Invalid Traffic: Bots, Bad Clicks and Deductions

Invalid traffic is any ad impression or click that did not come from a real person with a genuine interest in the ad. Networks do not pay for it, advertisers do not want it, and enough of it can put your whole account at risk. The awkward part is that a lot of invalid traffic is nobody's deliberate fault.

This guide explains the main kinds, how networks take it out of your earnings, what it looks like in your reports and what you can do to keep your account clean. Policy wording changes over time, so treat the descriptions here as the well-established behaviour and confirm the current rules in your own account or the network's help centre.

What counts as invalid traffic

Ad networks use the term broadly. Google's publisher policies, for example, cover any clicks or impressions that artificially inflate an advertiser's costs or a publisher's earnings, whatever the intent behind them. In practice that includes:

• Clicks and impressions from automated tools, scripts and bots.
• Clicks by you, your team or your testers on your own live ads.
• Clicks prompted by encouragement, such as "click an ad to support us".
• Accidental taps caused by where an ad sits in your layout.
• Paid or exchanged traffic that never had any interest in your content.

The key point is that intent does not decide whether traffic is valid. An honest mistake still produces an invalid click, and it is still removed from your earnings. If the vocabulary around this topic is new to you, the glossary has short definitions.

General and sophisticated invalid traffic

Industry measurement bodies split invalid traffic into two broad groups, and many networks describe their filtering in the same terms.

General invalid traffic (GIVT)

This is the easy-to-spot kind. It covers known data-centre traffic, search engine crawlers, monitoring tools and other non-human visitors that identify themselves or match published lists. Filtering it is largely routine, and you rarely notice it because it is removed before earnings are counted.

Sophisticated invalid traffic (SIVT)

This is the harder kind. It includes bots built to mimic people, hijacked devices and browsers, malware that loads ads in the background, hidden or stacked ad units, and manipulated or incentivised activity. Detecting it takes deeper analysis across many signals over time, which is why deductions for it can arrive after the fact.

From your side, the distinction matters less than you might think. Neither kind is paid, and either can lead to action against an account if it keeps appearing.

The common causes on real sites and apps

Most publishers who run into invalid traffic problems are not running bot farms. The causes tend to be ordinary.

Accidental clicks from placement. An ad right next to a "next level" button, a banner that loads under a moving finger, or an in-article unit that pushes text down just as someone taps. Young users and fast-moving games are especially prone to this. Our guide to rewarded and interstitial ads covers placement timing in more detail.

Self-clicks. Tapping your own ads to check that they work, or a relative "helping out". Either can look like a pattern to a filter, and both break the rules.

Incentivised clicks. Offering a reward for clicking or viewing an ad, outside formats designed for it. Rewarded video is legitimate because the network controls the exchange and the advertiser knows what they are buying. A line of text asking readers to click ads is not.

Bought or exchanged traffic. Cheap traffic packages, pop-under networks, traffic exchanges and some low-quality promotion campaigns can deliver visitors who are not real, or who were never interested. You may not know a source is bad until deductions appear.

Bots you did not invite. Scrapers and automated tools can hit any public site. Most of this is general invalid traffic and is filtered quietly, but a persistent attack can still cause problems.

How networks deduct it

Networks protect advertisers first. When their systems decide that activity was invalid, the money goes back to the advertiser and comes out of your earnings. Google refers to these as invalid traffic or invalid activity deductions, and they are reflected when a month's earnings are finalised.

This is why the balance on your home screen during the month is only an estimate. Suppose an app's estimated earnings for a month are $500 and the finalised figure includes a deduction of 4%. That removes $20, leaving $480. On a small account, a deduction like that can be the difference between crossing the $100 payment threshold and waiting another month, as our AdMob payment guide explains.

Deductions can also touch more than one period if a pattern is found later. A line for invalid traffic in your payments history is not an accusation. It is a prompt to look at where that traffic came from.

Signs to watch for in your reports

You know your audience better than any filter, so your reports are your early warning. Patterns worth investigating include:

• A click-through rate that jumps well above your normal level, especially on one ad unit.
• Clicks rising while impressions and sessions stay flat.
• A burst of traffic from a country, referrer or app version you do not normally see.
• Very short sessions combined with heavy ad interaction.
• An eCPM that spikes during the month and then falls when earnings are finalised.

Compare like with like. A rewarded placement naturally sees more engagement than a banner, and the geographic mix moves prices a long way on its own. Our benchmark tables put a Tier 1 Android AdMob banner at around $0.90 eCPM against $0.10 in Tier 3, a ninefold gap. The eCPM explainer helps you separate price changes from volume changes, and our guide to reading the AdMob report shows where each figure lives.

How to protect your account

None of these steps are tricks. They are simply what a clean, compliant setup looks like.

Space your placements. Keep ads away from buttons, navigation, game controls and anywhere a finger or cursor is likely to be moving. Make sure ads are clearly distinguishable from content. On the web, avoid layouts that jump as ads load, which causes mis-taps. Our comparison of website ad formats covers which units are most prone to it.

Use test ads while you build. AdMob provides test ad units and lets you register test devices, so you can check placements without generating live impressions. Use them for every development build and switch to live units only for release. On websites, keep live ad code off development and staging copies, and never click your own ads to check them.

Know your traffic sources. Only buy promotion from sources you can verify, and watch quality signals such as session length and retention for each campaign. If a campaign delivers lots of visits and no engagement, stop it. A source that promises huge volumes at a tiny price is a warning sign in itself.

Keep incentives honest. Never ask, reward or nudge anyone to click or view ads, apart from properly implemented rewarded formats.

Report what you cannot explain. If you see suspicious activity you did not cause, tell the network. Google provides a way to report invalid activity through the AdSense help centre and the AdMob help centre. Reporting early shows good faith and gives their team useful data. You can also pause ads on an affected page or ad unit while you investigate.

When invalid traffic leads to limits

Repeated invalid traffic can lead a network to restrict the number of ads it serves to you, or in serious cases to disable the account. Our guide to why ad accounts get limited covers what those limits mean and what to do while one is in place. Read Google's publisher policies for the current definitions.

Keep in mind that every estimate from our AdMob revenue calculator and AdSense revenue calculator assumes clean, valid traffic. Our methodology explains the other assumptions behind those figures.