How we measure, and what our measurement cannot tell you.
Every firm in this industry claims results. Almost none of them will tell you how they arrived at the number, which means the number is a matter of trust rather than fact. We would rather be checked than trusted, so this page explains exactly what we do.
You are welcome to take this method and use it yourself, or to hand it to another firm and ask them to be measured the same way. We would consider that a good outcome.
Shortlist Share.
The percentage of relevant AI answers, across an agreed question set and an agreed set of assistants, in which your company is named.
Named means named. Not mentioned in a footnote, not present in a linked source, not implied. If a buyer reading the answer would come away with your name, it counts. If they would not, it does not.
- 01The question set is agreed before the baseline is taken and does not change during the engagement. If it needs to change, we take a new baseline and say so.
- 02Every reading is an average across a window of at least four weeks. AI answers move week to week and a single reading proves nothing.
- 03The baseline is dated, recorded and shared with you before any work begins.
- 04We report the number that comes out, including when it goes down.
- 05We do not measure ourselves on anything we could influence by choosing what to measure.
Publishing the limitations is what makes the rest of this page credible rather than merely confident.
It cannot tell you how many people asked.
The assistants do not publish that and anyone who claims otherwise is estimating.
It cannot draw a straight line from a mention to an invoice.
Some of the value of being named never appears in your analytics at all, because the person who read it types your name into a browser a fortnight later.
It cannot promise the mechanism will hold.
These systems change how they choose sources, sometimes without notice. Our method is versioned for that reason and we will tell you when it changes.
It cannot make you a good answer to a question you are not a good answer to.
If the honest response is that you should not be on that shortlist, no amount of work will fix it and we will not take the engagement.
In an industry with no scoreboard, everybody says they won. We built a scoreboard, published how it works, and agreed to be judged on it.
A defined outcome, backed in writing.
On every ongoing programme, we commit to a measurable increase in your Shortlist Share — a minimum +15 percentage point improvement against your dated baseline within 120 days. If the re-measurement does not confirm that improvement, we continue working at no additional advisory charge until it is achieved.