A Performance Revenue Optimization experiment

You're about to run a dealership's marketing — on the metrics every marketer is given.

Cost per acquisition. Conversion rate. Volume of sales. Six months. Five ads. One budget. Make the calls any competent marketer would make — then we'll show you what those calls actually did.

First, one idea you'll need: Rate of Revenue Return

Rate of Revenue Return @ T (RRR@T) measures the revenue a cohort returns relative to its cost — and the cost is whatever you load in: media, sales commission, the dev/ops of running the department. Each cost is locked to the month it occurred (annual and lumpy costs amortized into the months they belong to — the investment, or "INV," side of the model), and the revenue is then tracked forward into the funnel, to capture when it actually lands — read at 1M, 3M, 6M, and 12M. A cohort destined to be valuable pulls above the line early; a weak one never gets there. Because the denominator is the full cost, cheap clicks don't rescue a cheap customer.

RRR building over time · overperformer, average, underperformer

The overperformer — pulls away early The average our forecasts assume — volume-weighted, so it sits low The underperformer

Read in summary

1M / 3M / 6M / 12M are the standard fields of view. The shape of the early curve tells you where the cohort is headed long before the 12-month number lands.

Normalized across cohorts

We index every creation-timeline cohort to the same baseline, so a cohort acquired in March is comparable to one from last year — and you can flag over/underperformance early.

The catch

The marketing chair never sees RRR. It sees CAC and conversion. That's the entire experiment.

In this experiment your monthly budget stands in for that fully-loaded cost basis — held fixed, so you can feel the mix do the damage. In the real model, commission and ops ride along with every deal you close.

The launch

Your team is kicking off five ads — one for each kind of buyer.

Here's the creative going live this quarter. Take a good look — this is the last time you'll see the people behind the numbers.

From here you'll manage them the way most teams actually do — not as faces, but as rows in a spreadsheet: a label, a CAC, a conversion rate, a car count, and a budget slider. Pick your winners.

Your department

Month 1 of 6

Monthly budget: $120,000

This is your quarter, the way the team actually works it — rows in a spreadsheet. Lower CAC and more cars is the goal. Set each row's budget share; cut the laggards. A row at 0% drops off the sheet.

Your dashboard this month

Six months later · the debrief

You optimized exactly as asked. Here's the bracket you built.

Every month you cut the rows with the worst CAC and the lowest volume, and you fed the winners. You stopped seeing them as buyers the moment the spreadsheet opened — so here's who they actually were:

What the chair saw — and what it never showed you

Your dashboard, month 6

Blended CAC

—

Cars sold / mo

—

CAC down, volume up, every month. By the numbers you were given, this was a flawless six months.

The layer you were never shown

Your book's RRR@12M vs a 4.0× baseline

—

The cohorts you built are set to return far less than the averages the business forecasts on.

RRR@12M by car type — the value behind each ad you ranked on CAC

4.0× baseline — what the dealership's forecasts assume

Ad A — the Connoisseur — was the engine you cut first. Its LuxuryX cohort returns 8.5× by month 12. The champion you crowned, Ad E / Free Lamborghini, returns 1.4×. You ranked them on the one number that pointed the wrong way.

The global number, despite every "right" call

Month	Blended CAC	Cars sold	Cohort RRR@12M	Revenue vs forecast

Lower CAC every month. More cars every month. And the book moved from beating its forecast to missing it badly — because optimizing CAC and volume quietly changed who walked onto the lot. The lift was real. It was also the population shifting under you, fed back as proof it was working.

The PRO method would have flagged it in Month 2

Same six months, but with revenue attributed by signal and every cohort normalized to the baseline so comparisons stay honest. Underperformance shows up in the early RRR curve — long before the 12-month damage is booked.

That's the whole difference: the marketing chair measured activity and called it performance. PRO measures the revenue each signal actually returns, normalized so the moving baseline can't disguise it — and tells you in month 2, not month 8.