ChatGPT Ads
India Playbook
2026

Table of Contents

• 1. Why ChatGPT Ads Matters for Indian Healthcare04
• 2. How the Auction Actually Works06
• 3. Conversation-Completion Attribution, Explained08
• 4. The AI Assistant Channel in GA4 — Reading the Signal10

• 5. Intent-Bucket Framework for Healthcare Buyers12
• 6. Intent-Bucket Templates (Copy-Ready)14
• 7. Specialty & Segment Playbooks18
• 8. City-Tier Budget Allocation21

• 9. The 12-Week Engagement Playbook23
• 10. Weekly Action Cadence Table26
• 11. Compliance Checklists (MCI/NMC-aligned)29
• 12. Reporting Cadence & Dashboard Design31
• 13. Worked Examples — Three Category Clients33

• 14. Glossary36
• 15. Resources & Templates38
• 16. Contact & Engagement Tiers39

Why ChatGPT Ads Matters for Indian Healthcare

ChatGPT Ads went live in India in mid-2026 as a conversational, sponsored-response advertising surface layered inside a chat interface that hundreds of millions of Indians now use as a first research stop for medical decisions — before a Google search, before a hospital website visit, sometimes before a phone call to a clinic. For healthcare marketers this is not an incremental channel bolt-on. It is a structural shift in where the first meaningful brand impression happens, and it rewards a different kind of preparation than paid search or paid social ever did.

Three forces make this urgent for hospitals, diagnostic chains, fertility centres, dental groups, aesthetic clinics and healthcare D2C brands operating in India today:

1.1 The research-before-search shift

A patient's family member investigating a knee replacement, a couple exploring IVF, a corporate HR head comparing annual health check packages for 400 employees — all three increasingly open a conversational assistant before they open a search engine tab. They ask multi-turn, follow-up-laden questions ("what's the difference between a cemented and uncemented replacement, and which one do doctors in Delhi generally recommend for a 62-year-old") that a ten-blue-links search result cannot answer in one shot but a conversational assistant handles naturally across three or four turns. A sponsored response inserted into that conversation — correctly targeted, compliantly worded — reaches the buyer at the exact moment their mental model of the decision is forming, which is earlier and more influential than the moment they type a query into Google.

1.2 The auction is intent-first, not keyword-first

Google Ads and Meta Ads both ultimately resolve to a keyword or an audience segment. ChatGPT Ads resolves to an inferred intent bucket derived from the shape of the conversation itself — the entities mentioned, the qualifying language used, the turn count, and signals like location and device. This means the same campaign structure that won on Google Ads (broad match + smart bidding) does not transfer. Healthcare advertisers who treat ChatGPT Ads as "just another paid channel" and copy-paste their Google Ads account structure consistently underperform advertisers who rebuild campaigns around intent buckets from scratch. This playbook exists to give you that rebuilt structure.

1.3 The data already shows a category-leading channel

Across ICG's own healthcare-marketing property, the GA4 "AI Assistant" channel — sessions arriving via conversational-AI referral and sponsored placements — converts at a 10.49% key-event rate. That is roughly double the organic-search key-event rate and roughly seventeen times the direct-channel rate on the same property, measured over a rolling 90-day window. This is not a small-sample anomaly; it reflects a simple mechanic: users arriving from a conversational assistant have already had their objections partially pre-answered by the assistant itself, so they arrive as warmer leads than users arriving from a ten-blue-links click.

1.4 What "getting it right" looks like

Getting ChatGPT Ads right in the Indian healthcare context means four things done together, not in isolation: (1) intent buckets mapped to the real multi-turn questions Indian patients and caregivers actually ask, in the phrasing they actually use — English, Hinglish, and regional- language-influenced English; (2) sponsored copy that reads as helpful within a conversation rather than as an interruption, because the assistant's own tone sets the bar for what feels native; (3) a compliance layer that respects MCI/NMC advertising restrictions on outcome claims, before-after claims, and patient testimonials, applied with the same rigour a hospital's own medical-legal team would apply to a hoarding or a newspaper ad; and (4) attribution built around conversation completion rather than click-through, because a large share of conversational-ad value shows up as an assisted, delayed conversion that click-based attribution simply cannot see.

