What a customer is worth

CAC = sales and marketing spend in a period ÷ new customers won in that period. Two choices sit inside that line, and every disagreement about CAC is one of them.

1. Fully loaded or paid media only. Fully loaded counts everything you spent to win customers: media, content and events, the salaries of the people who sell and market, their tools, the agency. Paid media only counts what the ad platforms billed. For Cadence (constructed) that is $1,800 against $500 for the same new account.
2. Blended or per channel. Blended divides all spend by all new customers, including the ones who arrived on their own. Per channel divides one channel's spend by the customers that channel won, which depends on how you credit a customer who touched three channels: that is the attribution question, owned by learn-marketing-attribution-with-phoebe.

Which one for which decision. To decide whether the business can afford its growth, use fully loaded and blended: it is the number the bank account feels. To decide where the next dollar goes, use per channel, and use the marginal version where you can, because the last customer a channel won cost more than its average. Paid media only is a useful number for the person running the ads and a dangerous one in a board deck. Whichever you pick, write the definition next to the number; a CAC with no definition is two people agreeing on a word and disagreeing on the money.

One timing trap. A Teams deal that takes two months to close was paid for by last quarter's spend and counted in this quarter's customers. Long cycles make a single month's CAC jump around; a rolling three-month figure is steadier.

The funnel bench in session 4 is a CAC calculator with the definition fixed: one $20,000 monthly budget, constructed cost per lead by channel, and close rates measured on the real 2018 Olist cohort (5,998 leads, 781 closed deals). At the printed defaults, the same $20,000 buys a customer for $368 under the best split and $627 under "cheapest leads first". Nothing about the budget changed. Only where it went.

That is the second lesson of CAC: it is not a property of your product, it is a property of your spending. The spend is constructed on the bench; the close rates are real. So the bench tells you the shape of the trap, cheap leads that rarely close, and not the dollar value your company would see.

The bench also shows the definition problem in miniature. It counts only media spend at a cost per lead, so it is a paid-media-only CAC. Load it with salaries and tools and every preset rises, and the ranking survives.

A Cadence Teams account pays $120 a month. It does not hand the company $120 a month to spend: hosting the recordings, transcription, payment fees and support come out first. With a 75% gross margin (constructed), what is left to pay back the acquisition cost is $120 × 0.75 = $90. Divide by revenue instead and payback looks like 15 months, which is a cheerful answer to a question nobody asked. The money that repays CAC is the money left after delivering the product.

So Cadence's payback is $1,800 ÷ $90 = 20 months. Measured against Skok's own reference points, that is well past his 12-month ceiling and nowhere near his 5-7 month band. It does not mean Cadence is doomed. It means every Teams account must stay at least 20 months before it has made the company a dollar, and that 1 in 40 accounts leaving each month (2.5%) is going to keep a lot of them from getting there.

Payback has one big virtue over the other metrics tonight: it needs no forecast. CAC is measured, gross profit per month is measured, and the answer is a count of months you can check against real accounts.

The CAC is paid now; the gross profit arrives over 20 months. Someone funds the gap. Win 50 Teams accounts this month at $1,800 each and $90,000 leaves the bank today, while those accounts return $4,500 a month between them. Growth with a long payback is a business that gets poorer the faster it succeeds, until the early cohorts start paying back. That is why finance teams ask about payback before they ask about lifetime value: payback is the shape of the cash hole.

Gross margin and cash timing are finance's subject, and learn-finance-with-phoebe teaches both, along with the three statements where the hole shows up. It does not teach CAC or LTV; those stay here. If your gross margin is a guess, fix that first: every number below inherits it.

An annual prepaid plan changes the cash shape without changing the payback arithmetic. If the customer pays twelve months up front, the hole is smaller even though the profit per month is the same.

The common simple form, the one Skok uses: LTV = monthly gross profit per customer ÷ monthly churn rate, which is the same as average revenue per account × gross margin ÷ churn. For Cadence: $90 ÷ 0.025 = $3,600. Against a $1,800 CAC that is a ratio of 2.0, below Skok's 3.

Dividing by churn is a compact way of saying "multiply by the average lifetime". At 2.5% leaving each month the average Teams account stays 1 ÷ 0.025 = 40 months. That one number carries the whole forecast. Four assumptions sit inside it:

1. Churn is constant. Real cohorts usually lose more customers early and fewer later. A single rate averages over that shape.
2. No expansion. The formula assumes the 8 seats stay 8 seats. Teams that add seats grow the value; teams that shrink cut it.
3. No discounting. A dollar of gross profit in month 38 counts the same as one this month.
4. The churn rate came from somewhere. On a young product it was measured on a young cohort, often over fewer months than the 40 it is now predicting.

Practitioners disagree about the second and third. Some add expansion revenue and report a larger LTV; others leave it out on purpose, because counting on growth you have not yet seen flatters the number used to justify spending. Some discount future profit at a cost of capital; others argue the discount rate is one more assumption and keep the simple form. There is no settled answer. The professional norm is to say which version you used, every time you quote the number.

The ratio has three inputs that matter and each belongs to a different team. All numbers below are constructed for Cadence, one lever moved at a time with the others held where they are.

• Price and seats (monthly gross profit). 10 seats instead of 8: $150 a month, $112.50 of gross profit, payback 16.0 months, LTV $4,500, ratio 2.5. Pricing and how demand responds to it are taught in learn-economics-with-phoebe.
• Churn (the lifetime). 1.5% instead of 2.5%: LTV $6,000, ratio 3.33. Payback is still 20 months, because payback does not look at churn at all. What moves churn, and how retention reshapes LTV, is learn-customer-retention-with-phoebe.
• CAC (the cost). $1,200 instead of $1,800: payback 13.3 months, ratio 3.0. It reaches 3.0 and stops there. This is the lever session 4 put on the bench.

Notice what the figure shows: the lever marketing controls most directly, CAC, is the one that only just reaches the line. The biggest move came from churn, which marketing influences only through who it brings in. That is the strongest argument for session 1's not-for list: the cheapest way to raise LTV is to stop acquiring customers who will leave.