Independent measurement, and nothing that depends on the result.

Every engagement is measurement against a public record. None is an optimization program, and none can be. That constraint is the product.

Engagements

Every engagement below includes a reading — a scheduled session in which we walk your team through the record. What each platform said, what the filing says, which findings are contradicted, which are unverifiable, and which are noise. Raw responses are not findings until someone who knows the instrument has read them with you.

The baseline

One firm. One or more questions grounded in a dated public source — a regulatory filing, a federal disclosure, or another verifiable account of the firm's work.

The questions are tested repeatedly across five AI platforms on a stated date.

You receive:

  • The questions and their source trail
  • Full transcripts
  • The firms recommended instead
  • The sources each platform cited or surfaced
  • Hand-verified findings
  • An explanation of what the evidence can and cannot support

Best for a firm that needs a defensible starting point, or an agency that needs an independent client read.

$5,000

One firm, delivered in two to three weeks, including the reading.

The cohort read

A defined market measured as a whole. Where possible the panel comes from a public register rather than a list someone assembled for visibility.

Each firm receives a question derived from its own record. The analysis shows the competitive pattern: which firms are found, which are recommended, which sources recur, and where the AI account differs from the documented market.

Best for an agency serving a vertical, or a firm that needs to see its result in context.

$15,000 to $35,000

A defined market, scaled to panel size. Quoted on the register and the screens.

The re-measure

The same questions and the same platforms, repeated at a stated interval after your team or agency has acted.

AIWorthy does not perform the work between measurements. That independence is what makes the comparison useful.

Best for testing whether an observable result changed, without asking the implementer to grade the implementation.

$2,500

The same questions and platforms at an agreed interval, including a second reading.

The firm record

For a firm already measured in a published AIWorthy panel. Your firm's complete record, extracted from the frozen corpus — every response, verbatim, scored against your filing — with the same reading. The measurement was completed before this was available to buy, and buying it changes nothing about what was measured.

Best for a firm that has seen a published finding and wants its own record in full.

$3,500

One firm, from an existing panel.

What we do not sell

No optimization program. No content production. No page building. No listings management. No retainer designed to raise a proprietary score.

Your agency or internal team decides what the evidence justifies, and performs the work. AIWorthy can measure again afterwards.

A measurement has a stake in the answer if the same party sells the remedy. That party can produce good work and still cannot be the one to certify it. The refusal is not modesty about what we can do — it is the condition that makes the measurement worth having.

How an engagement runs

  1. 1

    We agree on the question and its source

    You see the evidence behind the question before collection begins.

  2. 2

    We freeze the method

    The questions are fingerprinted before any results arrive.

  3. 3

    We collect on a stated date

    Five platforms, repeated asks, fresh sessions, documented conditions.

  4. 4

    We verify every finding by hand

    Automated extraction can flag a possible match. A human checks the raw response before it becomes data.

  5. 5

    You receive the evidence, not just the conclusion

    The deliverable includes the document, the transcripts, the sources, the competing recommendations, and the limitations.

  6. 6

    We read it with you

    A scheduled session with the people who need to act on it — or need it on file. We answer what the evidence supports and decline what it does not. What to change is your decision and your work; we can measure the result afterwards. Then we stop.

Two baseline studies are in publication, run to this method. Both panels, both screens, and the collection conditions are already published — including the questions as sent and the fingerprints of every frozen artifact.

See the panels and the method →

On being told the truth

If we find nothing, we will say so. If the sample is too small to support a conclusion, we will not dress it up as one. If a platform refuses to answer, we record a refusal — not an absence.

A credible measurement sometimes ends with "we do not know yet."

Tell me the firm and the market. I will tell you what the record allows us to ask.

hharreld@aiworthy.ai