Everything I write about AI in finance comes off a screen. Docs, demos, my own client pilots.

What I can't get from a screen: what everyone else already tried and quietly turned off.

Shift AI Europe runs on 1-2-1 meetings and roundtables instead of keynotes — 1,000 founders, hard-capped, 13-14 October.

I'll be there. If you're going, hit reply and let's get a coffee.

Consulting | Shop | Website | Newsletter | Speaking

Hey {{first_name}} 👋, I hope you’re having a great week! In the last issue, we discussed about Deferred Revenue, and now we are moving with the next topic from Reporting content.

Let’s talk about ⬇️

McKinsey looked at 100+ public SaaS companies and found something that should scare anyone at $5–50M ARR.

Top-quartile companies recover their customer acquisition cost in a median of 16 months.

Bottom quartile: 47 months.

Nearly four years of pre-financing a customer before seeing a dollar of gross profit come back. Most of those companies would tell you their LTV:CAC ratio looks fine.

And that's the whole problem with unit economics as most teams practice them. The ratio can be healthy while the cash timing quietly kills you.

Here's what actually matters, and in what order.

Every SaaS company with meaningful annual prepay has a smaller version of this. Most founders have no idea it's happening.

And here's the part that should worry you: the same mechanic that makes a healthy quarter look strange can make a sick quarter look healthy.

I'll show you how.

A CFO and a VP of Sales walked into the same board meeting with two different CAC numbers for the same period.

$18,400 and $11,200.

Neither was wrong. The CFO was running fully-loaded acquisition cost — SDR salaries, allocated marketing time, the whole thing. The VP was running direct sales commissions and program spend.

Two legitimate numbers. Two completely different decisions sitting on top of them.

The fix was a two-row table with both definitions written out. Sales-motion CAC for quota-setting and commissions. Fully-loaded CAC for capital allocation and the LTV:CAC ratio.

Four minutes to align.

If your team argues about CAC during board prep, the disagreement is upstream. It's about what a unit is, not about arithmetic.

Three ways to define it, and you have to pick one in writing:

A customer account. One contract, one unit. Fine if your customers look alike.

A cohort. Everyone acquired in the same month or quarter. This is the right level for churn behavior and LTV over time.

A seat. Works for per-seat pricing. LTV per seat needs expansion and churn assumptions that get ugly fast.

The definition decides whether expansion revenue counts in LTV (it should), whether your CAC denominator is new logos only (it should be), and whether a "churned unit" means a full cancellation or a seat reduction. Those have very different cash implications.

The slide: LTV:CAC of 3.2:1. Big number, center of the page, green.

What wasn't on it: which cohort. Blended across everything since inception.

Pulled the last two quarters separately. 1.8:1.

The 3.2 was real. It was also mostly historical — early customers acquired cheaply through founder-led sales, carrying a number that no longer described how the business acquires anyone.

The fix: never show a blended ratio without the most recent cohort next to it. Two columns instead of one.

CAC. Total S&M spend divided by new customers. Same period for both. The errors are predictable: excluding SDR and AE salaries understates it. Including customer success overstates it. Using contracted ARR as the denominator instead of customer count distorts it whenever deal sizes vary.

LTV. (ARPU × Gross Margin %) ÷ Churn. On gross margin, always. At 60% margins, a $30,000 revenue-based LTV is $18,000 of actual contribution. Different picture entirely.

Payback period. CAC ÷ (ARPU × Gross Margin %). This is the cash-timing metric. LTV:CAC tells you eventual return. Payback tells you how long you're financing it.

Gross margin. Not a standalone number. It's the multiplier that changes what every other figure means.

That last one deserves a second: a company at 80% gross margin with a 3:1 ratio is in a structurally stronger position than one at 50% with the same ratio. Same ratio. Roughly 60% more dollars coming back per dollar spent.

SaaS Capital has reviewed thousands of SaaS financials and their read is blunt — GAAP doesn't clearly define what belongs in SaaS cost of sales, so everyone's using judgment. Their benchmark for pure license revenue: 80–85%.

I'm opening a small group for SaaS finance operators. Every month: one finished financial model and one live session building it — ARR bridges, burn multiple, usage-based pricing, board packs.

Not a course. Just the models I build for clients, plus a room where you can ask why line 47 is doing that.

