All use cases
Inbound Lead Qualification
AI scores and routes 200+ inbound leads per week. SDR time on bad leads dropped 65%.
The problem
SDR team manually qualifying 200+ inbound leads per week. 70% turned out unqualified after initial outreach. Reps burning hours on leads that were never going to convert. Pipeline reviews were a graveyard of dead opportunities.
The result
SDR time spent on unqualified leads dropped 65%. Pipeline quality score improved 34%. Sales cycle shortened by 11 days on average for qualified leads.
The workflow
How the system actually runs.
- 01 New lead enters HubSpot via form submission or demo request
- 02 Clearbit auto-enriches with company size, industry, tech stack, funding data
- 03 Make sends enriched lead data to Claude with ICP scoring criteria
- 04 Claude scores lead 1–10 with written reasoning for each factor
- 05 Score 7+ auto-assigned to SDR. Score 4–6 enters nurture sequence. Below 4 archived.
Tools used
The stack behind it.
Claude APIMakeHubSpotClearbit
What worked
Why this setup held up.
- Combining Clearbit enrichment with Claude reasoning beat the old rule-based scoring by a wide margin
- Written reasoning for each score let sales managers audit and trust the system
- The nurture bucket (4–6) actually converted at 12% over 90 days — previously these leads were just ignored
What did not
The friction to watch.
- Claude initially scored too generously — had to add explicit negative signals and disqualification criteria
- Leads from new industries the model hadn't seen before got inconsistent scores
- Had to add a manual override flow for edge cases flagged by SDRs
Verdict
The short version.
If you're manually qualifying inbound leads, stop. This takes 2 days to set up and pays for itself within a week.