Custom sales CRM with AI lead scoring for a UK financial services firm
A 40-person UK brokerage was running on spreadsheets — leads dropped, leadership had no pipeline visibility, and reps wasted hours each week on manual triage. We built a Laravel + React.js CRM with AI-powered lead scoring that lifted team productivity 50% and shortened time-to-close by 35%.
The challenge
A 40-person UK brokerage had no unified CRM. Sales reps tracked leads in spreadsheets, follow-ups slipped through email threads, and contracts were chased manually across DocuSign and Hubspot. Leadership had zero pipeline visibility and no objective way to prioritise which leads got the next call.
The cost wasn’t a missing tool — it was a missing system. Every rep had their own ranking heuristic, every region had their own template, and every quarter ended in a frantic spreadsheet reconciliation.
What we built
A purpose-built CRM that automates the entire sales lifecycle and ranks every inbound lead the moment it lands:
- AI-powered lead scoring that combines firmographics, intent signals, and prior conversation context into a single 0–100 score
- Pipeline management with stage-gated automation (auto-nudges, escalation rules, contract triggers)
- Automated follow-ups with templated cadences and per-rep override
- DocuSign-integrated contract generation — quote-to-signature in under 10 minutes
- Live leadership dashboard with rep, region, and product-line drill-down
AI scoring surfaces high-priority leads 3× faster than the manual review the team had been doing.
Architecture
We built on a Laravel + React.js stack with PostgreSQL and a Stripe billing layer:
- Backend: Laravel API with role-based access control and per-tenant data isolation
- Frontend: React.js SPA with optimistic updates and offline-tolerant queue
- AI/ML: OpenAI API for scoring + summarisation, with deterministic fallbacks for compliance review
- Integrations: DocuSign, Stripe, Calendly, Slack — plus webhook surface for future ones
- Audit trail: Every lead state change recorded with actor, timestamp, and reason — required for FCA reporting
Outcomes
- 50% productivity lift across the 40-rep sales team
- 35% faster average lead-to-close cycle time
- 3× faster prioritisation via AI scoring vs. manual review
- 3 enterprise wins closed in the first quarter post-launch
- Zero pipeline drop-offs since go-live (vs. ~12% pre-launch baseline)
- FCA-compliant audit trail on every lead interaction
Why it worked
Three deliberate calls made the difference:
- Score every lead, ship the score. The AI scoring wasn’t a back-office model that informed reports later — it was a number visible in the queue, ranked, and acted on. Reps trusted it because they could see the inputs.
- Automate boring, never automate judgement. Follow-up cadences, contract drafts, and nudges are automated. Discounting, escalation, and qualification stay with the rep. The brokerage’s reps stayed in control.
- One rep, one queue. No duplicate views, no unsynced spreadsheets. The dashboard a rep sees on their phone is the same dashboard leadership sees in the boardroom.
The engagement is now in its second year as a maintenance + AI evolution retainer.