Multi-Channel Attribution for Indian Healthcare Clinics: From Last-Click to Patient Journey

The default attribution model for Indian healthcare clinics is last-click: whichever channel produced the last touch before consultation booking gets full revenue credit. This is wrong in ways that systematically distort budget allocation. Last-click:

• Undervalues YouTube (early-stage discovery)
• Undervalues organic search (mid-stage research)
• Overvalues Meta retargeting (last-stage closer)
• Overvalues branded Google search (intent already there)
• Makes WhatsApp invisible (often the final touch, but didn't generate the demand)

Real attribution requires multi-touch modelling that gives proportional credit across the full patient journey. This article covers what that looks like, why it matters, and how to implement it for Indian healthcare clinics.

Why last-click attribution is destructive

Consider a real patient journey ICG saw at a derm clinic:

1. Day 1: Patient watched YouTube video about hydrafacial (organic discovery)
2. Day 5: Patient googled "best dermatologist Gurgaon" (organic research)
3. Day 7: Patient clicked a Meta retargeting ad
4. Day 7: Patient clicked WhatsApp link, started consultation enquiry
5. Day 8: Patient booked consultation
6. Day 12: Patient visited, treatment quoted ₹85K

Last-click attribution gives WhatsApp 100% of the ₹85K credit. The clinic's marketing analyst sees this and concludes: WhatsApp drives revenue; YouTube doesn't.

Next budget cycle: clinic increases WhatsApp spend, decreases YouTube spend. Six months later: enquiry volume drops 25%; revenue drops 18%; analyst blames "WhatsApp burn" without understanding the actual cause.

The actual cause: YouTube was generating the awareness. Once YouTube spend dropped, fewer patients ever started the journey. Last-click made YouTube invisible until cutting it broke everything.

The healthcare patient journey

Indian healthcare patient journeys typically have 4-8 touchpoints across 14-90 days:

Awareness (touchpoints 1-3): YouTube, organic social, friend referral, Google organic search

Research (touchpoints 3-5): Hospital website, doctor profile pages, Justdial / Practo, ChatGPT / Perplexity

Comparison (touchpoints 4-6): Multiple clinic shortlist, cost comparison, doctor video research

Decision (touchpoints 5-7): Meta retargeting, branded Google search, WhatsApp enquiry

Action (touchpoints 6-8): Phone call, WhatsApp consultation, booking, visit

Last-click sees only the final 1-2 touchpoints. Multi-touch attribution sees the whole journey and credits proportionally.

Multi-touch attribution models

Linear attribution

Every touchpoint gets equal credit. Simple, fair, but undervalues high-conversion touchpoints.

Time-decay attribution

Touchpoints closer to conversion get more credit, but earlier touchpoints still get some. Most common middle-ground model.

Position-based attribution

First touchpoint and last touchpoint get most credit (e.g. 40% each), middle touchpoints share the rest. Useful for highlighting both demand generation and demand capture.

Data-driven attribution (Meta, Google native)

Platform-specific machine learning model that assigns credit based on observed conversion patterns. Best when there's enough data; not always available for healthcare verticals due to data restrictions.

ICG's custom model: Stage-weighted attribution

ICG's preferred model for healthcare: assigns credit based on which funnel stage the touchpoint moved the patient to.

• Touch that moved patient from Awareness → Research: 15% credit
• Touch that moved patient from Research → Comparison: 20% credit
• Touch that moved patient from Comparison → Decision: 25% credit
• Touch that moved patient from Decision → Action: 25% credit
• Touch that closed the booking: 15% credit

This model rewards both demand generation (the first touch that moved a stranger into Awareness) and demand capture (the final touch that closed the booking) without over-rewarding either.

Implementation requirements

Real multi-touch attribution requires:

1. Cross-channel identity resolution

The same patient touching YouTube + Google search + Meta + WhatsApp must be resolvable as one identity. Requires:

• First-party cookie tracking
• Cross-device matching (probabilistic at minimum, deterministic where possible)
• Email or phone hash matching across platforms

Beacon handles cross-channel identity resolution via hashed match keys.

2. Full-funnel event capture

Events captured at every touchpoint, not just the last:

• YouTube watch + engagement (via YouTube Studio API)
• Google search visit (via GA4 + GSC integration)
• Meta ad impression + click (via Meta CAPI)
• Website page view + time on page (via GA4)
• WhatsApp message (via Business API webhooks)
• Phone call (via call tracking integration)
• CRM stage progression (via Nexus CRM webhooks)

3. Attribution model engine

Code that ingests the events, resolves identity, applies the chosen attribution model, and outputs proportional credit per touchpoint per conversion.

4. Reporting layer

Dashboard that shows attribution by channel + by campaign + by specialty + by time period. Lets the analyst see what's actually working.

5. Budget feedback loop

Insights flow back into Meta/Google bidding, content investment, channel mix decisions. Without feedback loop, multi-touch attribution is just nicer reporting that nobody acts on.

How ICG implements

Across HealthApex OS engagements:

• Beacon handles cross-channel identity resolution + event capture
• Nexus CRM captures stage-progression events
• YODA captures YouTube watch + engagement events
• Agency OS dashboards present attribution by channel + campaign + specialty
• Quarterly attribution review (part of Client Alleviation Programme) feeds insights into budget decisions

Across deployments: budget reallocation from last-click to multi-touch typically produces 15-30% conversion lift on same total spend within 90 days.

Common implementation mistakes

1. Implementing attribution without identity resolution. Touchpoints captured but not connected to the same patient — produces meaningless data.

2. Using only platform-native attribution (Meta + Google). Each platform reports its own contribution. Combined, they over-count. Multi-touch needs unified attribution layer.

3. Setting up attribution without budget feedback loop. Beautiful dashboards that nobody acts on. The point of attribution is decision support, not reporting.

4. Switching models without explanation. Stakeholders see different numbers; lose confidence. Pick a model, document it, stick with it.

5. Trying to model everything. Some channels are very hard to attribute (offline events, family referrals). Acknowledge gaps; don't fake the numbers.

Related reads

• Nexus CRM
• Beacon — attribution engine
• Healthcare CRM India 2026 guide
• CRM + CAPI + GCLID integration
• HealthApex OS