Fake Clicks vs False Clicks vs Invalid Clicks: What Each Term Means

Three terms, three different situations

Start with what each phrase usually means in practice. False clicks is the loosest term: it normally refers to clicks that carry no real interest, such as a thumb brushing an ad on a small screen or a visitor who closes the page after one second. Fake clicks implies something manufactured: scripts, click farms or hidden elements that generate a click event no person intended. Invalid clicks is the platform's own category, the umbrella that ad networks use for any click they decide should not count toward billing.

The overlap is why confusion is common. A false click can end up classed as invalid, a fake click is always a candidate for invalid, and some invalid clicks are neither: duplicate clicks from a double tap, or clicks from the advertiser's own team. Our click fraud glossary lists the neighboring terms, and the overview of click fraud explains the wider picture.

Accidental clicks: the human-error bucket

Accidental clicks come from layout and device behavior rather than bad actors. Common sources are ads placed directly beside navigation targets, sticky banners that shift under a scrolling thumb, interstitials with tiny close buttons, and in-app placements where the ad slot loads late and pushes content into the tap point. The visitor never meant to engage, so the session looks like a click with near-zero dwell time and no downstream events.

Platforms filter some of these automatically, and Google says it credits advertisers for clicks it later identifies as invalid. But a high rate of one-second sessions on a single placement is also a signal about your own inventory choices. If the pattern clusters on one placement, one app or one device class, review the placement before assuming fraud. Our invalid clicks article covers how Google Ads treats these in reporting.

Automated clicks: software, not people

Automated clicks come from scripts, headless browsers and compromised devices. Some are benign and expected, such as link scanners, preview fetchers and monitoring tools. Others are not benign: botnets that load pages and fire click events, or malware that clicks ads inside an app without the user noticing. Industry bodies group these under general invalid traffic (GIVT) for the easy-to-spot cases and sophisticated invalid traffic (SIVT) for the cases that imitate people; the GIVT vs SIVT breakdown goes through both.

The practical difference from accidental clicks is consistency. Automated traffic tends to show uniform timing, identical screen and browser properties, datacenter or unusual network origins, and no meaningful interaction after landing. Detecting it relies on signals gathered across many requests, which is what the bot traffic detection guide walks through.

Deliberate manipulation: clicks with intent

The third case is a person or service clicking on purpose to waste someone's budget or to inflate someone's earnings. Competitor clicking, publisher self-clicking and paid click services all fall here. These clicks can come from real devices on residential connections, which is why they are the hardest to separate from genuine traffic using any single field. Patterns across time, repeated visitors, and geography that does not match the campaign targeting are the usual leads.

Because intent is hard to prove, platforms rely on aggregate evidence, and so should you. Keep dated exports of the affected campaign, the click timestamps, and the landing-page session data, since a credit request is judged on evidence rather than suspicion. The competitor click fraud detection article shows what that evidence typically looks like.

How to triage a spike in suspicious clicks

When click numbers jump without a matching rise in conversions, work through the buckets in order of cost to check. First compare click counts across platform, analytics and server logs; a gap points to counting differences or duplicate firing rather than fraud. Second, segment by placement, device and app; concentration suggests accidental clicks from a layout. Third, look at session quality for the same segments: dwell time, scroll, events fired. Fourth, look at network and timing patterns for signs of automation or coordination.

Only after those steps does it make sense to apply controls: exclusions for the placements that misbehave, filtering rules for traffic that fails basic checks, and a credit claim with the evidence you collected. The click fraud protection guide and the ad fraud types overview help match the pattern you found to a response.

Quick comparison

FAQ