Performance Max Architecture for Indian D2C — 2026 Operator's Guide

Performance Max replaced Smart Shopping + Discovery + most Display campaign types in 2022-2023, and by 2026 it's the default Google surface for Indian D2C brands. The catch: PMax is a black-box ML optimiser, which means structure + audience signals + exclusions matter more than ever.

This is the Frameleads PMax architecture playbook for Indian D2C as of mid-2026. Anchored to the Performance Marketing Operations pillar.

PMax structure that actually works for Indian D2C

1. One asset group per product category (or per hero SKU bundle). Mixing categories (e.g. apparel + accessories) confuses the ML optimiser and dilutes performance.
2. Separate campaigns by funnel stage — Brand-defense (branded queries only), Prospecting (new customer acquisition), Retargeting (existing audience re-engagement).
3. Use Final URL Expansion sparingly. Off by default for D2C; on for content-heavy sites where Google can find topical depth.
4. Set tROAS / tCPA targets aligned with Shopify-actual margins, not in-platform-reported ROAS. PMax's reported ROAS over-reports by 20-40% — adjust targets accordingly.
5. Lock customer-list audiences to Customer Acquisition Goal — tells PMax to prioritise new customers over existing ones. Critical for D2C where existing-customer re-acquisition kills incremental CAC.

Asset signals — what to feed PMax

• Audience signals: customer-match lists, Shopify Audiences, GA4 first-party audiences (purchasers / high-AOV), Lookalike-equivalents.
• Creative assets: 15-30 images, 5-10 videos (vertical + horizontal), 5-10 headlines, 5-10 descriptions per asset group. PMax tests combinations.
• Product feed: clean Merchant Center feed with rich attributes (color, size, gender, age_group, custom_label_0-4). Custom labels enable inventory segmentation.
• Conversion signals: Enhanced Conversions + Offline Conversions (purchase events from CRM/Shopify with order value). Without these, PMax optimises on signal-loss inflated data.
• Final URLs: product page URLs (not category pages) so PMax can attribute conversions to specific products.

Exclusions that protect margin

1. Brand keyword negatives at the account level — prevents PMax from cannibalising your branded Search campaigns. Add [your brand name], [brand misspellings], [brand + variant] as account-level negatives.
2. Low-margin SKU exclusions via custom_label_0 in the feed — flag products with margin <25% and exclude from PMax (run them via Standard Shopping instead).
3. Placement exclusions for YouTube + Display surfaces — exclude obvious low-quality categories (made-for-kids content, low-quality content sites).
4. Geographic exclusions — exclude geos you don't ship to, and exclude geos with consistently poor CAC (typically Tier-3 / Tier-4 cities for premium D2C).
5. Demographic exclusions rarely — PMax restricts demographic exclusions, and over-exclusion hurts more than it helps.

Reporting — inspecting the black box

• Search Terms Insights — shows search themes PMax matched against. Mine this weekly; convert high-performers into Standard Search campaigns where you can control bids precisely.
• Asset Reports — tells you which creative assets PMax is showing most + their performance. Pause low-performers; ship more variants of high-performers.
• Audience Insights — shows which audience segments PMax found valuable. Use to inform Lookalike + customer-list audience expansion.
• Brand-defense overlap report — quarterly check on whether PMax is cannibalising branded Search (despite account-level negatives, some leakage occurs).
• Shopify-actual reconciliation monthly — compare PMax-reported conversions vs Shopify-actual orders. Spread above 30% signals attribution issues or modelled-conversion inflation.

Common PMax mistakes in Indian D2C

• Running PMax without Enhanced Conversions wired — under-attributes 30-60% of conversions.
• Setting tROAS targets based on PMax-reported ROAS instead of Shopify-actual.
• Mixing product categories in one asset group (apparel + accessories + footwear in one group).
• Not excluding brand keywords at account level (PMax cannibalises Search).
• Setting Customer Acquisition Goal to OFF (PMax over-spends on existing customers, kills incremental CAC).
• Skipping Search Terms Insights mining (missing the chance to harvest winning themes into Standard Search).

Frameleads runs PMax architecture as a standard part of Google Ads engagements at Scale tier. Book a free audit — we'll score your current PMax structure against the framework above on the call.