Performance Max for lead generation: check the leads you can accept

What should a lead-generation PMax campaign count as success?

Start with one business outcome that the team can identify consistently. A low reported cost per conversion is not enough to establish whether the campaign brings suitable enquiries.

Google’s lead-generation guidance recommends lead-specific goals such as Submit Lead Form, Qualified Lead and Converted Lead, with deeper-funnel actions where possible. That is guidance about useful inputs, not a guarantee that every reported lead will qualify.

Our suggestion: write a short acceptance rule with the sales team. For an invented repair company, an accepted enquiry might mean a reachable person requesting an offered service in the service area, with enough information to arrange the next step. Define duplicates, existing jobs and unanswered contacts separately.

Record who applies the rule and when. “The office dislikes the leads” is too vague to diagnose; “these contacts requested a service we do not offer” gives an operator something specific to investigate.

If recording itself is doubtful, first use the conversion-tracking checks. Changing campaign targets cannot establish whether a broken form or duplicated event is being measured correctly.

Is the campaign bidding towards the outcome you meant?

Check the actual campaign goal and included actions before interpreting its conversion total. The action’s name alone is not enough.

Google distinguishes primary actions, used for bidding when their standard goal is used by the campaign, from secondary actions reported for observation in All conversions. The exception matters: actions included in a custom goal are used for reporting and bidding regardless of their primary or secondary setting.

Our suggestion: ask the account operator to identify the campaign’s selected goals, the included actions and any custom goal. Keep supporting events such as page views distinct from the agreed lead outcome; do not assume changing a label alone removes an action from bidding.

For a tCPA strategy, Google recommends one funnel stage to bid towards; combining website visits and converted leads under one target can be duplicative. That is a reason to inspect the goal definition, not permission to remove actions without checking dependencies.

The same guide distinguishes lead-volume bidding with Maximize conversions from lead-value bidding with Maximize conversion value, optionally with a target ROAS. Its tip to choose a goal closest to a sale with at least 15 conversions in the last 30 days sits under value-based bidding. It is not stated there as a universal minimum for launching any PMax lead campaign.

If you need qualified or closed outcomes fed back from a CRM, use the separate offline-conversions guide. This article evaluates the outcome definition and evidence; it does not replace that implementation.

Do the reported leads survive a business-quality check?

Use a small, consistent acceptance table rather than replacing one unexplained total with another. Our suggested check keeps confirmed outcomes and missing evidence visible.

This invented example assumes one agreed group of 80 enquiries, a matching $2,400 campaign cost and a consistent qualification rule. These are teaching inputs, not client results or Google benchmarks. The enquiry count is assumed to have been reconciled for this example; do not divide spend by an unrelated CRM total.

The unresolved group makes a confident campaign verdict premature. The calculation is still useful: it shows what has been established and why the cheap recorded-lead headline is incomplete.

For a fair comparison, use the same qualification rule, campaign scope, enquiry dates and follow-up maturity. Our suggestion is to compare groups with equivalent time for assessment. Do not treat recent uncontacted enquiries as worse than an older group that has already completed its sales follow-up.

Keep enough evidence to identify the problem without publishing personal contact details. The useful output is a reason, responsible person and next check; a public article or client narrative does not need a list of names and phone numbers.

Which PMax evidence and controls address the actual problem?

Match the proposed intervention to a confirmed failure. A channel label alone does not establish whether an enquiry is suitable.

Google’s channel report is reached through Campaigns → Insights and reports → channel performance. Its Conversions view concerns the conversions the campaign optimizes for; Results also covers goals it is not optimizing for. Keep the selected view beside any exported number.

The report helps locate channel delivery and diagnostics. It is not a CRM qualification list. Google also says you cannot directly control budget allocation per channel, and a channel with no evident issue may receive little delivery because PMax is prioritizing other inventory. For the separate Search Network partner setting and its reporting limits in a Search campaign, see Search partners and lead quality.

Performance Max negative keywords are applicable to Search and Shopping inventory only.

These controls and qualification suggestions are documented in Google’s PMax lead guidance. For the full query workflow, use the search-terms guide rather than treating one exclusion as a complete quality strategy.

How long should you wait before a major change?

The current Google Ads Help page says to allow at least 1–2 weeks after measurement foundations are in place for the campaign to run and exit the learning phase before major changes, or up to 6 weeks for complex setups or low conversion volume.

The Accelerate article instead says at least 6 weeks after those foundations before major changes. Both were checked on 19 Oct 2026; their wording differs.

Our reading: do not turn either recommendation into a universal pass/fail date. A waiting period does not repair an evidenced form failure or prove that rejected leads will become suitable. Preserve the setup, volume, reporting maturity and specific issue when deciding whether the next step is observation or a controlled correction. For the general Learning status and how it differs from lead quality, see bid strategy Learning: wait, check or repair.

What can Variafy help establish?

Variafy offers read-only Google Ads access after your beta invitation. Its reading scope includes campaign type, status, bidding-strategy type and budget, campaign performance, enabled conversion-action settings and aggregate conversions by action, with returned-data limits.

Those readings can help separate campaign settings from reported action totals. They do not by themselves confirm each selected campaign goal, reproduce the channel report, qualify individual CRM contacts or send qualified-lead data to Google.

Our suggested reading request is: “For this PMax campaign and period, separate returned campaign settings, conversion-action settings and action totals. Retain completeness limits. List the campaign-goal and sales-quality evidence a person still needs before judging results.”

The responsible operator must obtain that missing evidence and approve any intervention. Variafy does not change Google Ads campaigns.

What decision should the acceptance check produce?

Finish with a bounded decision, not simply “PMax works” or “PMax is bad”. Our suggestion is to name the evidence, uncertainty and next owner.

There are several useful outcomes: the measurement needs repair; a specific mismatch needs investigating; the accepted-enquiry economics need comparing with the business limit; or assessment is incomplete and a conclusion must wait. None requires pretending that every unresolved lead is worthless.

Use the cost-per-lead guide for the wider economics. Here the deliverable is a traceable acceptance decision with a next check, rather than another list of PMax settings to change at once.

How we checked

Questions

Related guides

• Google Ads offline conversions: send CRM sales back
• Google Ads search terms report: exclude, keep or wait?
• Google Ads cost per lead: what leads cost in 2026 and how to lower it
• Brand vs non-brand Google Ads: separate the results before changing the budget
• Google Search vs Display ads: choose by the customer task, not the cheapest click