Case study · B2B Services
Smart lead-routing for a national B2B services firm
An AI-assisted lead-routing and qualification system that sorts every inbound enquiry by territory, deal size and source, drafts a first response in the right tone, and lands it on the right rep's desk, usually before they finish their coffee.
The challenge
What we had to solve.
Inbound web leads were landing in a shared inbox that 14 reps could see, and almost no one was treating as their own. Important enquiries sat for hours; cold-fits got chased while warm-fits went stale.
Worse, there was no signal in the lead record about why a rep should care: no territory match, no deal-size estimate, no source attribution. Every lead looked the same on arrival.
Sales leadership wanted speed-to-first-response under 10 minutes, but only on leads that actually deserved that urgency.
Our approach
How we tackled it.
We started with a two-week discovery: shadowed reps, tagged 200 historical leads by hand, and mapped what a "good" routing decision actually looked like.
Then we built a routing engine on top of the existing CRM. Leads come in through web forms, get enriched (firmographic lookup, source attribution, intent signals from page history), and a classifier scores them on fit and urgency.
An LLM drafts a first reply matched to the lead's industry and stated problem, using language pulled from the rep's own past winning emails, so the response sounds like the rep, not like a bot.
The rep gets a Slack ping with the lead summary, a one-click "send as-is" button, and an edit window. Sales leadership gets a dashboard of routing decisions, response times, and override patterns.
Deliverables
What we built.
Specific, named outputs, not vague "strategy".
Routing engineRules + ML scorer that classifies leads on fit, urgency and territory and assigns to the right rep.
AI reply drafterLLM-generated first-touch email matched to industry and intent, with one-click send.
Enrichment layerFirmographic + intent-signal lookup that fires before the routing decision.
Slack + CRM integrationInline notifications to reps and structured lead records with all scoring exposed.
Leadership dashboardRouting decisions, response times, override patterns and lead-to-deal attribution.
Outcomes
What it returned.
- Average first-response time fell from 4.2 hours to 1.3 hours, a 68% reduction, hands-off.
- Sales-qualified-lead rate went up 22% in the first quarter, with no extra ad spend.
- Reps reported "feeling on top of the inbox" for the first time in years, the morale unlock was as valuable as the metric.
- Override rate on AI-drafted replies stabilized at 31%, which gave leadership confidence in the system's judgement.
The takeaway
What we learned.
The fastest AI productivity wins come from removing a hand-off, not from removing a human. We didn't replace the rep's judgement, we just made sure they saw the right lead at the right time with most of the work already done.
Services used
How we got it done.
Related work