Measure · Clio · In development

Clio

What AI was saying about you — then.

Clio establishes what AI systems were saying about a brand, or a whole sector, at a chosen moment in the past. Named for the muse of history, it gives the present a baseline: not only where you stand today, but where you stood — and the direction you have been travelling.

Register interest → How it works ↓
How it works — said honestly

We do not pretend to interrogate a past machine

We reconstruct the information environment as it was at the target date, then measure the perception it produces.

A brand improving is not the same as a brand that has arrived.

The method
We reconstruct the sources, the coverage and the record as they were at the target date, hold a current model against only that period-correct evidence, and measure the perception it produces. Where a reading is reconstructed rather than directly recorded, we say so, and attach the confidence to it.
Two depths
Index configuration — a periodic, sector-level time series. Audit configuration — deep and brand-specific, triangulating archives and reconstruction with per-finding confidence. Same engine, two depths.
What it returns

A line, not a dot.

One reading is a position. Two readings are a direction.

The number you have today means very little until you know which way it was travelling to get there.

The series
Period-correct ARES-C readings at the dates you choose, scored on the same instrument that produces the present-day reading — so the comparison is like for like rather than a change of ruler dressed up as a change of standing.
  • Scores at each target date, across the five ARES-C dimensions
  • The same named competitive cohort, held constant across the series
  • Confidence attached to every reconstructed reading, individually
  • The evidence base for each date, so a finding can be argued with
The reading
Where the line moved, and the events it moved around. A rebrand, a leadership change, a recall, a campaign, a competitor’s acquisition — Clio does not prove causation, but it establishes what the machine record looked like on either side of a date that mattered to you.
Where it is used
Establishing a baseline before an intervention, so its effect can later be judged. Testing whether a decline is recent or long-standing. Settling an internal argument about whether a position was ever held. Supporting a case that a competitor’s advantage is newer, or older, than assumed.
What we will not claim

The honest limits, stated first.

  • We are not querying a past model. The systems that answered in 2024 are gone, and the ones that replaced them cannot be made to forget. Anyone claiming otherwise is selling a reconstruction and calling it a recording.
  • We reconstruct the evidence, not the machine. Perception is modelled as a function of the information a system is standing on. Hold a current model against period-correct evidence only, and the reading it produces is a defensible estimate of what that period would have yielded.
  • Reconstructed readings are labelled as such. Every figure carries its confidence. Where the archive is thin, the band is wide, and we say the band is wide rather than quietly narrowing it.
  • The further back, the softer the reading. Coverage thins, archives lapse, and the record becomes patchier. There is a date beyond which we will decline to put a number on it at all.

These constraints are the reason Clio is worth having. An instrument that admits what it cannot see is the only kind whose readings can be relied on where it can.

Instrument ARES-C · Clio configuration · In development
In development

A brand improving is not the same as a brand that has already arrived.

Clio is how you tell them apart. Tell us you would like to be in the first cohort, and we will bring you in as the capability matures.

Register interest → Back to Measure →