The rest of this playbook works through each of these in the order a 12-week engagement actually runs: foundations, then strategy, then week-by-week execution, then the compliance and reporting scaffolding that keeps the programme defensible to a hospital's board and to regulators alike.

How the Auction Actually Works

Unlike a search auction, where a discrete query triggers a discrete ad slot, the ChatGPT Ads auction resolves against a rolling model of the conversation. Understanding the mechanics changes how you write copy, how you structure campaigns, and how you set bids.

2.1 The intent-bucket resolution step

As a conversation unfolds, the platform continuously re-estimates which of a finite set of commercial intent buckets the user's questions belong to. A conversation that opens with symptom description ("my father has been having chest pain on exertion for two weeks") resolves toward a diagnostic/consultation-seeking bucket; a conversation that opens with comparison language ("which hospital in Gurgaon is best for cardiac bypass") resolves toward a provider-comparison bucket; a conversation that opens with logistics ("how much does a bypass cost in India, does insurance cover it") resolves toward a cost-and-insurance bucket. Advertisers bid at the intent-bucket level, not the keyword level — this is the single biggest structural difference from Google Ads that healthcare marketers need to internalise.

2.2 Auction inputs

Note the absence of a pure click-through-rate input. A response that gets "clicked" a lot but leads nowhere degrades an advertiser's completion-quality score over time — which is a deliberate design choice that rewards advertisers who write honest, useful sponsored responses over advertisers who write clickbait-style copy. Healthcare advertisers, who are used to CTR-optimised search ad copy, need to unlearn that instinct here.

2.3 Bidding models available

• Cost per qualifying conversation (CPQC) — you pay only when a conversation reaches a defined completion event (e.g., the user asks for the clinic's phone number, requests a callback, or clicks through to a booking page). This is the recommended default for hospital and clinic advertisers because it aligns spend directly with lead-quality outcomes.
• Cost per response shown (CPRS) — you pay for impression-equivalent placement regardless of completion. Useful for top-of-funnel brand-presence campaigns (e.g., a diagnostic chain establishing category presence in a new city) but riskier for lead-generation budgets.
• Value-based bidding (VBB) — available once an account has at least 90 days of conversion history; lets the platform bid toward a target return using historical completion- to-revenue data you supply. Recommended only after Weeks 1–6 of the engagement (see Chapter 9).

Conversation-Completion Attribution, Explained

Conversation-completion attribution credits a conversion to a sponsored response if the completion event occurs any time the conversation is judged to be a continuous thread — even across multiple sessions, multiple days, and a device switch — as long as the platform can tie the thread back to the same underlying assistant conversation. This is fundamentally different from last-click attribution and requires a different mental model for hospital marketing teams used to Google Ads' click-window logic.

3.1 Why this matters more in healthcare than in most categories

Healthcare decisions — especially high-consideration ones like a surgery, a fertility treatment, or a long-term diagnostic relationship — are rarely decided in a single sitting. A caregiver typically researches across three to seven days, often across two or three devices (phone during the day, laptop in the evening), often looping in a second family member partway through. Click-based attribution systematically undercounts healthcare conversions because the device or session that completes the booking is frequently not the device or session that saw the ad. Conversation-completion attribution — because it follows the conversational thread rather than the click — recovers a meaningfully larger share of true conversions, which is part of why the GA4 AI Assistant channel shows the elevated key-event rate discussed in Chapter 1: it is not purely that the channel converts better, it is also that it is being measured more completely.

3.2 The completion event ladder

Define completion events in a ladder from soft to hard, and configure GA4 to capture all of them as distinct key events so your reporting (Chapter 12) can show the full funnel rather than only the bottom rung:

1. Soft completion — user asks a qualifying follow-up question after the sponsored response is shown (e.g., asks for the doctor's name or clinic timings).
2. Mid completion — user requests contact information or clicks through to the advertiser's booking page or WhatsApp link.
3. Hard completion — user completes a form fill, books an appointment slot, or initiates a WhatsApp conversation that a human on the advertiser's side responds to.
4. Business completion — the lead is qualified by the advertiser's front desk or call centre as a genuine appointment or admission enquiry (fed back via offline conversion import — see Chapter 12).

