How to measure thought leadership.
Almost nobody does, which is why almost nobody can defend the budget. This is the whole method, written down, including the parts that make it inconvenient.
Reputation is normally discussed as a feeling. Somebody senior says the market does not know who we are, somebody else says we should do more on LinkedIn, and a year later nobody can say whether it worked, because nobody wrote down where it started.
That is the actual problem. Not that thought leadership cannot be measured, but that it is almost never measured before the work begins, which makes any claim afterwards unfalsifiable.
What follows is how we do it. It is not the only defensible method, but it is a defensible one, and it is written down so that somebody else can check it.
Start with three questions, not one score
Everything worth knowing sits under three questions, and they fail in that order:
- Recognised. Do the right people know you exist? This drives inbound enquiry and job applications.
- Respected. Do they rate your judgement? This drives win rate and the price you achieve.
- Renowned. Do they come looking for you? This drives speaking invitations, media, and acquisition interest.
Three numbers, not one. A single blended figure with a decimal point implies a precision nobody has, and it hides the thing you most need to see, which is which of the three is actually broken.
Decide who you are being measured against, first
This is the most consequential decision in the exercise and it is usually skipped.
Before scoring anything, agree a list of 15 to 20 firms you genuinely compete with, and write down why each one might beat you. That list decides where the middle sits. Get it wrong, flatter yourself with a weak list, and every number afterwards is fiction.
Then 50 is the median of that list, not a pass mark. A 20 does not mean bad. It means most of the firms you lose to are ahead of you on that dimension. That distinction matters, because it turns an insult into a piece of commercial information.
Nine places to look, and 65 things to count
Six of the nine need nothing from the firm being measured, which is why it is possible to score a company that has never spoken to you.
What comes back for the company name, the leadership team's names, and the phrases buyers type when they do not yet know who to call.
Recorded, and weighted low on purpose. It is what everyone pictures when they hear reputation, and it is nine things out of 65.
Frameworks, original research, first-party data. Whether there is a position of their own, or a version of everybody else's.
Coverage across the previous twelve months, and whether journalists in the sector call them when they need a view.
Where they speak, what they judge, which boards and industry bodies they sit on.
Abandoned profiles, stale claims, the search results nobody wants found. Cheap to fix and routinely overlooked.
Inbound share, cycle length, win rate, price achieved. Their systems, so this one needs them in the room.
Direct applications, agency spend, offer acceptance, and what candidates knew before they applied.
Clients, the buyers who chose somebody else, recent joiners, and the people who turned the job down. The part no software will ever do, and the part that decides what to do next.
The weighting is the argument. Social is nine signals out of 65 because it is the most visible and least reliable evidence of standing. The interviews are seven, and they change more decisions than the other 58 combined.
Turning a finding into a number
Three routes, and which one applies is decided in advance rather than afterwards.
Against rivals
The same measurement run across all 15 to 20 competitors. Whoever lands in the middle gets 50. Used wherever there is a real number to compare.
Against a written guide
Where there is judgement rather than a number, what a 0, a 50 and a 100 look like is written down before scoring starts. Two analysts working separately should arrive at the same place. If they do not, the guide is not good enough yet.
Against your own past
For figures out of internal systems, where no competitor equivalent exists. Record today, measure the movement, and never dress it up as a comparison against rivals, because it is not one.
What should never be counted
Every one of these was considered and rejected. Somebody always asks about at least one, so here are the answers in advance.
- Follower count as a headline. Recorded, scored low. Forty thousand followers who will never buy are worth less than nine hundred who might.
- Software that grades coverage positive or negative. Not accurate enough on a sample this small. Where it matters, read the articles.
- One blended score. See above. Three numbers next to a named competitor beats one number with a decimal point.
- Anything you could not measure again in ninety days. If it cannot be shown moving, it cannot justify a monthly fee.
- Awards that are paid to enter. Record them, then score on whether anybody actually judged the entry.
- Anything you cannot help fix. A bad score with no work attached is just an unpleasant observation.
Then re-score, on a schedule, forever
Quarterly. Same 65 signals, same order, same written rules, same competitor list. Record every source so somebody else can check the working.
Measurement that happens once is a report. Measurement that happens every quarter is an instrument, and only the second one tells you whether any of the work is doing anything.
The honest limit
Reputation cannot be attributed as cleanly as paid media, and any firm claiming otherwise is selling a dashboard. What is defensible is this: measure the leading indicators rigorously, correlate them against commercial outcomes, and be explicit about the difference between correlation and cause.
That candour is worth more than a confident number that is wrong, and it is the reason the method above is published rather than kept behind a login.
We will run this on your firm.
Including the competitor list, which is the part that decides everything else. You get three numbers next to a named rival, and an estimate of what the gap costs you a year.