The problem this solves
A fit score is only useful when the person reading it can argue with it. A 74 next to a company name means nothing on its own; a 74 that says industry match, headcount inside band, region match, one of three technologies detected, no intent activity on <topic> in the last 30 days is a claim a rep can check in a minute and a RevOps lead can tune. That is the difference between a score people sort by and a score people ignore.
Most scoring lives in the CRM as a formula over whatever fields happen to be filled. Half the companies have no headcount, technologies were never recorded, and the score ends up rewarding data completeness more than fit. Scoring against live firmographics, technographics and intent data removes that bias: every company is evaluated on the same current evidence, and a blank means the evidence was looked for and was absent.
The ICP in your prompt has five dimensions - <industry>, <headcount>, <region>, <technologies> and intent on <topic> - and they are not equally important. Timing signals deserve weight because they change; firmographics deserve weight because they rarely do. The mission makes those weights explicit so the ranking reflects a decision you made rather than a default you inherited.
How the mission runs
- Turn the ICP into a scoring rubric. The Lead Verification Agent reads the five criteria and proposes a rubric: points per dimension, how partial matches are handled (a headcount just outside the band scores lower rather than zero), and how intent on <topic> is weighted against the static criteria. You accept the proposal or adjust the weights before any credits are spent.
- Refresh firmographics for every company. Industry, headcount, HQ region and revenue band are pulled from live company data for every row on list <name>, with waterfall enrichment filling the gaps: sources are tried in order until the field is filled, and you pay only for the source that answered. Companies whose domain cannot be resolved are held in a separate group with the reason.
- Check the technology stack. Each company is checked against technographic data for the technologies you named, and the row records which were detected, where and when they were last seen. A company running two of your three technologies scores on the two detected, and the third is stated as missing rather than assumed.
- Read the intent on <topic>. Intent data supplies the level of research activity on <topic> for each company over the recent window - surging, steady or none observed - and the rubric rewards recent activity. The window sits on the row so the freshness of the signal is visible to whoever sorts by it.
- Score, rank and write the reasons. Every company receives a 0-100 score, a reasons column listing each dimension with its result and points, and a top-reason column for a one-glance read. The list is sorted by score with ties broken by intent recency, and the scored columns are saved back to list <name> so the score persists and can be refreshed.
- Export the ranked list to Google Sheets. A sheet is created with the ranked rows, the reasons and the evidence columns, plus a second tab holding the rubric and the weights used, so anyone opening it later knows what the numbers mean. The sheet link is returned with the credit total for the mission.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
List <name> in the workspace, now scored 0-100 with a reasons column, a top reason and the underlying firmographic, technographic and intent evidence on every row, ranked highest fit first. A Google Sheet with the same ranked rows on one tab and the scoring rubric on another, so the weights are documented next to the results. Rows that could not be resolved to a company are listed with the reason, and the score can be refreshed later against the same rubric as the evidence changes.
Make it yours
- Score people instead of companies by adding title and seniority criteria, so a VP of <function> at a matching company outranks a manager at the same company.
- Ask for tiers as well as the number - A above 80, B from 60 to 79, C below - and route each tier differently in the next mission.
- Send the top 50 straight to HubSpot or Pipedrive as a reviewed batch, and keep the rest on the sheet for a later pass.
Frequently asked questions
Can I change the weights after seeing the results?
Yes. The evidence columns already sit on list <name>, so re-scoring with new weights is a recalculation over saved data rather than a fresh round of enrichment. Most teams adjust the weights once after reading the top and bottom twenty rows.
What happens when a company has no intent data?
It scores zero on the intent dimension and the reasons column says so explicitly. Intent is one dimension of five, so a strong firmographic and technographic match still ranks well; the rubric decides how much a missing signal costs.
How current is the technology evidence?
Each detection carries a last-seen date on the row, and the rubric can discount detections older than a threshold you set. The platform reports what was found and when, so a rep can judge whether a two-year-old detection still means the tool is in use.