" rule circulates constantly. It comes from sales-led SaaS benchmarks, primarily from Bessemer Venture Partners' frameworks published in their cloud indices, which are built on sales-led company data. Applying it to PLG is like using enterprise sales cycle benchmarks to judge your self-serve conversion rate, the inputs are different.

For PLG companies with a meaningful self-serve motion, 18 months is the correct ceiling for blended payback. Here is why:

PLG companies typically acquire at lower ACV than sales-led. The whole model is premised on land-small-expand-later. A $50/month converted account does not look great on a 12-month payback target. It looks fine on 18 months once you factor in that the account will expand to $180/month by month 8 if your expansion motion works.

The expansion revenue adjustment is the one most founders forget. PLG payback should be calculated against expected MRR at 90 days post-conversion, not MRR at conversion. If your median converted account expands 2.4x in the first 90 days (which is a realistic figure for seat-expansion products), and you are calculating payback against month-zero MRR, you are making your unit economics look worse than they are.

The adjustment is simple: replace the denominator with "expected MRR at 90 days" based on your cohort data. Run the last 6 cohorts, calculate the median MRR at day 90 for each, and use that figure. Your payback number will drop, accurately.

What does not save you: viral coefficient. Yes, if each paid user refers 0.3 additional free users and some fraction of those convert, your effective CAC per account is lower. But viral coefficient is a second-order adjustment. Get the first-order formula right before you add viral credit, and be honest about whether your referral loop is actually systematic or whether three loud customers made it look that way.

The account expansion mechanics are closely tied to retention outcomes. If your onboarding is leaving activation on the table, payback will stay elevated no matter how you calculate it, the onboarding-retention correlation in B2B SaaS is worth understanding before you optimize acquisition cost.

Three Places PLG Founders Inflate Their Apparent CAC

Seeing a bad CAC payback number usually means one of three things went wrong in the calculation.

Counting free-tier infrastructure inside CAC. AWS costs for free accounts, Cloudflare bandwidth for free users, the Segment events fired for non-paying users, these are product costs. They belong in your cost of goods for the free tier, not in your cost of customer acquisition. The test: would this cost disappear if you had zero free users? Yes = product cost. Would this cost disappear if you had zero paid users? Yes = acquisition cost. Keep them separate.

Using MRR at conversion instead of stabilized MRR. Accounts converted from free frequently have billing hiccups, failed card charges, and plan downgrades in the first 30 days. An account that converts at $99/month and downgrades to $29/month by day 20 did not actually contribute $99/month to your denominator. Use MRR at day 31 post-conversion as the denominator input. It adds a month's lag to your reporting but it's accurate.

Ignoring channel-level cost differences. Organic SEO-driven free signups that convert have a fundamentally different CAC than paid social signups. If 60% of your free signups come from organic and 40% from paid, and you blend the acquisition cost, you are hiding the fact that your paid acquisition converts at a different rate. The blended number masks a channel that may be destroying your economics. Breaking down inbound SQL cost by channel is the same discipline applied upstream, you need channel-level visibility before you can trust any aggregate number.

When PLG CAC Payback Signals a Broken Model

Fixing the formula sometimes reveals the model is fine. Sometimes it reveals a real problem.

Payback over 24 months without an expansion revenue floor. If your median account stays flat at conversion MRR for 12 months and payback is at 24, you are not building a PLG business. You have a low-ACV sales-led business that happens to use a free trial. The distinction matters because the fix is different: you need either higher ACV at conversion or a real in-product expansion trigger, not a faster sales cycle.

PQL-to-paid conversion under 3%. If fewer than 3 in 100 users who hit your activation threshold are converting to paid, the funnel between activation and purchase is broken. This is not a CAC payback problem, it is a pricing or onboarding problem presenting as a unit economics problem. Fixing the payback formula will not help. You need to track product-market fit metrics at the activation layer before spending more on acquisition.

ACV that cannot absorb even clean CAC. If you are spending $400 to convert a $19/month account on a 75% gross margin product, payback is 28 months. No formula adjustment will rescue that. The fix is pricing, not measurement. Specifically: look at what your top 20% of accounts actually need and price the product to reach them, rather than optimizing your conversion funnel for the $19 segment.

The honest signal that your model is broken rather than miscalculated: when you fix the formula, the payback gets worse. If you were counting free-user costs in CAC and removing them makes payback look better, you had a measurement problem. If removing them does not change the number meaningfully, your actual conversion spend is the issue.

Tracking This Without a Data Team

You do not need a data engineer or a BI tool to run this correctly. Four columns in a spreadsheet will get you 80% of the signal.

Column	Source	What it captures
Cohort month	Stripe or your billing system	When the account first paid
Acquisition channel	UTM data or your CRM	Which channel drove the free signup
Conversion-touch cost	Sales team time log + CRM activity	Hours × blended rate for any human touch
MRR at day 31	Stripe MRR movements	Stabilized first-month revenue

Pull this monthly. The accounts where conversion-touch cost is $0 (pure self-serve) are your self-serve cohort. Everything else is your sales-assist cohort. Calculate payback separately for each group and track the trend, not the absolute number. A self-serve payback moving from 9 months to 7 months over two quarters is a signal that your product is improving conversion. A sales-assist payback moving from 14 months to 19 months is a signal that your ACV is not keeping up with your sales team cost.

This is the same logic that applies to channel-level unit economics: the aggregate hides the story. The per-cohort view tells you where to act.

One area worth connecting to this workflow: the content driving your free signups matters enormously to the cost side of the equation. Organic-driven signups arrive with near-zero acquisition cost. If you're not tracking which blog topics and keywords are producing signups that actually convert to paid, you're making acquisition decisions blind. The waitlist is live at morbiz.ai/marketing-engine, MorBizAI closes that loop by pulling your Search Console striking-distance keywords, drafting SEO posts in your brand voice, and publishing directly to WordPress, so the content feeding your PLG funnel actually gets shipped instead of sitting in a Notion backlog.

Keep the spreadsheet live. Update it the first Monday of each month. Run it for three months before drawing conclusions about your payback trend, a single month is too noisy to act on.

CAC payback for product-led growth is not a more complicated version of the same calculation. It is a different calculation. Run it right, and it tells you exactly which part of your acquisition engine is working and which is quietly burning the cash you need to reach your next milestone.