Google Ads · Performance Max
Performance Max reaches across Search, YouTube, Display, Discover, Maps, and Gmail from a single campaign, and it optimizes toward whatever conversion you feed it. Feed it form fills and it will find you form fills, from bots if necessary. Run Performance Max for lead generation on downstream revenue signals with real guardrails and it becomes an incremental demand engine. Here is the checklist we require before it spends a dollar.
Performance Max shows you almost nothing about where your money goes, and in lead generation that opacity is expensive. Optimizing on raw form submissions invites click farms and bot traffic, because a fake form fill costs the machine nothing and satisfies the goal you set. The campaign type is not broken; the signal usually is. Everything below exists to make the signal impossible to fake before the black box gets a budget.
Non-negotiable. The campaign bids on high-integrity conversions bots cannot fake: offline-imported sales-qualified leads, qualified calls with a 60-plus second buffered duration and IVR validation, or pipeline revenue. Raw form submits disqualify the account from running PMax at all.
A minimum of 15 to 30 qualified conversions per month per campaign, ideally 50 or more. Below that the algorithm is guessing, and PMax guesses expensively across six channels.
Core exact-match and high-intent Search campaigns should sit near 100 percent impression share before PMax launches. PMax exists to capture incremental demand. It does not replace Search, it feeds on whatever Search leaves uncovered.
A daily budget of 3 to 5 times target CPA, in practice $100 to $150 per day at minimum, so the algorithm can explore YouTube, Display, and Discover inventory without choking its own learning phase.
Offline conversion feedback tied to click IDs and hashed CRM data, so the campaign optimizes only toward qualified leads and revenue. Location targeting set to Presence, not the Presence or Interest default. Brand exclusions and shared negative lists, so PMax cannot inflate its numbers on demand you already own. And landing-page friction: native PMax lead forms disabled in favor of multi-step pages with phone and email verification and qualifier questions.
Bidding on raw form submits, which invites click farms and bots. Basic reCAPTCHA alone, which modern bots pass. Chasing an Excellent Ad Strength rating, which measures asset diversity, not lead quality. And treating audience signals as targeting, when they are hints the algorithm is free to ignore, not limits it must respect.
The feedback loop behind the real controls: offline conversion tracking →
Three tells, checked in order. A sudden impression-share drop in dedicated brand or exact-match Search campaigns right after PMax scales. Branded or already-targeted queries appearing in the PMax Search Terms Insights. And the classic: PMax reporting a flattering CPA while blended cost per lead rises or the qualified pipeline stays flat. Our defaults prevent most of it before it starts: account-level brand exclusions on every PMax campaign, a dedicated exact-match brand Search campaign held at high impression-share targets, shared global negative lists covering competitors, job seekers, support queries, free, pdf, and login terms, and customer-match exclusions so the campaign cannot pad its numbers retargeting people who already bought. One nuance worth knowing: Google prioritizes exact-match Search over PMax, but PMax will take those queries the moment the Search campaign is budget-constrained.
The two get conflated constantly, and they should not be. Performance Max is a cross-channel campaign type spanning Search, YouTube, Display, Discover, Maps, and Gmail, driven by asset groups. AI Max is a feature layer inside Search campaigns: broad match paired with Smart Bidding, final URL expansion, and automated copy. For lead-gen accounts hesitant about a multi-channel black box, AI Max on Search is the better testing ground, because it delivers the AI query expansion with keyword guardrails, query transparency, and channel isolation intact. It is also the version with published proof: in the Google-published test on our health insurance client, AI Max text customization delivered a 35.8 percent uplift in clicks, a 28 percent uplift in conversions, and a 23.4 percent uplift in conversion value over 45 days, all while holding the client’s strict target ROAS. AI features earn their keep when they run on real revenue signals. That story holds for PMax too, which is the whole point of this page.
The Google-published case studies → · How the health insurance account runs →
Smart Bidding Exploration now runs inside Performance Max and AI Max, letting target-constrained campaigns reach adjacent search categories with controlled bidding flexibility. It does not change the prerequisites on this page: exploration amplifies whatever conversion signal you feed it, so downstream signal quality matters more, not less. Announced among Google Marketing Live 2026 lead generation launches.
Google is bringing multi-stage learning to bidding natively: the system learns from every stage of the lead-to-sale journey, not just the final event. This is the principle our offline conversion hierarchy already runs; if you feed PMax staged, revenue-valued events today, the beta is a complement rather than a replacement. Background coverage.
Yes, under conditions most accounts skip. It needs a conversion signal bots cannot fake, a minimum of 15 to 30 qualified conversions per month per campaign, core Search near 100 percent impression share first, and a daily budget of 3 to 5 times your target CPA. Run it without those and Performance Max for lead generation becomes a machine for buying junk form fills across six channels at once.
Downstream signals only: offline-imported sales-qualified leads, qualified calls with a 60-plus second buffered duration and IVR validation, or pipeline revenue. Never raw form submissions. The campaign type reaches across Search, YouTube, Display, Discover, Maps, and Gmail, and whatever you reward is what it multiplies.
With controls that actually bind: offline conversion feedback tied to click IDs and hashed CRM data, location targeting set to Presence instead of the default, brand exclusions and shared negative lists, and multi-step landing pages with phone and email verification instead of native lead forms. Basic reCAPTCHA alone, chasing Excellent Ad Strength, and treating audience signals as targeting are placebos.
Check three places. A sudden impression-share drop in your dedicated brand or exact-match Search campaigns right after scaling Performance Max. Branded or already-targeted queries showing up in the Search Terms Insights. And a flattering campaign-level CPA while blended cost per lead rises or the qualified pipeline stays flat. If any of the three shows up, the campaign is taking credit for demand you already owned.
They are different products. Performance Max is a cross-channel campaign type driven by asset groups. AI Max is a feature layer inside Search campaigns: broad match with Smart Bidding, final URL expansion, and automated copy. For lead-gen accounts hesitant about multi-channel black boxes, AI Max on Search is the better testing ground because it keeps keyword guardrails, query transparency, and channel isolation. In the Google-published test on our health insurance client, AI Max text customization delivered a 35.8 percent uplift in clicks, 28 percent in conversions, and 23.4 percent in conversion value while holding a strict target ROAS.
A daily budget of 3 to 5 times your target CPA, which in practice means $100 to $150 per day at minimum, so the algorithm can explore YouTube, Display, and Discover inventory without starving its own learning. Our engagements start at $30,000 per month in ad spend, where those exploration budgets fit inside a properly structured account.
Next Step
If you are spending $30,000 or more per month on Google Ads and Performance Max is either off the table or out of control, we should talk.
Request a Proposal →