How to stop junk traffic in Google Ads

Junk traffic in Google Ads has three sources, and the fix for each is different. Diagnosing which source is leaking before you start blocking saves hours of misdirected work.

The three sources

Source one: broad-match expansion. You bid on plumber phoenix as broad. The matcher serves your ad against plumbing services arizona, best plumber tempe, and emergency drain repair scottsdale. Some are good. Some are junk. Diagnostic: search terms report shows queries that are tangentially related but not what you bid on. Fix: tighten match types or add phrase negatives.

Source two: irrelevant placements. This is mostly a Display Network or Performance Max problem. Your ad appears on a website or app where the audience does not match. Diagnostic: placements report shows traffic from low-quality publishers, mobile games, or unrelated content sites. Fix: placement exclusions, content category exclusions.

Source three: junk audience signals. Your audience targeting is too broad or includes signals you did not intend. Common in Performance Max where the bidding can drift toward demographics that click but do not convert. Diagnostic: audience report shows poor conversion rate from specific demographic or audience segments. Fix: audience exclusions, narrower targeting.

Source one is the most common

For Search campaigns, broad-match expansion is the dominant source of junk traffic. The other two matter mostly for Display, Shopping, and Performance Max campaigns.

If your account is mostly Search, focus there. The fix has two components:

When to use phrase vs exact match for the keyword side

The trade-off:

Source two: placement exclusions for Display and PMax

Placement junk is its own kind of pain. Common offenders:

• Mobile gaming apps where the audience taps the ad accidentally
• Low-quality content farm sites
• Sites in adjacent but wrong languages
• Domain-parking sites that auto-show ads

The placements report (under Reports → Predefined → "Where ads showed") lists every placement with cost and conversion data. Filter to high-cost, low-conversion-rate placements and add them to the campaign's placement exclusions.

For Performance Max, placement exclusions are limited; you cannot do per-campaign placement exclusions the way you can for Display. The closest equivalent is the "Account-level placement exclusions" list, which applies across all PMax and Display campaigns.

Source three: junk audience signals

This is most common in Performance Max where the bidding can drift toward audiences that click but do not convert. Examples:

• Demographic skew toward an age range where your service is uncommon
• Affinity audiences that overlap with your real customers but include a noisy adjacent group
• Custom audiences built from generic keywords

The fix: in the audience report, find segments where conversion rate is more than fifty percent below account average. If consistent across multiple weeks, exclude.

In Performance Max specifically, the controls are coarser. Audience signals are positive hints to the bidder rather than constraints. PMax can still bid outside the signals. The mitigation is layered: signal what you want, exclude what is clearly wrong, and accept that the matching will be imperfect.

What good looks like, after a sweep

After the three-source sweep on a typical small-business account:

• Search campaigns: junk traffic share drops by half within two weeks. Cost-per-conversion improves.
• Display campaigns: placement exclusions clean up the worst publishers. Click-through rate goes up because the impressions concentrate on better surfaces.
• Performance Max: audience signal tweaks shift the distribution. The improvement is slower and harder to see because PMax controls are weaker.

The sweep is not a one-time fix. Junk patterns shift over time as the auto-bidder explores. A monthly fifteen-minute sweep keeps the worst offenders out and lets the rest of the account work.