The AI Assistant Channel in GA4 — Reading the Signal

GA4's default channel grouping surfaces conversational-AI referral traffic under an "AI Assistant" channel once your property has enough session volume from qualifying referrers. Reading this channel correctly — and not confusing it with generic Google Search Console AI Overview reporting, which measures a completely different surface — is essential before you can trust any ChatGPT Ads reporting downstream.

4.1 AI Assistant channel ≠ AI Overview

This distinction trips up nearly every marketing team new to the space, so it is worth stating plainly: Google Search Console's "AI Overview" reporting shows impressions and clicks where your organic content was cited inside a Google AI Overview summary. It has nothing to do with ChatGPT, and nothing to do with paid conversational advertising. GA4's "AI Assistant" channel, by contrast, is a session-level channel grouping based on referral source and campaign parameters, and it is where both organic conversational referrals and paid ChatGPT Ads sessions land (paid sessions are further split out via campaign/source/medium parameters you set on your tracking links — see the UTM convention in Chapter 12). Never present AI Overview clicks as ChatGPT Ads performance in a client-facing report; the two are unrelated surfaces measuring unrelated things.

4.2 Reading the 10.49% figure correctly

The 10.49% key-event rate referenced in Chapter 1 is a blended figure across ICG's own healthcare-marketing property's AI Assistant channel over a rolling 90-day window, inclusive of both organic conversational referral and paid ChatGPT Ads sessions. It should be read as a directional benchmark — "this channel behaves like a warm, pre-qualified channel" — not as a guaranteed number for every advertiser's account. A hospital account in its first three weeks, still building bucket-level completion history, will typically show a lower rate that climbs toward benchmark by Week 6–8 as the auction's compliance-confidence and completion-quality scores mature (see Chapter 2.2).

Figures illustrative of ICG's own property benchmark set; individual advertiser results vary by specialty, city tier, and account maturity.

4.3 What to set up in GA4 before you spend a rupee

• Confirm the AI Assistant channel grouping is active and correctly picking up your conversational referral traffic (check Explore > Traffic Acquisition, filter to the channel).
• Define all four completion-ladder events from Chapter 3.2 as GA4 key events, not just the hardest one.
• Set up a dedicated UTM convention for every ChatGPT Ads campaign — source=chatgpt, medium=paid-conversational, campaign=<bucket-name> — so paid and organic AI Assistant traffic can be split in every report.
• Connect offline conversion import (call-centre qualification data) on at least a weekly cadence so Business Completion (Chapter 3.2, rung 4) is measurable, not just a soft signal.

Intent-Bucket Framework for Healthcare Buyers

ICG's account structure for Indian healthcare advertisers organises every campaign around six core intent buckets, developed from the 250 URLs of ChatGPT Ads-adjacent content ICG has already published and validated against real conversational query patterns across pillar, segment, specialty and city-level content. Every hospital, clinic, diagnostic chain or D2C healthcare brand's account should map to some subset of these six, weighted by specialty and buyer profile.

5.1 Mapping buckets to specialty

Not every specialty distributes evenly across the six buckets. Cardiac and oncology skew heavily toward Symptom & Diagnostic and Second Opinion & Escalation — these are fear-driven, urgency-inflected categories where trust and speed dominate the copy. Fertility and aesthetic categories skew toward Procedure Explainer and Cost & Insurance — these are consideration-heavy, research-driven categories where detail and transparency dominate. Diagnostic chains and annual health check providers skew toward Corporate & Bulk and Cost & Insurance. Dental and orthodontic groups skew toward Provider Comparison and Cost & Insurance. Set initial bucket budget-weights per specialty before launch, then let performance data reweight from Week 4 onward (Chapter 9).

Intent-Bucket Templates (Copy-Ready)

Below are working copy templates for each of the six buckets. Every template follows the same three-part discipline required to pass both the platform's conversational-relevance scoring (Chapter 2.2) and ICG's internal compliance review (Chapter 11): (1) it continues the conversation's tone rather than interrupting it, (2) it makes one factual, verifiable claim rather than a superlative, and (3) it offers a next step rather than a hard sell.

