Measurement · Philosophy
The metric follows the objective.
There is no single metric we optimise every client towards.
A brand with 40% gross margin and a warehouse full of last season's stock does not have the same measurement problem as a subscription brand whose economics live in month nine. Picking one number for both is how accounts end up optimised precisely towards the wrong thing.
01 / Objective to metric
Six objectives. Six ways of being right.
Each of these is the correct primary metric for one commercial question, and a misleading one for the other five.
PRODUCT PROFITABILITY
POAS
Contribution margin per pound of ad spend. The right metric when the question is whether the products being sold are worth selling at the price the auction demands.
POAS explainedCUSTOMER ACQUISITION
nCAC, payback, LTV
New-customer acquisition cost, the months to recover it, and the value that follows. POAS on a first order will always understate a Gateway product.
INVENTORY AND CASH
Cash recovery, sell-through
When capital is trapped in ageing stock, the objective is not margin. It is converting units back into cash before the next buying cycle, at a defined floor.
INCREMENTALITY
Holdouts and experiments
Attributed revenue is not necessarily incremental revenue. Geo holdouts, brand pause tests and campaign experiments answer what would have happened anyway.
DEMAND CAPTURE
Brand and generic decomposition
Blended account performance is mostly a report on how much brand demand you have. Splitting it turns a vanity number into a diagnostic.
BUSINESS OUTCOME
Contribution
What reaches the P&L after COGS, fulfilment, returns and media. The only number that survives contact with a board meeting.
02 / Where POAS fits
POAS is the answer to one question, not all of them
It answers product profitability better than any other single number. It does not answer acquisition, cash or incrementality.
POAS ANSWERS
Whether the products this account is selling generate contribution at the price the auction charges. It is the right lead metric for most catalogue-led ecommerce, which is why it leads almost everything we publish.
POAS DOES NOT ANSWER
Whether a Gateway product losing money on its first order is worth acquiring. Whether £250,000 of ageing stock should be discounted to release cash. Whether the revenue would have arrived without the ad. Those need different numbers.
03 / Anti-patterns
Measurement that flatters rather than informs
- 01One blended ROAS target applied across a catalogue with a 50-point margin range
- 02Revenue reported as conversion value with returns never netted off
- 03Brand and generic reported together, then read as account performance
- 04Platform-reported conversions reconciled against nothing
- 05Attribution model changes presented as performance improvements
- 06A single north-star metric held constant while the commercial objective moves
Measurement questions
Tell us the commercial objective and we will tell you which number should be governing the account.
Book a commercial review