Performance Max Fake Leads: Why It Happens and How to Fix It (2026)

Performance Max is a single campaign type that serves across the entire Google network: Search, Display, YouTube, Discover, Gmail, and Maps. Google describes it as a goal-based campaign that uses automation to find converting customers across all channels. That breadth is its selling point, and it is also the reason it is so exposed to fake and low-quality leads.

Opaque placements. A standard Display campaign lets you see and manage exactly which sites and apps your ads run on. PMax gives you far less visibility and far less direct control. You cannot see the full picture of where your spend is going, which makes it harder to spot the placements that are driving junk.

Search Partners and Display inventory. Much of PMax volume comes from the Display network and Search Partners sites rather than Google Search itself. This inventory is vast and varied, and it includes a long tail of low-quality sites, some of which exist mainly to serve ads.

Made-for-advertising sites. A portion of that long tail is made up of made-for-advertising sites: pages built to maximise ad impressions rather than to serve a real audience. These environments attract automated traffic, and automated traffic is where a lot of fake form fills originate.

Broad reach plus automated bidding. PMax casts a wide net and then lets the algorithm decide where to spend. The bidding does not have a concept of "good lead" versus "bad lead." It only knows what you have defined as a conversion. If your conversion is a form submission and that form can be filled by a bot, the door is open.

The scale of the underlying problem is worth keeping in mind. According to the Imperva 2025 Bad Bot Report, automated traffic made up 51% of all web traffic in 2024, and bad bots alone accounted for 37%. When more than a third of the web is hostile automation, a campaign type that reaches as broadly as PMax will inevitably encounter a lot of it.

This is the part that makes PMax fake leads more dangerous than fake leads in a manual campaign. PMax does not just report junk conversions. It learns from them.

The campaign uses your conversion data as the steering signal for its automated bidding. Every time a conversion fires, PMax treats the audience, placement, device, time of day, and creative that produced it as a pattern worth repeating. It then shifts budget towards more of the same.

Now imagine those conversions are junk: bot form fills from a made-for-advertising site, or spam submissions chasing a lead magnet. PMax cannot tell the difference. It sees conversions, so it doubles down on the placements and audiences that produced them. The result is a feedback loop where the algorithm is actively trained to find more junk.

The damage compounds quietly. Your reported conversion volume goes up, which looks like success. Meanwhile real buyers receive less of the budget, your cost per genuine lead climbs, and your sales team wastes time on enquiries that never had any intent. By the time the disconnect between reported conversions and real revenue becomes obvious, the campaign has often spent weeks optimising in the wrong direction.

This is why fixing PMax fake leads is not just about cleaning your CRM. It is about cutting off the bad signal at the source before the bidding model bakes it in. For the broader picture of how lead fraud works across paid channels, see our lead generation fraud guide.

PMax does not give you the granular search-term and placement reporting you get in a standard Search or Display campaign. That does not mean you are blind. There are several signals you can use to infer where fake leads are coming from.

The placement report. PMax does expose a placement report showing where your ads appeared, even if it is less detailed than a Display placement report. Scan it for obviously irrelevant sites, parked domains, low-quality mobile apps, and anything that looks like a made-for-advertising page. These are your prime suspects.

The brand versus non-brand split. PMax tends to absorb branded search traffic, which converts easily and makes the whole campaign look efficient. Separate branded conversions from non-branded ones. If your headline ROAS collapses once you strip out brand traffic, the non-brand portion, which is where PMax is doing its real prospecting, may be full of junk.

Channel-level signals. Look at how performance breaks down by network and channel where the interface allows it. A spike in conversions concentrated in Display or in a particular time window is a classic fingerprint of automated traffic rather than genuine demand.

Account-level conversion auditing. Cross-check reported conversions against your CRM. Match lead records to the contacts that actually replied, qualified, or bought. The gap between conversions Google reports and leads that turn out to be real is the single most important number in this whole exercise.

Lead patterns. Gibberish names, mismatched geographies, disposable email domains, and bursts of submissions within seconds of each other all point to automation. We have seen this first-hand: on our own forms we deployed Cloudflare Turnstile plus gibberish-name detection and cleaned 62 bot and spam leads.

There is no single switch that stops PMax fake leads. The fix is a layered set of changes that together starve the bad feedback loop and steer the algorithm back towards real buyers.

1. Exclude brand traffic where it distorts ROAS

If branded search is inflating your PMax numbers, separate it out so you can see the true performance of your prospecting spend. Many advertisers run a dedicated brand campaign and use brand exclusions on PMax so the algorithm is judged on the non-brand demand it actually generates, not on customers who were already searching for you.

2. Rebuild asset groups around real themes

Loose, catch-all asset groups give the algorithm too much freedom to wander into low-quality inventory. Structure asset groups around tightly defined themes, products, or audiences, so the creatives and signals point PMax at the customers you genuinely want. Tighter themes give the model less room to chase junk.

3. Add account-level placement exclusions

Apply placement exclusion lists at the account level so they cover PMax as well as your other campaigns. Build a list of known low-quality sites, parked domains, made-for-advertising pages, and irrelevant app categories. Our GDN exclusion list and placement exclusion guide gives you a ready starting point you can apply across the account.

4. Tighten audience signals

Audience signals tell PMax who your good customers look like. Feed it strong, high-intent signals: your converting customer lists, qualified-lead segments, and tightly defined custom audiences. Weak or generic signals leave the algorithm guessing, and a guessing algorithm is more likely to settle on cheap junk traffic.

5. Harden your forms

This is the most direct lever and the one most advertisers ignore. If a bot cannot complete your form, it cannot create a fake conversion. Add a CAPTCHA or invisible challenge such as Cloudflare Turnstile, validate input server-side, reject gibberish names and disposable email domains, and rate-limit submissions. For a deeper walkthrough, read our guide on how to stop spam leads from Google Ads.

6. Feed back only validated conversions

This is the change that breaks the feedback loop. Instead of counting every form fill as a conversion, count only leads that pass validation, or use offline conversion imports to feed back the leads that actually qualified or bought. When PMax optimises on validated conversions rather than raw submissions, it learns to find real buyers instead of bots.

Sometimes the cleanest way to confirm that PMax is the source of your fake leads is to turn it off and watch what happens.

A controlled pause test works like this: pause the PMax campaign for a defined window, keep everything else running, and monitor your lead quality. If the junk submissions stop while PMax is paused and return when you re-enable it, you have isolated the source. If they continue regardless, the problem lies elsewhere, perhaps another channel or a site-wide bot issue, and you have saved yourself from blaming the wrong campaign.

Pause tests cost you volume, so time-box them and run them when you can tolerate a temporary dip. Use them when the diagnostic signals are ambiguous, or when you need a clear before-and-after to justify a bigger restructure to a client or stakeholder.