A launch we can account for.
Baseline pending validationLast evidence review:
We are preparing our own launch using relayup. This page describes the measurement protocol. We have not yet published a validated baseline or claimed an improvement.
| Evidence | Status |
|---|---|
| Validated baseline | Not available |
| Comparable follow-up | Not available |
| Attributed customers or revenue | Not available |
What we will measure
- Technical website checks, separate from AI mentions.
- A fixed set of discovery questions without our brand name, separated by language and market.
- Successful answers and failures, with the model, date, search mode and original answer preserved.
- Enquiries, qualified opportunities and revenue, with the attribution method stated.
What counts as evidence
Failed calls do not count as an absence. Missing data does not become zero. A model without web search cannot validate that it has seen a recent website change.
Before and after comparisons require the same questions, model configuration, language and search conditions. Changes over time do not establish that relayup caused them. API results are observations, not a reproduction of every assistant app.
Preparation log
Prepared on 5 September 2026: Portuguese and English homepage, campaign labels on lead submissions, content and campaign drafts. Publication, campaign activation and the first measurement must be recorded separately once completed.