Data Nexus

Check · 30 minutes · 5 questions

Conversion is up 6.4×. Revenue grew 40%. Abandoned carts down 65%.

Ask for these first

In writing, before the meeting. The request is half the check: everything below is one line to supply if it exists.

The period, with both dates
A multiple with no interval is not a measurement. Two months and two years are different claims and the number is identical.

The base — the absolute figure the multiple is calculated from
6.4× on a rate of 0.3% is a rate of 1.9%, which may be below the category floor. The multiple is large because the base was small, and the base is the part that was omitted.

The definition of the numerator and the denominator, in one sentence each
Almost every disputed marketing number is a dispute about what was counted, not about the arithmetic. Ask what one conversion is and what population it is divided by.

The same figure for the equivalent prior period
Separates a real change from a seasonal one. A florist reporting February against January is reporting Valentine's Day.
01/The questions
  1. 01

    Does the number carry a period and a base?

    These are the two things a supplier omits when the number is weaker than it sounds, and both are one line to supply if they exist.

    If insteadThe reply describes the improvement again in different words instead of giving the two figures.


  2. 02

    Was anything else changed in the same window?

    A rebuild, a price change, a new channel and a seasonal peak all move conversion. Attributing the whole move to the work is a claim about causation from an uncontrolled comparison.

    If insteadSeveral things changed and the report attributes the result to one of them without saying why.


  3. 03

    Does the denominator population match the numerator population?

    Conversion measured against sessions and conversion measured against people are different numbers, and a switch between them produces a large improvement with no change to the business.

    If insteadThe two are described in different units, or the definition changed part-way through the period.


  4. 04

    Is the revenue figure attributed or settled?

    Attributed revenue is a model's opinion about which touchpoint deserves credit. Settled revenue is what the payment processor received. They differ, sometimes by a lot, and only one of them is money.

    If insteadThe revenue in the report cannot be reconciled to the payment processor within a few per cent.


  5. 05

    Do the funnel steps multiply out to the headline?

    A real funnel is arithmetically closed: contacts, qualified, attended, paid, revenue. If the stages do not compose into the top-line claim, one stage is being measured differently from the others.

    If insteadThe stages are not supplied, or supplying them is treated as an unusual request.

02/What the answers mean
It is real
Period, base and definitions arrive without argument, the funnel composes, and the revenue reconciles to the processor. The number can go in a board pack.

It is not
Any of the four cannot be supplied. This does not mean the work was bad — it means the number does not measure it, and a different measurement is needed before anyone decides anything on this one.

Cannot tell
The figures arrive but several changes coincided. The honest report says the outcome improved and the cause is not isolated, which is a normal state of affairs and a reasonable thing to say out loud.
03/Where we failed it

Our own home page and case record, in the BHM Store block

Three figures — conversion up 6.4×, abandoned carts down 65%, revenue up 40% — carried no base and, for two of the three, no period. Held to this check, we would have rejected them. The multiple was ambiguous as well as unbaselined: against that store's 2.3% starting rate, 17% is 7.4 times as much while the increase is 6.4 times, so the label said one thing and the arithmetic another. Three blocks below on the same page the cold-traffic record passes in full — AED 1,111.04 spent, 66 first contacts at AED 16.83, 6 attended at 9.1% of contacts, 4 paid at an average enrolment of AED 8,900, AED 35,600 received, March–April 2025 — every figure deriving from its neighbours.

SinceAugust 2026. The multiple is replaced by the pair it was computed from, 2.3% → 17%, which needs no interpretation. The abandoned-cart figure was withdrawn rather than restated, because no baseline or window for it exists to publish; the catalogue reconstruction that took its place — 8,500 SKUs — is checkable against the live store. The revenue figure stays with its period on the label. All of it remains the operator's analytics rather than our measurement, and the case record says so.

The words this uses
Conversion rate
The share of a defined population that completed a defined action — meaningless until both the population and the action are stated.

Vanity metrics
Numbers that rise without anything else improving — followers, impressions, likes — and which, when bought, actively damage the account that carries them.

Attribution
The rule that decides which touchpoint gets credit for a sale — a modelling choice, never a fact.

ROAS
Revenue attributed to advertising divided by the advertising spend — a channel efficiency ratio, not a profitability one.

Cohort analysis
Grouping customers by when they arrived and following each group separately, so that changes over time are visible instead of averaged away.

Session, user, event
The three units analytics counts in, each measuring something different and none of them a person.

Every checkThe whole glossary

Next

A supplier who cannot supply a base and a period has told you something, and it is not that they are disorganised. We do this for a living and the questions land differently when they come from outside.