6.1 Symptom & Diagnostic bucket

6.2 Provider Comparison bucket

6.3 Cost & Insurance bucket

6.4 Procedure Explainer bucket

6.5 Corporate & Bulk bucket

6.6 Second Opinion & Escalation bucket

Specialty & Segment Playbooks

Each specialty carries its own dominant bucket mix, seasonal pattern, and language register. This chapter gives a working playbook for the specialties ICG most frequently manages for Indian hospital and clinic clients.

7.1 Cardiac care

Dominant buckets: Symptom & Diagnostic (40%), Second Opinion & Escalation (25%), Cost & Insurance (20%), Provider Comparison (15%). Cardiac conversations are fear-driven and often initiated by an adult child researching on behalf of a parent — copy should acknowledge the caregiver, not only the patient. Peak conversation volume tracks with post-festival-season dietary excess (late October through December) and with heatwave months (April–June) when cardiac events spike. Bid up 15–20% in these windows.

7.2 Fertility & IVF

Dominant buckets: Procedure Explainer (35%), Cost & Insurance (30%), Provider Comparison (25%), Symptom & Diagnostic (10%). This category is the most detail-hungry — couples read extensively before a single consultation. Long-form procedure explainer landing pages with success factor transparency (without numeric outcome claims — see 6.6 compliance note) consistently outperform short pages. Avoid any messaging that implies guaranteed success; NMC guidance and advertiser self-regulation in this category are unusually strict.

7.3 Dental & orthodontic

Dominant buckets: Provider Comparison (40%), Cost & Insurance (35%), Procedure Explainer (25%). Highest price-sensitivity of any specialty in this playbook; cost transparency on the landing page (not necessarily in the ad copy itself) is the single biggest lever on completion rate. EMI and no-cost-EMI messaging performs strongly.

7.4 Aesthetic & dermatology

Dominant buckets: Procedure Explainer (45%), Provider Comparison (30%), Cost & Insurance (25%). Before/after imagery is heavily restricted in advertiser policy for this category — templates should lean on process explanation and credential transparency rather than visual transformation claims.

7.5 Diagnostics & annual health checks

Dominant buckets: Corporate & Bulk (35%), Cost & Insurance (35%), Symptom & Diagnostic (30%). The only specialty in this set where B2B (HR/admin persona) conversations regularly exceed 30% of volume — see Chapter 6.5 template and budget a distinct bucket weight for it rather than folding it into consumer campaigns.

7.6 Ayurveda & integrative wellness

Dominant buckets: Procedure Explainer / treatment-explainer variant (50%), Provider Comparison (30%), Cost & Insurance (20%). This segment carries its own compliance layer beyond standard MCI/NMC guidance — AYUSH advertising rules apply, and outcome-claim restrictions are, if anything, stricter. Always route Ayurveda accounts through the AYUSH-specific compliance checklist appendix (Chapter 11.4) in addition to the standard one.

City-Tier Budget Allocation

Conversational-AI adoption in India is not uniform across city tiers, and neither is the competitive density of the ChatGPT Ads auction. ICG's recommended starting allocation, refined across the 140 city×specialty programmatic pages already live on the ICG content platform, is shown below. Treat this as a Week 1 starting point, not a fixed rule — reweight from Week 4 using actual bucket-level completion data (Chapter 9).

The 12-Week Engagement Playbook

This is the backbone of every ICG ChatGPT Ads engagement, organised into four three-week phases. Each phase has a single primary objective; resist the temptation to chase secondary objectives early, since intent-bucket accounts need clean, uncontaminated data in the first phase to make good decisions in the second.

Phase 1 — Weeks 1–3 · Foundation & Launch

Objective: get a compliant, correctly bucketed account live and gathering clean data. No optimisation decisions are made from Phase 1 data — it exists purely to seed the auction's understanding of the account.

• Week 1: Compliance review of all claim language (Chapter 11); GA4 key-event and UTM setup (Chapter 4.3); landing page audit against the six intent buckets.
• Week 2: Campaign build — one campaign per intent bucket, conservative CPQC bids, City-tier split per Chapter 8; internal QA pass on every sponsored response against the templates in Chapter 6.
• Week 3: Launch; daily spend-pacing checks only (no bid changes); begin the weekly reporting cadence (Chapter 12) from day one so the client sees data immediately, framed clearly as "learning phase, not yet optimised."

