← All work
Automation HubSpot Claude AI Paradyse Homes

Lead Decay Monitor

High-intent leads on a $199K+ product were going silent in reps' inboxes: no one watching, no system alerting. I built a five-status decay tracker that catches leads before they go cold and an AI queue that drafts the re-engagement message itself.

Auto
Detection: no manual monitoring
2
Systems integrated: HubSpot + Claude
01 · The problem

Hot leads were going cold, silently

Recency bias is baked into how sales pipelines work. The newest lead always looks most urgent. The result: a qualified lead from three weeks ago, still interested and just busy, slips out of sight while the rep calls fresh enquiries.

Before
  • No automatic detection of aging leads
  • Reps manually checked every contact to spot who'd gone quiet
  • High-intent leads treated same as low-intent in follow-up queue
  • Follow-up timing depended entirely on rep memory
  • No visibility across the team's pipeline health
After
  • System detects leads crossing decay thresholds automatically
  • Task + Slack alert fires the moment a lead starts to drift
  • Tier A leads escalated before they reach cold
  • AI drafts the re-engagement message, rep approves in one click
  • Dashboard shows every agent's pipeline health at a glance
02 · The system

No lead lives in two places at once.

Every open lead sits in exactly one status. HubSpot updates automatically. Reps know exactly where each lead stands without opening a single contact record.

Active
Contacted ≤7 days ago, or next step booked
No action needed. You're on it.
At Risk
8–14 days quiet, nothing booked
Task created. Tier A leads get a Slack alert.
Decaying
15–30 days quiet, nothing booked
Going cold fast. High urgency alert. AI queue drafts the message.
Dormant
30+ days quiet, nothing booked
One last personal attempt, then moved to nurture.
Parked
Future activity booked 30+ days out
Alert fires on the booked date. Silent until then.
03 · Watch it work

One lead, day by day

Step through a real-shaped lead as it moves through statuses, and watch the system catch it before it goes any colder.

James Whitfield · Prospect

Budget: $260K · Investment + own use · Visited Bali 2022
Active
Days since contact
0
last contacted
Lead status
Active
System action
Monitoring. No action needed.
Step 1 of 7
04 · Daily Workflow

The other half: an AI that drafts today's follow-ups before anyone logs in

Every morning, before the sales team logs in, it's already pulled today's priority leads and drafted what to say.

📋
Pull today's leads
20-30 per rep, ranked by fit & urgency
🧠
Read the history
CRM activity, notes, WhatsApp, email
👀
Agent reviews
Approve, or edit first
📤
Sends & logs
Only after approval
Today's queue · 17 leads due, ranked by priority
#LeadTierStatusLast contact
1James WhitfieldA Decaying16d
2Priya SharmaB At Risk9d
3Lars BergC Dormant34d
4Yuki TanakaB At Risk10d
5Amara OseiC At Risk8d
+ 12 more, queued below in priority order
AI-drafted follow-up
James Whitfield
CRM notes
Investment + own use, Seminyak. Budget $260K. Asked about ROI projections on day 4, no reply since.
WhatsApp history
Last message from James: "Thanks, will have a look." 16 days ago, unanswered.
Drafted message
Hey James, just checking in on the Seminyak opportunity we discussed. I know you mentioned weighing timing around your Bali visit. Has anything shifted? Happy to make it a quick call this week if that's easier.
05 · Results

What changed

15h+
Per rep, per week saved
No more manually scanning contacts for who's gone quiet. The system does that work automatically.
100%
Of aging leads detected
No lead can drift past a threshold without a task being created and the appropriate alert firing.
0
Cold-start messages written
Every re-engagement draft is pre-written by Claude with full context. Reps edit if they want, never start from blank.