Case study · B2B lead intelligence
1,555 prospects. One number that says who to call first.
A specialist UK CRM provider had a full prospect book and no way to rank it. The judgement lived in people's heads. We built a lead-intelligence system that scores every prospect from real company data and verifies every contact before outreach.
System
Lead-intelligence dashboard
Client
Specialist UK CRM provider
Scope
1,555 prospects · 2,043 contacts
Status
In production
Client and product names withheld. We name clients only with their written permission. Every number on this page comes from the production system.
The problem
The data existed: 1,555 prospect companies, 2,043 contacts, years of accumulated relationship history, all spread across CSV exports that couldn't answer the only question that matters: who's most likely to buy, and are we able to reach them?
Prioritisation was experience-based, which meant it lived in certain heads, didn't scale, and disappeared on holiday. The answer wasn't another CRM. The client sells one. What was missing was intelligence on top: scoring, enrichment and verification that turns a static book into a ranked pipeline.
What the system does
Real filings in. Ranked, reachable pipeline out.
Enriched from statutory filings
Each prospect is matched to its Companies House record and its filed accounts are read automatically, so scoring runs on real financials, not guesswork.
Scored, tiered, prioritised
An 8-factor model scores every prospect up to 130 points and sorts the book into hot, warm and cold. The sales question stops being "who's next?" and becomes "how fast can we work the hot list?"
An assistant that can't make things up
The team asks questions in plain English, like "which hot leads haven't been contacted this month?", and the assistant answers only from live pipeline data. If it can't trace an answer to a source, it refuses.
Outreach with guardrails
Personalised B2B outreach with daily send limits, unsubscribe handling and full delivery tracking built in. Volume discipline that protects the client's sending reputation.
The scoring model's factors and weights stay private. That's the client's edge. What we'll happily show you is the same approach pointed at your prospect book.
The numbers
1,555
prospects scored and ranked
The entire prospect book, scored on a 130-point scale and sorted into hot, warm and cold tiers, so the day starts with who to call, not with a spreadsheet.
130-pt
scoring scale across 8 factors
Company financials, sector fit, relationship warmth, network signals and more, combined into one number a salesperson can act on without needing to know how it's made.
SMTP
live handshake email verification
Every outreach address is confirmed against the receiving mail server before a single email is risked on it. A live check, not a guess.
Why it holds up in production
A sales team will only trust a score they can interrogate. Every figure traces back to its source: a filed account, a public record, a logged interaction. The AI assistant inherits the same rule, no answer without a source, ever. When the data can't support an answer, the system says so instead of improvising.
Outreach carries the same discipline. Send limits, suppression lists and delivery tracking were built in from the start. Those guardrails are what made it safe to let the system touch a real sending domain on day one.
Sitting on a prospect book like this?
See the fully autonomous version of this idea in the creator lead engine case study, or work out the right route for you in agency vs in-house vs no-code.