BUSINESS EXPERIMENTS · 2026 · SAMPLE BUILT · VERIFIED CUSTOMERS 0
I tried turning a site scan into a report that still left room for human judgement. Automated checks gathered repeated signals; a final manual pass decided which issues belonged in the pack and how they should be prioritised.
The white-label sample and service structure were built. Messages from several sales experiments were recorded together, however, so I cannot establish how many prospects received this particular pack. Verified customers and revenue were zero.
A scan was not yet a diagnosis
A long export of automated findings looked substantial but did not tell a reader what deserved attention first. The sample grouped an observation with its evidence, likely consequence and next action. Items that could not be judged confidently were left for manual review.
That final review also meant the service was never fully unattended. It did, however, put a boundary around false positives and findings stripped of context. As an internal artifact, that division of labour was the clearest part of the work.
The outreach ledger could not support a count
Other product and job-related outreach happened in the same period. Delivery, bounce and reply events were not fully separated by campaign, so I do not publish a number for diagnostic-pack proposals. A general lack of response is not turned into evidence that this offer was tested properly.
A cleaner test would begin with one rights-safe public site in a narrow industry, then measure the report scope and the time required for manual review. The surviving result is a sample pack and a service design, not a sale.
Project record
Period: 2026
Role: service scope, sample-report design, and the boundary between automated checks and manual QA
Activity: built a sample white-label site-diagnostic pack
Internal evidence: sample pack and service-structure documents
External validation: outreach and response attribution is unresolved
Final outcome: customers 0; revenue 0