Bot Traffic in Google Ads: What It Is and What It Costs You

Bot traffic in Google Ads is any click or impression produced by software rather than a person, and it is a permanent feature of paid search, not an occasional glitch. Some of it is harmless and declares itself. Some of it is built specifically to look like a customer. The first kind gets filtered before it ever reaches your invoice. The second kind is the reason this page exists. If you run paid search for a business where a click costs more than a coffee, non-human traffic is quietly taking a cut of every campaign you run, and the damage goes well past the line item on your bill.

What bot traffic in Google Ads actually is

The industry splits non-human traffic into two buckets, and the split matters more than most people realise.

General invalid traffic (GIVT)

GIVT is traffic from declared crawlers and known automated tools. These bots identify themselves in the user agent string. They are not hiding. Search engine crawlers, uptime monitors, SEO tools, preview fetchers and link checkers all sit here. Filtering GIVT is close to a solved problem because the traffic tells you what it is. Ad platforms strip most of it out before billing.

Sophisticated invalid traffic (SIVT)

SIVT is traffic that actively disguises itself. It does not declare anything. It arrives with a browser user agent that looks like a real Chrome install on a real phone, from an IP address that looks like a real household, and it may move a cursor and scroll a page before it leaves. This is where the money goes. Everything difficult about bot traffic in Google Ads lives in this second bucket, and it is the category that generic filtering was never designed to catch.

The distinction is worth holding onto because vendors, ad platforms and blog posts all use the word "bot" loosely. A crawler that says "I am a crawler" and a click farm script running a headless browser behind a spoofed user agent are not the same problem and do not have the same fix.

Where the fake clicks come from

There is no single source, which is part of why this is hard to reason about. In practice the traffic hitting a typical advertiser's account comes from a handful of recognisable places.

• Automation tooling. Scripts, scrapers and testing frameworks pointed at your site or at the search results your ads sit in. Much of this is not aimed at you at all. You are collateral.
• Competitive clicking. Someone in your market, or an agency working for them, draining your daily budget so their ad sits higher for less. In tight local markets this is not paranoia, it is Tuesday.
• Affiliate and lead-gen fraud. Traffic manufactured to hit a payout threshold somewhere downstream.
• Proxy and residential IP networks. Infrastructure rented by the hour that makes automated traffic look domestic. The difference between datacenter and residential IP addresses explains why some of this is obvious and some is not.
• Accidental and incentivised human clicks. Not bots at all, but still invalid. Repeated clicks from the same person, clicks from someone paid to click, clicks from a device farm staffed by people.

What bot traffic in Google Ads costs you

The wasted click spend is the part everyone talks about. It is also the smallest part of the bill.

Budget that never had a chance

A daily budget is a finite thing. Every fake click consumes a slice of it, and once it is gone your ads stop showing to the people who would have converted. In a high-cost vertical this is not a rounding error. If your cost per click runs into double or triple figures, it takes remarkably few bad clicks to remove a real enquiry from your day.

Data you then make decisions on

This is the expensive one. Non-human traffic does not just cost money, it lies to you. Bounce rate climbs, average session duration falls, conversion rate drops, and none of it reflects anything your actual customers did. You then respond to that noise. You pause a keyword that was working. You rewrite a landing page that was fine. You cut a campaign because the numbers looked soft, when the numbers were soft because a third of the sessions were never human.

Smart Bidding learning from fiction

Automated bidding optimises toward the signals it receives. Feed it clicks that never convert and it draws conclusions: this audience is weak, this time of day is weak, this geography is weak. It then bids accordingly. The model is doing its job correctly on corrupted input, which is worse than the model being broken, because nothing looks wrong.

Time that does not show up anywhere

Fake form fills get called. Junk leads get chased. Sales people spend their morning on records that were never people. That cost never appears in any ad platform report, and in a service business it is often larger than the media waste.

Google already filters invalid clicks, so what is left

Google does run invalid click detection, it does remove clicks it identifies as invalid, and it does issue automatic credits that appear on the following month's bill. That system is real and it does genuine work. It is also deliberately opaque: Google does not disclose its detection criteria, so there is no way for an advertiser to audit what was caught, what was missed, or why.

The gap is not that Google is lazy. The gap is structural. Filtering happens after the fact, at platform scale, against criteria you cannot see, and it addresses the billing question rather than the traffic question. A click that gets credited back still entered your analytics, still occupied your budget for the day, still influenced your bidding signals, and still sent a bot to your landing page. We wrote a full breakdown of exactly where that line falls: what Google's invalid click refunds cover and what they leave behind. If you read one thing on this site, read that.

Signals that bot traffic in Google Ads is hitting your account

You will rarely get a smoking gun. What you get is a pattern that does not fit human behaviour.

• Click volume that spikes at hours when your customers are asleep
• Repeat clicks from the same network inside a short window
• Sessions with zero scroll, zero engagement, and an immediate exit
• Form submissions with plausible formatting but disconnected phone numbers or nonsense job titles
• Click-through rate holding steady while conversion rate falls off a cliff
• Geographic clusters that have nothing to do with where you sell

None of these prove fraud on their own. Together, in the same account, in the same week, they are worth taking seriously.

Check the Invalid Clicks Column Before Anything Else

Google Ads will show you how much traffic it filtered on its own, but the column is hidden by default. Open Campaigns, click the Columns icon above the table, choose to modify columns, search for "Invalid clicks" and apply it. The number you see is the invalid clicks Google filtered before billing. You were not charged for those.

Read it as a floor, not a ceiling. A rising invalid clicks count tells you the account is attracting automated traffic and that Google is catching some of it. It says nothing about the traffic that looked normal enough to be billed. If the invalid count climbs while conversion rate falls on the same campaigns, that combination is the reason to look at your own click records rather than relying on the platform's filter alone.

What you can actually do about it

There is a manual path and an automated path, and both have a place.

Manually, you can pull your traffic data, look for repeat offenders, and add the worst networks to your exclusion list. That approach has real limits and a real ceiling, which we cover in detail in our guide to using IP exclusions in Google Ads. The short version: the list is capped, addresses rotate, and you are always working from last week's problem.

Automatically, a dedicated layer watches traffic as it arrives and blocks the sources that are demonstrably not customers before they cost you another click. That is the job click fraud protection exists to do, and it runs continuously rather than whenever you next remember to check.

Whichever route you take, one rule is absolute. Do not block indiscriminately. Googlebot and AdsBot-Google have to reach your site or your organic indexing and your ad landing page checks both break, which is a self-inflicted wound far more expensive than the fraud. Our guide to the good bots you must never block covers which crawlers need clear access and why.

Some verticals feel this harder than others. Where a single conversion is worth thousands and the click prices reflect it, the maths changes completely, which is why we treat click fraud in legal advertising as its own case.

The practical take

You are not going to reduce non-human traffic to zero, and anyone promising that is selling something. What you can do is stop treating the credit line on your invoice as the whole story. The wasted spend is recoverable in part. The corrupted conversion data, the mis-trained bidding model and the sales hours burned on phantom leads are not. Start by looking honestly at your own traffic patterns, then decide how much of your budget you are comfortable donating to software.