Six markets, two languages, twenty agents. One system that never drops a lead.

The problem

Paid campaigns were performing. The system underneath them wasn't. Every lead, across six campaigns, two languages and four buyer regions, landed in the same undifferentiated queue. What happened next depended on whoever noticed first.

• Blind routing Leads arrived without a language or a project attached. An English-speaking enquiry and a Spanish-speaking enquiry for a different market looked identical in the CRM.
• Duplicates worked twice The same buyer enquiring on two campaigns got two agents, two follow-up calls, and one bad first impression.
• Stagnant pipeline A contact that arrived without an owner stayed ownerless until a human spotted it.
• Limited engagement window Follow-up stopped when the agent's day did. Anyone who didn't reply to the first message was, in practice, gone.
• Zero visibility Nobody above the agent knew a lead had landed until it showed up in a pipeline review.

Money was going into the top of the funnel faster than the funnel could handle it.

What we built

Seven interlocking workflows that take a lead from ad click to assigned agent to nurtured opportunity, without anyone touching it manually.

1. Campaign routing and enrichmentEvery inbound lead is identified by the form or lead ad that produced it, then tagged by market, language, and buyer region, with the project written to a custom field. Six campaign branches, each with its own segmentation path.
2. Duplicate detectionBefore a lead reaches an agent, the system checks it against four tag families for prior contact. Matches are flagged, the existing owner is notified by email and in-app, and the contact is pulled out of every competing workflow so it can't be worked twice.
3. Assignment and escalationLeads without an owner are routed by segment and budget band, with high-value enquiries flagged for senior agents. Assignment is mirrored onto the opportunity, so the whole pod sees it. A separate layer matches the assigned agent to their team and notifies the team lead within seconds.
4. Verification with split testingNew leads enter a four-variant SMS test. Replies are captured, the lead is marked verified, and the response is written back to the CRM. Non-responders move into a timed follow-up that respects business hours: messages queued after 9pm hold until 9am rather than arriving overnight.
5. AI nurtureVerified leads with a WhatsApp number enter an AI-generated nurture sequence. The system generates a message tailored to the contact, sends it, then waits up to three days for a reply. A reply moves the opportunity to Engaged Lead, updates the record, and notifies the agent. No reply triggers a second attempt, then a hand-off to long-term nurture.
6. Long-term nurtureA three-message sequence spread over several weeks, each message waiting a full week for a reply. Contacts who go the distance without responding wait thirty days and re-enter the sequence from the top. Leads don't get discarded; they get parked and picked back up.
7. Two-way integrationA bidirectional connection with the client's lead distribution platform. Qualified leads are pushed out by segment; inbound leads come back through a webhook, are matched to the right agent by name, and enter the same pipeline as everything else.

What changed

• Every lead gets an owner Assignment happens on arrival, by rule, not by whoever is looking at the CRM.
• Duplicates are caught before an agent picks up the phone Every duplicate the system flags is an agent-hour that isn't spent on a lead a colleague already called, and a buyer who doesn't get two versions of the same pitch.
• Follow-up outlives the agent's working day Verification, nurture and re-engagement run on their own schedule, in the contact's language, within sensible hours.
• Non-responders come back around Leads that go quiet aren't written off. They cycle back after thirty days.
• Team leads see it as it happens Escalation is automatic, so pod leads know about a lead before the pipeline review, not after.