Building Paperturn's SaaS unit economics from zero

Executive summary

Why measurement had to come first

Without a measurement layer, no benchmark could apply and no target would be fair.

SaaS targets revolve around customer and trial acquisition (both volume and cost), and they only mean something compared against benchmarks fair to your business. Measurement is the infrastructure that makes both possible. With it, you can build decision frameworks that greenlight or kill campaigns and segments.

We entertained two “true north” KPIs (CAC payback and a 3:1 LTV:CAC ratio) and settled on a 9–12 month CAC payback target, country-dependent.

Strategic consultant Martin Bjørn flagged the reasoning: a product-led-growth company (where customers sign themselves up rather than going through a sales team) typically has higher cancellation rates than B2B SaaS. That uncertainty makes CAC payback the more fiscally responsible metric: it ensures the cash actually comes back. LTV:CAC targets were reserved for time-tested winners: campaigns and country segments where we could afford a higher acquisition cost.

The methodology

Two flavors of payback.

I shipped both because hiding the contrast would have been its own dishonesty. The simple equation is good for speed and quick comparisons, but the strict nature of the empirical method allows you to understand cohort behavior across time. That’s the version that survives a conversation with a financial officer.

When you run these equations separately for monthly vs annual cohorts (groups of customers acquired in the same period, paying monthly or yearly), by country, by acquisition window, the segmentation reveals where efficiency actually lives — and where it doesn’t. Same standards applied, just with different inputs.

A formula that gives you a target.
The most useful question for a marketing team is also the simplest: at our cancellation rate, what’s the worst acquisition cost we can afford and still hit a 12-month payback? Once the empirical model is set up, that question collapses into a single equation:

At Paperturn’s 86% gross margin and a 12-month target, this simplifies to a rule of thumb that fits on a sticky note:

A 5% monthly cancellation rate needs an efficiency ratio of ~1.43× to hit 12-month payback. A 10% rate needs ~1.85×. Plug in your number, get your target. The same formula run backwards tells you the maximum cancellation rate any observed efficiency ratio can sustain.

Segregating monthly vs annual.
Blending was the default. No one at the company was operating on subscription-business thinking. Same acquisition cost per account, same 86% gross margin, different cancellation patterns over time: annual cohort earned its acquisition cost back in ~10 months; monthly cohort took ~18–20. Same dollars in, meaningfully different dollars out. The reallocation toward annual-producing geographies followed downstream.

The 5-tier country framework

Top 10 markets accounted for 84.6% of country-attributable spend across 79 total markets. Each had a different efficiency, payback, and trial-to-paid conversion profile. Naming them one-by-one was unwieldy. Meetings got stuck on edge cases. Tiering forces the action: the label dictates the decision.

22-month window with Meta excluded (2024-06 → 2026-03), n = 2,283 accounts. Bifurcated = annual customers earn payback, monthly customers don’t, in the same market.

The reallocation

By late 2024, blended acquisition cost was up ~110%. Google’s AI Overviews had pushed organic traffic into paid search and bid prices with it. Cancellations and revenue per customer were stable; the problem was acquisition. The framework named which markets to act on.

Adding up the cuts across underperforming geographies and monthly-heavy campaigns gave a reallocatable budget of ~25,000 DKK/month: a portfolio shift, not a budget cut, trading unprofitable volume for durable quality. The founder approved.

The Q1 2026 recovery, and country-level honesty about it

Portfolio recovery was real and material: empirical payback 19 → 11 mo (−42%), ad-spend efficiency +53%, blended acquisition cost −26%.

But it was country-uneven. Denmark posted its best-ever quarter (acquisition cost 543 DKK, 3.20× efficiency, ~5-month empirical payback). Australia and France also recovered. The UK had been declining since its Q3 2025 peak. The US’s Q1 2026 was the second-worst US quarter on record. Competitors had begun aggressively bidding on our branded search terms and running comparison SEO campaigns, and click prices in our highest-value market reflected that.

Reporting Q1 2026 as “we recovered” would have been technically true and substantively misleading. Denmark’s best-ever quarter doesn’t fix the US’s continuing problem. The methodology’s point was to surface what blended numbers hide; reporting the blend alone would have been the dilution it was built to prevent.

What persists

All four artifacts I built are still running: the customer-lifetime-value formula, cancellation definitions and dashboards, the methodology for tracking whether customers spend more over time, and the CAC payback model with its 5-tier framework. The trial-conversion and acquisition-cost targets I set are the paid agency’s operational benchmarks today.