Paid Advertising

Performance Max, run without losing control

Performance Max hands Google your budget and most of the steering wheel. For a remodeler chasing real jobs, the difference between a money pit and a real channel is the handful of inputs you still control. Here is exactly which levers matter.

8 min read Updated June 2026

71% Share of surveyed advertisers running Performance Max in 2025, up from 60% in 2024 (Optmyzr State of PPC, 2025)
28% Cost-per-lead reduction after a clinic aligned form value with offline sales data in PMax (Search Engine Land, 2024)
50 Search themes allowed per asset group after the 2025 increase from 25 (Google Ads Help, 2025)

Performance Max is one campaign type that serves across Search, YouTube, Display, Discover, Gmail, and Maps from a single budget, with Google deciding placement, audience, and bid in real time. Adoption is everywhere now: the share of surveyed advertisers running PMax rose from 60% in 2024 to 71% in 2025 (Optmyzr State of PPC, 2025). The problem for a home builder or remodeler is that PMax is a black box that optimizes toward whatever you mark as a conversion, and by default that means the cheapest form-fill it can find, not the homeowner who signs a $45k kitchen. The good news is you are not powerless. A specific set of inputs still gives you control, and used together they turn PMax from a guess into a managed channel.

Why Performance Max is a black box, and what that costs lead gen

Performance Max collapses six of Google's surfaces into one automated campaign and gives the algorithm control over placement, audience, creative combination, and auction-time bidding. You no longer choose keywords, you no longer choose where each ad runs, and until recently you could barely see what you were buying. That opacity is the trade: you hand Google the steering wheel in exchange for reach it claims you could not assemble manually.

For ecommerce the trade is tolerable because a purchase is a clean, valuable signal: money changed hands. A contractor's lead has no such gatekeeper. A form submission is the start of the estimate process, not the end of one, and a bot, a price-shopper who will never book, and a homeowner ready to remodel all fire the same conversion tag. If you let PMax optimize on raw form-fills, it does exactly what you asked and finds you the cheapest conversions in existence, which are usually junk.

Search Engine Land calls this the feedback loop of doom: spam submits forms, Google counts them as conversions, then optimizes toward more of the same low-quality traffic (Search Engine Land, 2024). The fix is not to avoid PMax. It is to control the inputs so the black box optimizes toward real leads instead of noise.

The real risks before you turn it on

Three risks hit lead-gen accounts hardest, and all three are baked into the defaults. The first is optimizing toward low-quality conversions, because the standard setup counts every form-fill equally and the algorithm chases volume of the cheapest one. The second is brand cannibalization: PMax will happily serve on searches for your own company name, claim those near-guaranteed conversions as its own, and make itself look efficient while spending on traffic you would have won for free.

The third is the lack of placement transparency. Historically you could not see which queries, channels, or placements drove your results, which made it nearly impossible to tell good spend from wasted spend. Google has since added channel-level reporting and full search-terms reporting in 2025, but the visibility is still partial and arrives after the money is spent (PPC Land, 2025).

Knowing these risks reframes the setup. You are not configuring a campaign so much as building guardrails around an algorithm that will otherwise optimize for the wrong thing. Every lever below exists to close one of these three gaps.

The three lead-gen risks PMax creates by default:

  • Optimizing toward junk: every form-fill counts equally, so it chases the cheapest conversion, not the best lead
  • Brand cannibalization: it serves on your own brand searches and claims credit for traffic you would win anyway
  • No placement transparency: until 2025 you could not see which queries or channels spent your budget
  • Final-URL drift: by default it sends traffic to pages it picks, not only the landing pages you chose
  • Budget concentration: one budget spreads across six surfaces with no per-channel control over the split

Fix the conversion goals first, then import offline value

The single most important control is what you tell PMax to optimize for. Mark only meaningful actions as primary conversions, and demote or remove soft actions like newsletter signups. Then add spam defenses before the tag fires: invisible reCAPTCHA or a honeypot field stops bots from teaching the algorithm to find more bots, and counting only phone calls longer than 60 seconds filters out hang-ups (Search Engine Land, 2024).

The bigger move is importing offline conversion value. Connect your CRM so that when a lead qualifies, books, or closes, that outcome and its value flow back to Google. Now the algorithm learns the difference between a form-fill that goes nowhere and one that becomes revenue. Search Engine Land documents a clinic that cut cost per lead by 28% after aligning form value with offline sales data (Search Engine Land, 2024). Import quickly rather than waiting months, because while you stall, PMax is learning from incomplete data.

A practical pattern is progressive lead scoring: assign a small value to a raw lead, a larger one when it turns into a booked estimate, and the full value when the homeowner signs the job. That teaches Google which leads progress and which stall, which is the entire point of value-based optimization for lead gen.

PMax does exactly what you ask. Mark every form-fill as a win and it will find you the cheapest junk in existence.

The levers that actually give you control

Once goals and offline value are right, the rest is steering. Asset groups are your structural lever: each holds its own final URLs, creative, audience signals, and search themes, so you can split distinct services like kitchen remodels, home additions, and basement finishes and read them separately. A campaign allows up to 100 asset groups, but for lead gen a few tightly themed ones beat a sprawling pile.

Audience signals and search themes are suggestions, not restrictions. Audience signals tell the algorithm who your best customers look like; it then finds similar people rather than limiting reach to that list. Search themes do the same for queries, and the 2025 increase to 50 per asset group, paired with a new usefulness indicator that flags whether a theme actually added traffic, gives you more honest input than before (Google Ads Help, 2025).

