Years of paid leads often sit untouched in an aging CRM. An AI agent classifies thousands of dormant contacts, checks in over WhatsApp referencing each one's original search, and routes warm replies to agents, turning a dead database into new deals.
Typical results
3,000
Dormant leads worked
21%
Reply rate
140
Warm handoffs to agents
Figures reflect typical results for this kind of automation.
The challenge
The brokerage had paid for thousands of leads over a decade — portal inquiries, open-house sign-ins, old valuation requests — and nearly all went cold after one or two calls. Nobody had time to work 3,000 stale contacts by hand, a mass email blast years earlier had earned a 2% reply rate, and the data kept aging.
Before and after
Outreach reply rate
10.5×Deals from dormant leads
Live conversations
The automation
We built an AI reactivation agent on Make, OpenAI, and the WhatsApp Business API. It read every dormant record — searches, budgets, notes — classified each contact, and sent a personal check-in referencing what they originally looked for. The agent handled replies alone: updating preferences, answering questions, scoring intent. Genuinely warm conversations were handed to a human agent with full context and a suggested next step, and every outcome synced back to the CRM.
How it works
- 01
Classify the graveyard
AI reads each dormant record — searches, budgets, old notes — and segments buyers, sellers, renters, and dead entries.
- 02
Personal check-in
Each contact gets a WhatsApp message referencing what they originally looked for, drip-fed to protect number quality.
- 03
AI handles the reply
The agent answers questions, updates preferences, and scores intent before any human touches the conversation.
- 04
Warm handoff
Genuinely interested contacts land in an agent's pipeline with full context and a suggested next step.
- 05
Learn and repeat
Every outcome syncs back to the CRM, and the flow re-runs quarterly on whoever went quiet again.
Starts from
- 3,000 dormant contacts
- Original search history
Delivers
- Personalized WhatsApp check-ins
- Warm leads to agent pipeline
- CRM re-tagged and updated
The impact
On a database of about 3,000 dormant contacts, roughly 21% typically reply and around 140 turn into live conversations, which can lead to about nine closed deals within six months from records that produced nothing in years. Future sellers are tagged and nurtured, and the quarterly re-run keeps the CRM from decaying.
Questions about this automation
How does the agent decide what to say to each old contact?
It reads each dormant record first: the original search, budget, property type, area and any agent notes. It then classifies the contact as a buyer, seller, renter or dead entry, and writes a short WhatsApp check-in that refers to what that person originally asked about, so the message reads like a personal follow-up.
What happens when a contact replies?
The AI agent handles the first part of the conversation on its own. It answers simple questions, updates the contact's preferences and scores how serious the intent is. When a contact is genuinely interested, the conversation passes to a human agent with the full history, the updated criteria and a suggested next step.
How does the flow protect the WhatsApp number when messaging thousands of contacts?
Messages go out in small daily batches using approved WhatsApp Business API templates, never as one mass send. Contacts who ask to stop are removed and tagged in the CRM immediately. Batch sizes grow gradually while reply quality holds, and the whole run repeats quarterly for contacts who have gone quiet again.