Bot traffic used to be easy to spot: visits from server farms, ancient browsers, the same device a thousand times. Today a lot of it comes through residential proxies, which route each visit through a real home connection in the target country. Each visit looks clean on its own. Together, they look like nothing a real audience produces.
This article explains what those patterns are, how ad networks look for them, what happens to a site that sends them, and why an honest site, even a small one, has nothing to fear. We have seen the pattern on our own network, so this is written from the inside.
Why proxy traffic exists at all
Someone buys cheap traffic, or runs a script, and points it at a site. The goal is usually to make the site look bigger than it is, to sell ad space, or to collect ad revenue for visits no human made. Residential proxies make it harder to catch, because the IP addresses belong to normal internet providers in the right country, not to a data center.
For an advertiser, these visits are pure loss: nobody sees the ad and nobody buys. That is why networks spend real effort removing them.
The signatures networks look for
No single signal proves anything. A network looks at combinations, over enough volume to be sure. In general terms:
- An IP address proves nothing on its own. Residential proxies give each visit a clean home connection in the right country, and a frequency cap already makes most visits in a report look unique. A network that judged on IP addresses alone would flag honest sites and miss the farms.
- Browsers that change on every visit. A real audience repeats itself: the same few phone models and the same current browser versions, thousands of times. Generated traffic tends to look a little too varied.
- Visits with no action behind them. Real people sometimes sign up or buy, even on weak placements. A lot of volume where nobody ever does anything, on offers that convert elsewhere, says nobody was there.
- An audience that does not look like an audience. Real visitors come from many places, at uneven times, with all the small inconsistencies of people. Scripted traffic is often too uniform where people vary, and too varied where people repeat.
Networks also run checks on every visit. On Lana Network they fall into four kinds: bots and automation, non-human and data-center traffic, abnormal behaviour, and source quality. Our traffic quality page explains each kind, without the exact checks. A visit that fails is recorded as invalid with its reason and not paid.
How the pattern check works on Lana Network
The per-visit checks catch the crude traffic. For the rest, we look at each zone, each site and each publisher account over a rolling period, and combine several signals. A site is held when the combination is clearly not human, not when one number looks odd.
We do not publish the exact thresholds. Publishing them would only tell a traffic seller how far to tune a script. What we can say is that each rule needs real volume before it fires, and that a normal audience, even a strange one, does not match the combination.
What it costs a publisher
For the site sending the traffic, the cost is not a warning email:
- Invalid impressions are not paid. The advertiser gets that money back at settlement, so the revenue shown during the day can disappear when the day is counted.
- The site and its zones are paused. The tag stops serving, so honest visitors on the same site stop earning too.
- Payouts are frozen. The publisher account is put on hold and any open payout request waits until a person reviews the case. On Lana Network this shows as "On hold", never "Rejected", because a review can lift it.
- New sites wait. A publisher on hold cannot put another site live on their own.
Other networks go further, with account closures and clawbacks of past payouts. The 7-day hold explained in our article on how a network pays its publishers exists precisely so the money for this traffic never leaves.
Why a real site does not trigger any of this
If your visitors are people, the patterns point the other way on their own: real audiences are varied in the ways people are, repeat themselves in the ways devices do, and a few of them convert. You do not need to do anything special.
Things that can put an honest site at risk, and that are easy to avoid:
- Buying cheap traffic to look bigger. Traffic exchanges and very cheap "visitor" packages are a common source of proxy visits. If you buy traffic for your site, tell your network which source, before it runs.
- Testing your own ads. Clicking your own ads or asking friends to do it creates exactly the patterns networks remove.
- Iframing or auto-reloading pages. Anything that loads ads without a person looking at the page.
Our guide to traffic quality rules lists what the network accepts and what it does not.
If your site is put on hold
- Read the note on your site card: it says the site is on hold for review.
- Write to us with what you know: your main traffic sources, a screenshot of your analytics sources and countries, and whether you bought any traffic recently.
- A person reviews the case. If the traffic is real, the hold is lifted and the held earnings follow the normal schedule.
If you are setting up a site for the first time, start with adding and verifying your site. Advertisers who want the same topic from their side can read how to buy adult traffic.
Frequently asked questions
What is residential proxy traffic?
Visits routed through real home internet connections, often in the target country, so each one looks like a normal visitor. Networks detect it from patterns across many visits, not from the IP address alone.
Can a small site be flagged by mistake?
The pattern rules need real volume and several signals at once before they fire, so a small site with a normal audience does not match them. If a site is held, a person reviews it before anything is final.
What happens to my earnings if my site is put on hold?
The site and its zones are paused and open payout requests wait for the review. Impressions found invalid are not paid; earnings from valid traffic follow the normal schedule once the hold is lifted.
Does Lana Network publish its detection thresholds?
No. The checks applied to every visit are listed on our traffic quality page. The thresholds of the pattern rules stay private, because publishing them would only help traffic sellers tune their scripts.