Final-URL expansion is the lever most lead-gen accounts get wrong. It is on by default and lets Google send clicks to pages it considers more relevant than the ones you built. Turn it off so traffic only goes to your chosen landing pages, then use URL exclusions and page feeds for finer control. Brand protection runs through two settings: account-level and campaign-level negative keyword lists, now up to 10,000 per campaign, plus self-serve brand exclusions, which since December 2024 block branded queries by recognized entity rather than exact match, catching misspellings and foreign scripts (Karooya, 2024; Google Ads Help, 2024).

The control levers worth setting on every lead-gen PMax campaign:

  • Asset groups: split distinct services or offers so you can read and steer them separately
  • Audience signals: feed your converters and CRM segments so it finds lookalikes, not just remarketing
  • Search themes: add up to 50 per asset group and watch the usefulness indicator to prune dead weight
  • Brand exclusions and negative lists: block your own and competitor brand traffic at account and campaign level
  • Final-URL expansion off: send clicks only to the landing pages you chose, then add URL exclusions
  • New-customer acquisition setting: bid up for new customers instead of paying full price for existing ones

Reading the limited reporting without fooling yourself

PMax reporting got materially better in 2025, but it still asks you to infer rather than see. Channel-level performance reporting, which rolled out to accounts by May 2025, breaks delivery across Search, YouTube, Display, Discover, Maps, and Gmail, so you can finally tell whether your lead-gen budget is funding Search intent or being parked on cheap Display impressions (PPC Land, 2025).

Full search-terms reporting, announced in April 2025, shows which queries triggered your ads, which is how you find the irrelevant searches feeding junk leads and turn them into negatives. Asset-level reporting now segments by network too, so you can see which creative carries each surface. Use these together: if conversions look cheap but channel reporting shows the spend sitting on Display and search terms are vague, your low cost per lead is a warning, not a win.

The honest caveat is that this visibility is partial and retrospective. You are reading what already happened across an algorithm you do not fully direct, so treat the reports as a steering aid for your inputs, not as the granular control a Search campaign gives you.

How WellBuilt runs Performance Max

WellBuilt treats Performance Max as a managed channel with guardrails, not a set-and-forget button. Before any budget moves, we audit conversion actions so the algorithm optimizes toward qualified leads, add spam defenses so bots never enter the learning data, and stand up offline conversion import from your CRM so deal value, not form-fill volume, drives the bidding.

From there we build the controllable inputs deliberately: tightly themed asset groups, audience signals fed from your real converters, search themes pruned against the usefulness indicator, final-URL expansion off, and brand traffic walled off with negative lists and brand exclusions so PMax cannot take credit for searches you already own. We do not promise a number we have not earned for you; PMax performance depends on your data, offer, and sales follow-up.

Then we read the 2025 reporting every week, channel splits, search terms, and asset performance, and adjust the inputs rather than the algorithm. The deliverable is a channel you understand and can defend, with the black box working toward your real leads instead of the cheapest ones it can find.

Key takeaways

  • Mark only qualified actions as primary conversions and add reCAPTCHA or a honeypot before the tag fires, so spam never trains the algorithm.
  • Import offline conversion value from your CRM with progressive lead scoring so PMax optimizes for closed revenue, not raw form-fills.
  • Turn off final-URL expansion so clicks only reach the landing pages you built, then tighten with URL exclusions and page feeds.
  • Wall off brand traffic with account and campaign negative keyword lists plus self-serve brand exclusions so PMax cannot claim searches you already own.
  • Read the 2025 channel and search-terms reports weekly and adjust your inputs, not the algorithm, since the visibility is partial and retrospective.

SourcesOptmyzr State of PPC, Performance Max adoption 60% to 71%, 2025 · Search Engine Land, Why Performance Max lead generation fails and how to make it work (28% CPL reduction, feedback loop, offline import, URL expansion), 2024 · Google Ads Help, Search themes increased to 50 per asset group and usefulness indicator, 2025 · Google Ads Help, About brand exclusions in Performance Max (self-serve, December 2024), 2024 · Karooya, Why negative keywords matter more than ever in Performance Max (up to 10,000 negatives), 2024 · PPC Land, Performance Max channel-level reporting and full search-terms reporting rollout, 2025 · Google Ads Help, About final URL expansion and asset groups in Performance Max, 2024

Questions, answered straight.

Can Performance Max work for lead generation at all?

Yes, but only if you change the defaults. Out of the box PMax optimizes toward the cheapest form-fill, which for lead gen means junk. With offline conversion value imported from your CRM, spam defenses in place, and brand traffic excluded, it can drive qualified leads. Search Engine Land documents a clinic that cut cost per lead 28% after aligning form value with offline sales data.

Should I turn off final-URL expansion?

For most lead-gen accounts, yes. Final-URL expansion is on by default and lets Google send clicks to pages it picks rather than the landing pages you built and tested. Turning it off keeps traffic on your chosen, conversion-optimized pages. You can layer URL exclusions and page feeds on top for finer control.

How do I stop PMax from spending on my brand searches?

Use two controls together. Account-level and campaign-level negative keyword lists block your own and competitor brand terms, with campaigns now supporting up to 10,000 negatives. Self-serve brand exclusions, available since December 2024, block branded queries by recognized entity, which also catches misspellings and foreign scripts that keyword lists miss.

Is PMax or standard Search better for my account?

Standard Search wins when you need keyword-level control, have low conversion volume, or cannot yet feed offline data, because it gives transparency PMax does not. PMax fits when you have solid conversion tracking, offline value flowing from a CRM, and enough volume to learn, and when you want reach across Google's surfaces from one budget. Many lead-gen accounts run both and let Search own high-intent brand and core terms while PMax expands beyond them.

Paid Advertising

Want this run for you, not just read about?

Make every ad dollar accountable to pipeline — lower cost per lead, sharper bidding, less waste.

Get your Blueprint Our Paid Advertising service