Performance Max fake clicks: the spend you can't see

Performance Max (PMax) lets you serve ads across Search, Shopping, YouTube, the Display Network, Discover, Gmail and Maps from a single campaign. You set the goal and the budget; Google's automation largely decides where, to whom and when the ads appear. The price of that convenience is visibility: the search terms and placements you see line by line in a classic Search campaign are reported far more sparingly in PMax. That is the real difficulty with Performance Max fake clicks; it is hard to control spend you cannot see.

This article covers where fake and low-quality traffic can enter PMax, which controls Google gives you, and why feeding the automation the right signals is the most important defence.

Why PMax has its own risk profile

1. Breadth of inventory

PMax gets impressions not only in search results but also on placements such as Display Network sites and mobile apps. Some of these placements may have interfaces prone to accidental clicks or low-quality traffic. Google's invalid click filters cover this inventory too, but low-quality clicks that do not count as invalid still consume budget.

2. Limited reporting

For a long time PMax offered only insights showing “search term categories” rather than actual search terms. Google has recently started rolling out search terms and channel-level performance reporting for PMax, but visibility is still not on a par with classic campaigns and can vary from account to account. The placement report long offered impression-level data only, so it is hard to see how many clicks and how much cost a given placement produced.

3. The automation's dependence on signals

PMax learns from your conversion data. If an action counted as a conversion is triggered by bots, for example forms filled with fake details, the algorithm takes that traffic to be “good” and spends more budget finding similar traffic. PMax bot traffic is therefore not just wasted clicks but a self-reinforcing source of error.

Traces of fake traffic in PMax

• Conversions rising while sales or qualified leads do not.
• Form submissions with nonsensical names, invalid phone numbers and disposable email addresses.
• PMax sessions in your analytics showing engagement time far below average.
• Sudden click spikes at certain hours, accompanied by zero-engagement sessions.
• Apps and sites in the placement report with heavy impressions and no connection to your brand or sector.
• With final URL expansion on, traffic arriving on unexpected pages and leaving immediately.

How to measure the spend you cannot see

The most reliable way to close PMax's reporting gap is to look at data on your own site. With auto-tagging on, PMax clicks also arrive with a gclid; by building a campaign-level segment in your analytics tool you can compare engagement time, scroll depth and conversion rate of PMax traffic with other campaigns. If there is a clear difference, look at which days and hours it concentrates in.

The second step is to measure conversion quality. Match the conversions PMax reports with real outcomes in your CRM or order system. If a significant share of reported conversions does not correspond to a real customer, the problem is not just clicks but signal quality. That distinction determines which control to use first: placement and negative lists fix clicks, while conversion definitions and form protection fix signals.

The controls Google gives you

Controls in PMax differ from classic campaigns, and most sit at account or campaign settings level. The table below summarises the main ones and what they address.

Placement exclusion lists

You can create an excluded placements list in the content suitability section of your account settings. The list applies to eligible campaigns in the account, including PMax. If you see unrelated apps or suspicious sites in the placement report, add them to the list. Do not set it up once and forget it; compare it with the report at regular intervals.

Brand exclusions

Brand exclusion lists prevent PMax from showing on searches that contain the brands you specify. Their main purpose is to keep your own brand traffic from shifting into PMax or to control whether you appear on competitor brand searches. For fake clicks there is an indirect benefit: if you do not appear on competitor brand searches, you are less exposed to deliberate clicks coming through them. We cover competitor-driven clicks separately in our competitor clicks article.

Account-level negative keywords

Negative keywords defined at account level also apply to PMax's search inventory. Keeping out queries without purchase intent, such as “free”, “jobs” or “how to”, cuts bad traffic. This reduces inefficient rather than invalid clicks, but it is one of the easiest ways to clarify where your PMax budget goes.

IP exclusions

For a long time PMax did not support IP exclusions at campaign level. Google has started offering account-level IP exclusions; if the option is available in your account, use it for static, repeating sources. Even so, because of the 500-address limit, shared IPs on mobile networks and VPN rotation, IP exclusion is not enough on its own in PMax either.

