Most publishers find out about bot traffic the hard way: a payment shrinks, a deduction appears, or ad serving gets limited. The signs were usually there weeks earlier. This guide covers what to look for in your analytics and ad reports, and how to tell a real problem from noise.
Why it's worth checking regularly
Bots don't just waste server resources. When they load pages with ads on them, they generate impressions that buyers don't want to pay for. Ad platforms filter a lot of this automatically, but what gets through can lower the price buyers are willing to pay for all of your inventory, and repeated problems put the account itself at risk. We cover the wider picture in invalid traffic explained.
A monthly look at the signals below takes twenty minutes. It's one of the cheapest habits a publisher can have.
Signals in your analytics
No single metric proves bot traffic. Look for several of these together, especially if they start on the same date.
- Sessions with zero or near-zero engagement. A sudden block of visits with no scroll, no clicks and a few seconds on page.
- Unnaturally high engagement. The opposite also happens: sessions that view dozens of pages at a perfectly regular pace.
- Traffic spikes with no cause. No article went viral, no newsletter went out, yet visits doubled overnight.
- Odd geography. A country you've never written for suddenly becomes a top source.
- Strange device or browser mix. Old browser versions, unusual screen resolutions, or one exact device model making up a big share of visits.
- Referrers you don't recognise. Sites with no reason to link to you, or "direct" traffic that jumps without explanation.
- Visits at fixed intervals. Real readers come in waves that follow time zones. Bots often run on a schedule.
Signals in your ad reports
Your Ad Manager or partner reports show the commercial side of the same problem.
- Impressions up, revenue flat. If buyers are discounting the traffic, eCPM falls as volume rises. See eCPM and RPM explained.
- Viewability drops. Bots often load ads that nobody ever sees. A sharp fall in viewable rate on part of your traffic is a warning sign.
- Click-through rate jumps. A CTR far above your normal range, especially on one unit or one country, deserves a close look.
- Deductions after the month closes. Earnings revised down for invalid activity mean the platform already found something.
Check where the traffic comes from
Real readers mostly arrive from home broadband and mobile networks. A meaningful share of visits from cloud hosting and data-centre networks is a classic bot signal, because that's where automated browsers usually run. VPNs and proxies are more mixed: some real people use them, but they're also used to disguise automated traffic. We go into this in VPN, proxy and data-centre traffic.
Your server or CDN logs can show the networks behind your visits. If you see the same hosting provider again and again, that's worth following up.
One more thing to keep in mind: bots change. A pattern that was obvious last year may be disguised this year, with traffic routed through residential connections and browsers that look perfectly normal. That is why a single check, done once, isn't enough. Keep a simple monthly log of your main numbers: visits, engagement, top countries, viewability and eCPM by country. Then a change stands out against your own history, not against some industry average that may not apply to your site.
How to confirm before you act
- Find the start date. Line up when the odd pattern began with anything you changed: new traffic sources, a redesign, a new partner.
- Segment it. Isolate the suspicious country, referrer or device and compare its engagement with the rest of your traffic.
- Check the pages. Bots often hit the same few URLs. Real readers spread out.
- Look at the money. If the segment earns far less per impression, buyers have probably already spotted it.
What to do once you find it
- Cut the source if it's a traffic vendor or a campaign you're paying for. Read the risks of buying traffic before you buy any more.
- Block obvious automation at the CDN or firewall level, for example known bad networks hitting you at high volume.
- Stop serving ads to it. Even if you can't stop every bot visiting, you can avoid showing ads to them. That's what a publisher-side filter is for.
- Tell your partner. If you're in an MCM relationship, your partner should hear it from you first, not from the platform.
At Bidora, IVT Shield classifies each visit as human, known crawler or suspicious, and withholds ads from the suspicious ones. You can read how it works on the IVT Shield page.
Talk to us about joining Bidora MCM. We send Manage Account invitations, Bidora's share starts at 8%, and IVT Shield is included for every site we monetise. Start on the Get started page or email partnership@bidora.one.