The Numbers, Defined

Every article here reads a book in money: what the accounts bring in, what they cost to win and serve, and when they pay it back. These are the definitions behind those numbers. Where a page uses a different setting, it says so on the page.

Net revenue

What a customer pays, less refunds, discounts and the pass-through costs you never keep. In a payments book, the pass-through costs are interchange, the card brands’ assessments and the processor’s costs, so net revenue is what’s left of each statement after them. It’s the top line for everything below.

Variable costs

The costs that rise and fall with each account, and go away when the account does: agent and partner residual splits, commissions paid as the account bills, chargeback and fraud losses, and the cost of servicing it.

Contribution

Net revenue less variable costs. It’s what each account contributes toward the fixed costs of the business, and then toward profit. It isn’t profit, because the fixed costs still come out of it. Contribution margin is contribution as a share of net revenue.

Contribution after acquisition also takes out what it cost to win the account. That’s the version to manage to month by month, because it’s the one that says whether growth is paying for itself.

Acquisition cost

The marketing and sales spend that wins a month’s new accounts: the media, the agencies and tools that run it, and the sales pay for signing them. Divide it by the accounts won and you have the cost per acquisition.

R2M, the revenue return at month m

Take a month’s new accounts, a cohort, and read it m months later: the net revenue the cohort has brought in so far, for every $1 invested in it. R2M₁₂ is the return at month 12.

The business sets what counts as invested, in one of three settings:

• acquisition only
• acquisition and variable costs
• acquisition, variable costs and operating expenses

The return to aim for follows from the setting. On this site, R2M counts acquisition and variable costs unless a page says otherwise. Read at months 1, 3, 12 and 60, the change from one checkpoint to the next shows how fast a cohort’s return is building. “Revenue return” on any page of this site means R2M.

The payback month

The first month a cohort’s R2M reaches 1, with variable costs counted in the investment. From that month on, the cohort has covered what it cost to win and to serve, and everything above 1 is contribution after acquisition: R2M minus 1, for every $1 invested.

When only acquisition counts as invested, R2M is a revenue return, and the cohort pays back once it reaches 1 divided by its contribution margin. At a 50% margin that’s 2. At 30% it’s about 3.3.

One cohort, three settings

Put illustrative numbers on one cohort. Its accounts cost $10,000 to win, $6,000 in marketing and $4,000 in sales pay. By month 12 they’ve brought in $24,000 of net revenue, with $10,000 of variable costs along the way, $8,000 of agent splits and $2,000 of losses and servicing. Their share of the operating expenses is $5,000.

What counts as invested	Invested	R2M₁₂	What it says
Acquisition only	$10,000	2.40	A revenue return. The cohort has paid back, because 2.40 clears 1 divided by its 58% margin, 1.71
Acquisition and variable costs	$20,000	1.20	Paid back, with $4,000 of contribution after acquisition, 0.20 for every $1
Acquisition, variable costs and operating expenses	$25,000	0.96	Not yet covering its share of the operating expenses

Contribution is $14,000 before acquisition, 58% of net revenue, and $4,000 after it, 17%. Same cohort, same year, three honest numbers. A target set on one setting and read on another is where most arguments about return start.

Retention

How much of what you had, you hold on to. Its opposite is attrition, or churn.

Net revenue retention (NRR) reads it in net revenue: the net revenue this year from the accounts you had a year ago, against what the same accounts brought in then, with new accounts left out. It counts the accounts that grew and shrank as well as the ones that closed.

Power Law Nested Means

The two-split test on what each account brings in over time: split the book at its average net revenue, then split each half at its own average. The four groups it makes are called Tiers A to D. A and B are the above-average accounts, C and D the ones below. Read the tiers on contribution too, where you have it, because a big account can carry a thin margin.

Baseline

The growth and expectations set by what’s come before, carried forward. For a cohort, a tier or a channel, it’s what that group was doing in the months before anything changed, how much of its revenue it kept from one month to the next and what each account brought in, carried forward at the same rate. For a year’s plan, it’s the book you have at its run rate, plus new accounts arriving at the pace and mix they’ve been arriving at. Every “above baseline” on this site is read against it, over time.

In an incrementality or lift test, the holdout is the baseline: the group left alone shows what would have happened without the change. Where you can run one, it’s the cleanest check that a change, and not the season or the market, is what moved a number.

Qualified-conversion bidding

Reporting only the conversions whose answers clear a line, such as an above-average band on a sorting question, to the ad platforms, and letting them bid on those alone. The signal is anchored to value: its answers are checked against what the accounts became before anything is bid on it. The bid strategy doesn’t change. What changes is which conversions the platforms are told about. The guide to signals and attribution has the build.

The value line

The deepest result that goes back to the ad platforms for a lead that qualified on the form: its account’s first full month of net revenue above the median, sent back as a later result. The stages before it go back too, for the same leads, as secondary conversions. Tiers and reporting keep the full quarter. It isn’t the line on a form’s answers, which the qualified conversion clears.