When Google launched Performance Max as the mandatory successor to Smart Shopping campaigns, the promise was a single campaign type reaching customers across Search, Display, YouTube, Gmail, Discover, and Maps — with Google's algorithm handling distribution. At ₹5L/month or ₹5Cr/month, PMax is now unavoidable for most accounts. The question is not whether to use it. It is how to run it in a way that does not hand your entire acquisition strategy to a black box that actively resists your oversight.
What PMax automates — and why some of that is genuinely useful
Performance Max consolidates six Google surfaces into one campaign. The automation operates on three levels. Budget allocation: PMax decides how much to spend on which channel each day based on where it projects the highest conversion probability for your declared goal. Audience: PMax builds audiences from your asset group signals and expands from there using Google's user data. Bidding: PMax runs Target ROAS or Target CPA bidding against your declared conversion action. The useful part: at ₹10L+ monthly Google spend, manual channel management across six separate campaign types is genuinely impractical overhead. PMax removes the coordination cost of managing separate Display, YouTube, and Discovery campaigns. The prerequisite almost every account skips: PMax performance is entirely dependent on the quality of the conversion signal you feed it.
What PMax breaks — the three failure modes to diagnose first
Failure mode one: branded search cannibalization. PMax will bid on your branded search terms unless you explicitly add brand exclusions, which are available but not enabled by default. If you run branded Search campaigns separately, PMax and your branded campaign compete for the same query. Failure mode two: search query opacity. In standard Search campaigns, you see the exact queries triggering your ads and can add negatives systematically. In PMax, the search terms report is partial — sufficient to identify obvious irrelevant queries, insufficient to manage query relevance comprehensively. Failure mode three: asset group conflation. If your catalogue spans multiple product categories with different audiences, price points, and conversion behaviors, running them in a single asset group means PMax is learning one audience signal for fundamentally different products.
The controls that actually influence PMax behavior
Three controls meaningfully change what PMax does. First, brand keyword exclusions: add your brand name and all common spelling variations as brand exclusions at the campaign level. This prevents PMax from competing against your own branded Search campaigns. Second, account-level negative keyword lists: PMax respects account-level negatives. Build this list by downloading the Search Terms report weekly, identifying irrelevant queries, and adding them systematically. Third, asset group segmentation: split asset groups by product category, not by presumed audience. PMax builds its own audiences — your job is to give it the right creative signals for each product cluster. The fourth control — the most important — is the conversion action itself. Feed it purchase events with actual revenue values passed via offline conversion imports or server-side tracking, not proxy events.
The test that shows you what PMax is actually doing in your account
The question no PMax dashboard answers directly: is this campaign finding new customers, or cannibalizing existing demand you would have captured anyway? Run this diagnostic: create an audience segment of existing customers and website visitors from the last 90 days. Use the Audience Insights report within PMax to see what proportion of your conversions are coming from this audience versus genuinely new users. If the majority of PMax purchases are attributable to people who already knew your brand, the campaign is functioning primarily as a retargeting and branded capture mechanism — not a prospecting engine. Google's new customer acquisition goal setting within PMax biases the algorithm toward first-time purchasers, but it is opt-in and off by default. Enable it and set the new customer value uplift to reflect the actual LTV premium of a first-time buyer.