$490 for the first year, capped at 30 people. $990 after. Closes Aug 30th.

A single LTV figure on a board slide is false precision dressed up as rigor.

The honest version has three:

Best case — top retention cohort, strong expansion
Base case — median cohort behavior
Stress case — cohorts running 20–30% worse churn than current median

"Our base-case LTV is $18,000, with a range of $12,000–$24,000 depending on cohort vintage" is a sentence that survives contact with an investor. $18,000 is not.

Same principle on the ratio. Here's what it actually signals:

But 3:1 is a conversation starter, not a grade. A 2.5:1 with 9-month payback and 120% NRR is a better business than a 4:1 with 36-month payback and 95% NRR. The ratio doesn't know about time.

McKinsey looked at 40 public B2B SaaS companies. Those with NRR above 120% traded at a median 21x EV/revenue. Below 120%: 9x.

That's not a marginal difference. It's the single metric with the strongest correlation to valuation in their sample.

One caveat worth carrying: OpenView's benchmark data showed top-quartile expansion-stage companies watching NRR compress from 119% to 107% year over year. Expansion is harder to sustain than it was in 2021. Benchmark against now, not the peak.

McKinsey's number: companies exceed Rule of 40 performance only 16% of the time. Fewer than a third achieve it consistently. Median revenue growth in their sample of public SaaS companies above $100M was 22% — well under the hypergrowth narrative most teams benchmark themselves against.

Which makes the pressure to hit 40 at $19M ARR particularly expensive.

I watched a company do exactly that before a Series C process. Score of 28. They cut two CS roles and paused a planned R&D hire.

Score moved to 34. One quarter.

Six months later: NRR down from 107% to 99%. Two enterprise expansion conversations gone quiet. Roadmap slipped a quarter on a feature three customers had been promised.

The Rule of 40 was built for public companies above $100M in revenue. At $5–15M ARR, a score of 20–30 with improving trajectory means more than a static 40 achieved by cutting.

The useful question at your stage isn't the composite. It's this: is your growth rate declining faster than your margin is improving? If yes, no dashboard fixes that.

Each one has an owner, because metrics without ownership are decoration.

1. Define the unit in writing. (Head of Finance or founder) Account, cohort, or seat. Inclusion and exclusion rules documented. Do this before recalculating anything.

2. Rebuild CAC from fully-loaded costs. (Finance) Four quarters of every S&M cost. Divide by new logos only. Compare to whatever you've been reporting.

3. Calculate LTV on gross margin, by cohort. (Finance + RevOps) Last six quarters, actual retention and expansion data. If the newest cohort is declining, name the cause.

4. Map payback period to cash runway. (CFO or founder) If payback is approaching your available runway, that's the most urgent slide in your next board deck. Not a footnote.

5. Track NRR monthly. (RevOps + Finance) Decline shows up in monthly data 60–90 days before it damages the annual headline. Quarterly tracking means you find out too late to react.

6. Run Rule of 40 for four quarters, not one. (Finance) Trend over level. If it's improving — growth or margin? If deteriorating — which component is moving faster?

7. Present as a range with stated assumptions. (Founder + board prep) Replace the single-number slide with base, upside, stress. Owner named per input. That's the difference between reporting history and framing a decision.

Unit economics isn't a reporting exercise. It's the analytical basis for every capital allocation call you make — S&M hiring, which segment to double down on, whether payback is compatible with your cash position, how to frame growth versus profitability in your next raise.

The companies that get this right at $5–50M ARR don't track more metrics. They have better-defined inputs, ownership at the assumption level, and they present ranges instead of points.

The failure mode I see most isn't bad math. It's that CAC and LTV get calculated for the deck and never used to decide whether the next two AE hires extend or compress payback.

That's the gap worth closing.

Reply "UNIT" and I'll send the template — channel-level LTV:CAC, payback by segment, the lagged CAC calc, and the decision worksheet that keeps board meetings from becoming definition arguments.

Made espresso at 6am to write this and it was the best decision of the day.

{{first_name}} You can get free stuff for referring friends & family to my newsletter 👇

50 referrals - Cash Flow Models Bundle 💰

10 referrals - SaaS Financial Model 📊

{{rp_personalized_text}}

Copy & Paste this link: {{rp_refer_url}}