Turning a raw directory into a ranked prospect list
The challenge
A remote-work directory holds hundreds of companies — some perfect-fit prospects, most noise. Reviewing them by hand means hours per batch, inconsistent judgment, and a list that's stale by the time it's finished.
The brief: turn 767 raw company entries into something an outreach effort can actually run on — ranked, reasoned, and verified.
The approach
- Fit criteria first. Before any code, the definition of a good prospect was written down in plain language — industry signals, size signals, remote-work posture — so the scoring logic was explainable, not a black box.
- A scoring engine, not a spreadsheet marathon. A Python engine scored all 767 companies against those criteria and sorted them into tiers.
- Live verification at the top. The highest-scoring targets were checked against their live web presence, so the top of the list could be trusted enough to act on immediately.
- A deliverable the client already knows how to use. Everything landed in a tiered Excel workbook — no new tool to learn, no dashboard to log into.
Outcomes
Why this matters
This is what "applied AI" means in practice: not a chatbot bolted onto a website, but judgment encoded into a system so it can run at a scale no human afternoon can match — while staying explainable enough that the client can defend every ranking.
The same pattern fits any repetitive evaluation work: vendor shortlists, applicant screening, content triage. If a person does it with a checklist, an engine can do it at volume.