Phase 2 — Weeks 4–6 · Signal & Reweight

Objective: let real completion data reweight bucket budgets and bids for the first time.

• Week 4: First reweighting pass — shift budget toward buckets showing above-benchmark soft- and mid-completion trendlines (Chapter 3.2), even before hard completions catch up.
• Week 5: Raise CPQC ceiling specifically on the Second Opinion & Escalation bucket if data supports it (this bucket converts fastest in almost every account ICG has run).
• Week 6: First graduation candidates identified for Value-Based Bidding (Chapter 2.3) — typically the one or two buckets that have crossed 150 qualifying conversations.

Phase 3 — Weeks 7–9 · Scale & Specialise

Objective: scale what is working, layer in specialty-specific refinements (Chapter 7), and begin testing city-tier reallocation.

• Week 7: VBB graduation for qualifying buckets; new copy variant testing within top- performing buckets (2–3 variants per bucket, not more — small accounts don't have volume to support wider variant testing).
• Week 8: City-tier reweighting using six weeks of tier-level CPQC and completion data; typically this is when a Tier-1 city outperforms the Metro tier on efficiency and earns a budget shift.
• Week 9: Corporate & Bulk bucket (where applicable) gets a dedicated push if Q4/Q1 corporate wellness budget cycles are approaching — this bucket has a strong seasonal calendar tied to corporate financial years.

Phase 4 — Weeks 10–12 · Prove & Plan

Objective: consolidate a defensible ROI story for the client's leadership and build the next-quarter plan.

• Week 10: Business-completion reconciliation (Chapter 3.2, rung 4) against front-desk/call- centre qualification data — this is the number that goes in front of a hospital board, not the platform's own completion count.
• Week 11: Full 12-week readout deck — bucket-level CPQC trend, city-tier efficiency, channel comparison against Google/Meta on the same property (Chapter 4.2 benchmark table as template).
• Week 12: Next-quarter plan — typically a budget increase recommendation for the buckets that graduated to VBB, plus a specialty-expansion recommendation if the account covers a multi- specialty hospital that only launched with 2–3 specialties in Phase 1.

Weekly Action Cadence Table

A full-page reference table for the account manager running the day-to-day of the engagement. Use this as the literal weekly checklist for the 12-week programme in Chapter 9.

Compliance Checklists (MCI/NMC-aligned)

Every sponsored response written for an Indian healthcare advertiser passes through the following checklist before it goes live. This checklist is deliberately stricter than the platform's own automated compliance-confidence scoring (Chapter 2.2) because ICG holds advertiser copy to the same standard a hospital's medical-legal team would apply to any public advertisement, in line with National Medical Commission (NMC) guidance on medical advertising and professional conduct.

11.1 Universal checklist — every sponsored response

• No outcome, cure, survival-rate, or success-rate percentage stated, implied, or referenced.
• No superlative claims ("best," "No. 1," "leading") without a currently valid, citable, third-party accreditation backing the specific claim.
• No direct or indirect disparagement of a named or identifiable competing hospital, clinic, or doctor.
• No patient testimonial, before/after reference, or implied patient endorsement inside the ad copy itself.
• No urgency-manufacturing language ("book now before it's too late," "limited slots remaining") applied to a medical decision.
• Every named doctor or specialist referenced holds a currently valid, verifiable registration; credentials are not overstated (e.g., "specialist" is not used for a general practitioner).
• Pricing, if referenced at all, points to a landing page kept current rather than stating a fixed number in the ad copy.
• Every claim is traceable to a source document the advertiser has supplied and signed off on — no claim is written from assumption.

11.2 Fertility & IVF-specific additions

• No implied or stated success/pregnancy rate, even in ranges.
• No language implying guaranteed outcomes ("achieve your dream of parenthood") without a clear, prominent qualifier about individual variation.

11.3 Aesthetic & dermatology-specific additions

• No before/after visual reference of any kind in ad copy, even described in text.
• No comparative body-image or appearance-shaming language framing the "before" state negatively.

