Referral tracking and re-engagement

What automation does

AI automation from Grow2.ai takes on two related tasks: tracking the fate of every referral and systematically re-engaging contacts that have dropped out of active work. A referral is any inbound direction: a client from a partner, a patient from a doctor, a lead from word of mouth. Once a referral enters the system, automation tracks every step of their progress, records gaps, and launches touch sequences when movement stops.

Re-engagement is the second loop. Automation periodically scans the contact base, identifies those who have not interacted with the company beyond a set threshold, and launches a re-engagement campaign: a personalized message via the preferred channel, an offer to book an appointment or schedule a call, automatic transfer to the active pipeline on response.

What is included in the automation loop

1. Referral intake from a source (partner, website form, call, EHR, CRM).
2. Assignment of status, owner, and SLA for each stage.
3. Monitoring: if a stage does not advance within N days — an alert to the owner and a follow-up to the client.
4. Segmentation of the base by recency of last contact.
5. Launching a re-engagement campaign based on the preferred channel (email, SMS, messenger).
6. Moving respondents to the active pipeline and notifying the manager.
7. Logging all events for reporting and compliance.

What automation does NOT do: it does not sell instead of the manager, does not negotiate price, does not make diagnoses, and does not replace a doctor or consultant. The decision on the content of each touch is made by the operator using templates — the AI agent executes and escalates.

Typical configuration options

Solo (1–5 people). Minimal configuration: one inbound referral channel, one pipeline, basic monitoring with two rules — 'no movement for 7 days' and 'inactive for 60 days'. Re-engagement in one wave, one template. Integration with a calendar for automatic call booking. Setup takes a few days, including base export and first run. Suitable for solo-practice consultants, single clinics, coaches — those who close deals themselves and lose leads in the chaos of inbound.

SMB (6–30 people). Referral segmentation by source and service type, owner assignment by rules, SLA by stage. Re-engagement in 2–3 waves with escalation to a live manager. Integration with CRM or EHR, calendar, and communication channels. Reporting on conversion at each stage and ROI of re-engagement campaigns. Setup takes approximately one business week. Typical case: a 3-doctor clinic with a coordinator, a consulting team with account managers.

Enterprise (30+ people). Multi-tenancy by branch or business unit, complex distribution rules, integration with multiple systems simultaneously (CRM + ERP + billing + EHR), compliance mode (HIPAA, GDPR) with audit and encryption. Re-engagement campaigns are segmented by lifetime value, purchase history, and preferred channel. Setup takes 2–4 weeks accounting for approvals. A dedicated manager from Grow2.ai leads the project to stable operation and hands off to the internal team.

How it works

AI automation is split into three layers: data collection, rule logic, actions. Each layer is configured for a specific company — no code, in config.

Step 1. Connecting sources

Automation connects to the systems where referrals and clients already live: CRM, EHR (for clinics), calendar, communication channels (email, SMS, messengers, telephony). Grow2.ai uses standard APIs and webhooks — if you have a vertical SaaS with an open interface, integration is standard. If the system is closed, an intermediate connector is added.

Step 2. Rule configuration

The config describes the funnel stages (for example: «referral received → initial contact → booking → visit → post-visit»), SLA for each stage in days, responsible roles, communication templates by channel, and re-engagement segments. The AI agent does not invent these rules on its own — it executes what is defined. Rule complexity ranges from a few conditions to dozens.

Step 3. Monitoring and alerting

Every N minutes the system scans all active referrals and checks their status against SLA. Violations are classified by priority:

1. Critical — the referral is losing value (for example, insurance expires in 3 days and the visit is not booked). Alert to the responsible manager in Slack or via email + follow-up to the client.
2. High priority — the stage has not progressed beyond the SLA. Follow-up to the client using a template, copy to the manager.
3. Scheduled — reminder one day before the meeting, confirmation request.

Step 4. Re-engagement loop

In parallel with active funnel monitoring, a reactivation process runs for inactive contacts. Inactivity criteria — a parameter (30, 60, 90 days without contact). The selection is segmented, and each segment is assigned a series of messages. A client response triggers a transition to the active funnel and notifies the manager.

Step 5. Reporting

Automation records all events: sends, responses received, link clicks, visit bookings, cancellations, returns. From this, a dashboard is built: conversion by stage, fallthrough rate (failed conversions), ROI of re-engagement campaigns, average time-to-contact.

Alternative approaches

The manual approach breaks down at scale: with even dozens of active referrals per week, a person loses control and gaps start appearing. No-code tools like Zapier or a workflow engine cover standard scenarios but cannot handle segmentation, non-standard logic, and reporting — each new scenario requires additional setup, and maintenance becomes burdensome. AI automation from Grow2.ai is different in that it covers the entire loop at once: rules, monitoring, communication, escalation, reporting — in one system, without stitching together 5–7 different SaaS.

Security and compliance

For clinics, HIPAA (USA), GDPR (EU), and local medical regulations are critical. Grow2.ai configures automation taking these requirements into account:

1. Personal data does not leave the CRM or EHR perimeter without explicit approval.
2. Patient communication goes through approved channels with encryption in transit.
3. Event logs are stored in an auditable format with the operator indicated (human or AI agent).
4. Access to configuration is role-based.

For consulting, NDA and control over client information leakage are important — automation runs within the client's infrastructure or in an isolated environment, and data is not transferred to third parties without approval.

Prerequisites for implementation

AI automation is a layer on top of your system, not a replacement for it. For the project to launch in a week and deliver results, basic conditions on the client side are required.

Technical

1. CRM or EHR with API / webhook (or at least scheduled CSV export).
2. A calendar with the ability to programmatically create and reschedule appointments.
3. A client communication channel accessible via API: email service, SMS gateway, messenger provider, or telephony.
4. Admin rights for the client's contact person — to grant Grow2.ai access without waiting on the IT department.

Organizational

1. Documented stages of the referral or lead funnel. If no stages exist — we work through them in the first session, but this adds a few days to the setup.
2. A designated owner — a manager or coordinator who receives alerts and handles escalations. Automation works not "instead of people" but "with people".
3. Consent for processing client data — Grow2.ai provides consent templates.

Data

1. A contact database with up-to-date emails and phone numbers. Database quality is critical: outdated contacts → low response → wasted SMS budget.
2. At least 6 months of engagement history — for segmenting inactive contacts and calibrating rules.

Potential pitfalls

• Overly aggressive re-engagement. Frequent outreach to inactive contacts triggers complaints and unsubscribes. A conservative setup in the first 30 days reduces the risk.
• Templates without personalization. A generic "Hello, it's been a while" yields a low response. Templates are adapted to the segment and channel, not taken from defaults.
• No designated owner for alerts. If alerts go to a shared chat with no designated person, they get ignored. SLA is assigned to a specific role.
• Setup without relying on real data. Rules set arbitrarily produce either a flood of false positives or missed triggers. Before launch, we run the last 3 months of history through the rules and review what fires.
• Ignoring compliance requirements in the healthcare industry. SMS mentioning medical details is a violation. Communication over unencrypted channels is a violation. Segregating channels by data sensitivity is mandatory.