Bot and click farm detection
Fake clicks come in a few shapes. Each one gives itself away.
Invalid traffic is rarely random. It is produced by an operation, and operations leave patterns: in their timing, their networks, their devices and the way they move. Maddet is built to read those patterns.
Automated bots
Software that loads pages and clicks ads without a person behind it, from simple scripts to full headless browsers.
What the engine looks for
- Abnormal click velocity from a narrow IP range
- Headless browser traits and automation framework fingerprints
- No pointer movement, or perfectly linear movement, before the click
- Machine-regular spacing between successive clicks
A single hosting IP produces 40 clicks per minute on one ad group, each landing 1.5 seconds apart with no mouse events.
Click farms
Real phones and computers run in bulk by people or rigs, rotating identities so each click looks like a new visitor.
What the engine looks for
- Many devices clustered in one /24 subnet or a small set of ranges
- One device fingerprint appearing behind many IPs or cleared cookies
- Synchronised bursts, often outside your audience's waking hours
- Sessions that never progress past the landing page
Eighteen devices on the same subnet click three campaigns inside four minutes, then fall silent together.
Competitor click abuse
Someone with a reason to drain your budget, clicking your ads on high-cost keywords until your daily cap runs out.
What the engine looks for
- The same device returning to click the same advertiser repeatedly
- Clicks concentrated on your most expensive keywords
- Activity during working hours, often from one city
- Zero engagement after arrival, repeated visit after visit
One device clicks the brand campaign 11 times over 36 minutes, always on the highest-bid keyword.
Also detected
The quieter categories that add up.
Not every invalid click comes from a dramatic attack. These patterns cost less per event and far more in aggregate.
Low-quality placements
Display and network sites that inflate clicks through layout tricks or bought traffic.
Tell Placement history, accidental-tap rates, instant bounces
Proxy and geo masking
Traffic routed through proxies to appear inside a targeted region.
Tell Proxy exits, timezone and language mismatch
Data center traffic
Clicks originating from cloud and hosting networks rather than homes or phones.
Tell ASN type, uniform device traits, off-hours bursts
Flag it, don't fake certainty
Borderline is a verdict too.
A university campus, a busy office and a mobile carrier can all put many real people behind one IP. A detection system that blocks on a single signal will punish them and your campaign reach with them.
Maddet only blocks when several independent signals agree. Clicks with mixed evidence land in a review band, visible in your dashboard with the signals that fired, so the call stays with you.
Exposure estimate
If there is an operation clicking your ads, it has a pattern. Let's find it.
Enter what you spend on paid media each month. The slip shows the range typically lost to invalid traffic at that level. The audit replaces the range with your real figure.
- Lower band (7.000000000000001%)
- ₹35,000
- Typical (14.000000000000002%)
- ₹70,000
- Upper band (22%)
- ₹1,10,000
- Typical, over 12 months
- ₹8,40,000
This is an illustrative estimate based on typical invalid-traffic rates, not a live scan of your account. Connect your account for your actual numbers.
Swap the estimate for the real figureGet a free traffic audit