11.4 AYUSH / Ayurveda-specific additions

• No claim positioning an Ayurvedic treatment as a replacement for, or superior to, a conventional medical treatment for a diagnosed condition without appropriate qualification.
• Practitioner credentials referenced must be AYUSH-registered and verifiable.
• No disease-cure claims for conditions classified as requiring conventional medical management under applicable Drugs and Magic Remedies (Objectionable Advertisements) Act guidance.

Reporting Cadence & Dashboard Design

12.1 Cadence

12.2 UTM convention

Every ChatGPT Ads link uses: utm_source=chatgpt&utm_medium=paid-conversational& utm_campaign={bucket-name}&utm_content={city-tier}. This convention is what allows the weekly report to split organic AI Assistant traffic (Chapter 4.1) from paid ChatGPT Ads traffic cleanly, and it is what allows Chapter 10's weekly cadence table to be built directly from GA4 Explorations rather than manual spreadsheet reconciliation.

12.3 The one number that matters to leadership

Every monthly and quarterly report leads with Business Completions (Chapter 3.2, rung 4) — reconciled against the advertiser's own call-centre or front-desk qualification data — not the platform's raw completion count. Platform completion counts are useful for the account manager's weekly optimisation work; they are not the number a hospital CFO should be asked to trust, because they do not yet reflect whether a completed conversation became a real, qualified enquiry. Getting this reconciliation running by Week 10 (Chapter 9) is a hard requirement of every ICG engagement, not an optional add-on.

Worked Examples — Three Category Clients

The following are hypothetical, category-representative composites built to illustrate the playbook end to end. Names and identifying details are illustrative, not drawn from any single real client.

13.1 A 300-bed multi-specialty hospital, Metro Tier

13.2 A single-city fertility clinic, Tier-1 city

13.3 A regional diagnostic chain, Tier-2 expansion

Glossary

AI Assistant channel (GA4)

GA4's default channel grouping for sessions arriving via conversational-AI referral or paid conversational campaigns, distinct from AI Overview reporting in Search Console.

Business Completion

The fourth and highest rung of the completion ladder — a lead qualified as genuine by the advertiser's own front desk or call centre, reconciled via offline conversion import.

Compliance confidence score

An auction-ranking input reflecting the platform's automated and policy-team assessment of an advertiser's claim safety.

Completion ladder

The four-rung framework (soft, mid, hard, business) used to define and report conversation-completion events.

Conversation-completion attribution

Attribution model crediting a conversion to a sponsored response across the full conversational thread, including multi-session and multi-device continuations.

Conversational relevance score

An auction-ranking input measuring how well a sponsored response continues a conversation's tone and specificity.

CPQC (Cost Per Qualifying Conversation)

A bidding model charging the advertiser only when a conversation reaches a defined completion event.

CPRS (Cost Per Response Shown)

A bidding model charging for placement regardless of completion; typically used for brand-presence campaigns.

Intent bucket

One of the finite set of inferred commercial intents (Symptom & Diagnostic, Provider Comparison, Cost & Insurance, Procedure Explainer, Corporate & Bulk, Second Opinion & Escalation) that the auction resolves conversations toward.

NMC

National Medical Commission — the Indian regulatory body whose guidance on medical advertising and professional conduct underpins ICG's compliance checklist.

VBB (Value-Based Bidding)

A bidding model available after ~90 days of conversion history, optimising toward advertiser-supplied conversion values.

Resources & Templates

ICG maintains a living library of resources referenced throughout this playbook, available to engagement clients:

• The full six-bucket copy template library (expanded beyond the samples in Chapter 6), with specialty-specific variants for cardiac, oncology, fertility, dental, aesthetic, diagnostics and Ayurveda.
• The Universal + specialty compliance checklists from Chapter 11 as a fillable sign-off document.
• A GA4 Exploration template pre-built to the UTM convention in Chapter 12.2, ready to connect to a client property.
• The Weekly Action Cadence Table from Chapter 10 as a shareable project-tracker template.
• 250 live ICG content URLs mapped to the six intent buckets — pillar, five segment pages, seventeen specialty pages, five Ayurveda pages, five Ayurveda×city pages, seven compliance pages, twenty-five glossary pages, twenty-five case scenario pages, twenty sub-specialty pages, and one hundred forty city×specialty programmatic pages — available as ready-made landing destinations for a new account's Phase 1 launch (Chapter 9).