In short
- No own name
- Your brand name appears in no question. Otherwise you measure whether the engine can read, and you know that already.
- Fixed composition
- The ratio between kinds of question is fixed. A set that grows each quarter with whatever went well measures nothing.
- Unknown is not zero
- An engine that did not search gets its own label. Otherwise a failure becomes a poor score.
How is a question set composed?
For a set of ten the distribution is fixed, and that ratio is why the outcome says something. A set consisting only of the third kind measures whether you exist among people who already know the trade word — that is the easy half, and the half most reports are built on.
The first kind is the hardest to win and the most important to measure: someone describes what they want to achieve without using a single word from your field. Whoever shows up there is found by people who were not looking for them.
| Kind of question | Out of ten | What it measures |
|---|---|---|
| need question | 4 | What the customer wants to achieve, in their words and without your jargon. The only kind that measures whether you are found by someone who does not know you. |
| comparative question | 2 | "Which is better, X or Y." This is where the choice is made, and where dropping off goes unnoticed. |
| category question | 4 | The question with the name of your trade in it. Important, and the least surprising: scoring here means scoring with people who already know the word. |
Why do you pick the hardest question?
Because otherwise the figure steers nothing. A score of ninety-one per cent achieved by putting your own name in the question gives you no indication of what to do — it is high and stays high whatever you undertake. A lower figure showing per question where you drop off is usable on the day you get it.
There is an uncomfortable side to it too, and it belongs here: a strict figure sells worse. That is why we explain it in a guide rather than leaving it out of the report.
What happens when a measurement fails?
Then it says we do not know, not that you were not there. An engine that did not search because of a failure or a limit gets its own entry and does not count in the average; the same goes for an engine with no reliable weight known in your market.
That makes coverage its own number beside the score, and it is the number saying what the score is worth. An excellent figure across thirty per cent of your market is worth less than a middling figure across ninety.
Can I recalculate your outcome myself?
Yes, and that is the intention. The question list is in the report, the engines are named, the weighting per market comes from a public source with a month attached, and the three outcomes per question are noted separately. Anyone asking the questions again should land nearby — not exactly, since answers vary, but within the same order.
Frequently asked questions
Can I add questions to the set myself?
Yes, and it is useful, because you know what your customers really ask while we are guessing. The only condition is that an added question then travels with every re-measurement — otherwise you compare two different sets and the movement between measurements means nothing.
Do you measure yourselves with this method too?
Yes, with the same questions and the same weighting, and the outcome goes on our own site as soon as it exists — including when it is poor. A measurement only applied to clients is not a method but a service.
Read on
Free, no account and no card.