A before-and-after process built to separate movement from noise.

1. Freeze the questions

Each benchmark begins with 30 questions tied to the business’s market, service, customer, and buying situations. Twenty-eight unbranded questions measure discovery and recommendation strength. Two branded controls check whether the assistants state material business facts accurately.

2. Establish the baseline

The full founding benchmark tests three AI surfaces and repeats each prompt three times, creating 270 planned observations. Every run records the surface, time, recommendation position, competitor, citations, factual accuracy, and supporting response evidence.

3. Score what happened

The headline score combines prominence, recommendation stance, citation support, and buyer-intent weight. We separately report inclusion, top-three position, factual accuracy, competitor share, source patterns, and run-to-run agreement. A score is a sampled indicator—not market share, search volume, or guaranteed demand.

4. Change only what can be documented

We select a small number of changes, map each one to the buyer questions it could influence, and record the owner, affected page or source, and publication time. Keeping the intervention set small makes the learning more useful.

5. Rerun the frozen panel

The same questions are repeated after the changes have had a reasonable opportunity to become discoverable. We report gains, losses, unchanged answers, factual regressions, and volatile cells. Durable movement should survive more than one rerun.

What makes the service economically viable

• Customer value: the founder can identify consequential buyer decisions the business is losing and what should change first.
• Demonstrated movement: improvements survive repeated testing without creating factual errors.
• Time to signal: the first baseline and intervention cycle produces useful evidence quickly enough to guide action.
• Delivery economics: software and operator time leave a credible path to sustainable gross margin at the selected plan.
• Commercial outcome: AI-referred inquiries and qualified opportunities are tracked separately from the recommendation score.

Current validation stage

Good Productivity is in a founding-pilot stage. The methodology, free snapshot, email workflow, pricing boundaries, and measurement workbook are operational. Public performance claims will wait until a pilot improvement survives repeated measurement.

Claims we do not make

• We do not guarantee placement in ChatGPT, Google, Perplexity, or any other assistant.
• We do not present sampled prompts as total market demand.
• We do not treat one favorable answer as durable improvement.
• We do not attribute revenue to AI without a separate lead-source record.