The strongest defence: give the automation the right signals

In PMax it is not always possible to block individual clicks, but you can control what the automation learns to treat as “good traffic”. This is where you get the most lasting effect against fake clicks.

1. Tighten the conversion definition: remove easily triggered actions such as page views or button clicks from primary conversions.
2. Strengthen form protection: filter out bot submissions with human verification, a hidden honeypot field and server-side validation before they are counted as conversions.
3. Feed back qualified leads: import enquiries that turn into real customers in your CRM as offline conversions so PMax optimises towards them.
4. Keep suspicious visits out of audience signals: stop bot sessions from entering your remarketing and audience signal lists.
5. Use value rules: assign higher conversion values to the actions that are genuinely valuable.

A weekly PMax checklist

Because visibility in PMax is limited, a short, regular check routine is more effective than large, infrequent analyses. Once a week:

• Review search term insights and, if your account has it, the search terms report; move categories with the wrong intent to the account-level negative list.
• Check the placement report for new apps and sites with heavy impressions; add suspicious ones to the exclusion list.
• If the channel performance report is visible in your account, note which inventory the spend is shifting to; sudden shifts deserve attention.
• Compare a sample of conversions with your CRM: are the conversions real people?
• Check the invalid click column and billing credits to see what Google filtered in the period.
• Track post-click behaviour on your site in a separate segment for PMax traffic.

When launching a new PMax campaign

The cheapest precaution against fake traffic is taken before the campaign goes live. Before starting a new PMax campaign, make sure your account-level placement exclusion list and negative keyword list are ready. Turn on final URL expansion as a deliberate choice and exclude pages unsuitable for ad traffic from the outset. Limit conversion goals to genuinely valuable actions and test your form protection before launch. During the learning period the automation is very sensitive to early signals; fake conversions that slip in then shape targeting in the weeks that follow.

The same principle applies to feed-based PMax campaigns in e-commerce accounts: a purchase is hard for bots to trigger, but the risk grows if intermediate steps such as add-to-cart or checkout page views are used as primary conversions.

What wall.click does in PMax

wall.click scores every visit coming from PMax on your site: behaviour score, device fingerprint, data-centre and ASN detection, repeat-visit pattern. It keeps suspicious sessions out of your conversion and audience signals, adds static sources to supported exclusion lists and makes visible which PMax traffic shows no engagement. That way you shed light, with your own data, on the area Google does not report. See how bot detection works on the bot detection page and the overall approach on the PPC protection page.

Google's own filter also runs on PMax traffic and does not charge for the invalid clicks it detects. wall.click does not replace it; by measuring post-click behaviour it complements the part the filter cannot see. For the boundaries of invalid clicks, see our guide to invalid clicks in Google Ads.

Frequently asked questions

Can I see which placements my PMax ads appeared on?

Partly. The PMax placement report in the report editor lists the Display Network and YouTube placements where your ads showed, but click and cost detail is not as rich as in classic campaigns.

Should I switch off PMax and go back to Search campaigns?

Fake clicks alone should not drive that decision. PMax can perform well when fed good signals. Fix your conversion definition, exclusion lists and signal quality first, then compare performance.

Does Google refund invalid clicks in PMax too?

Yes. The invalid click policy covers all campaign types, including PMax. For activity you believe Google missed, you can file a refund request.

How can I tell which channel PMax clicks come from?

If the channel performance report is enabled in your account, start there. If not, you can roughly tell from the landing page, device and referrer in your own analytics tool; traffic from in-app placements, for example, usually carries an in-app browser signature.

Is running PMax on a small budget risky?

On a small budget the automation may not get enough conversion data to learn, and a few fake conversions have a proportionally larger effect. In that case take extra care with signal quality, or start with more controllable campaign types until conversion volume is sufficient.

Should I turn off final URL expansion?

Only if your pages suitable for ad traffic are limited or you see that traffic to expanded pages shows no engagement. Excluding specific URLs instead of switching it off entirely is also an option.