LFH Lab Start a project
Work/BotAudit Live · MY PRODUCT

BotAudit

Forensic click verification for advertisers. Every paid click is scored human, suspicious or bot — and the flagged ones come with the evidence trail ad platforms ask for in an invalid-traffic dispute.

BotAudit screenshot
The problem

I audited my own ad spend and 28% of the clicks weren’t human. The tools that tell you this are priced for enterprises and stop at a number — a scary percentage with nothing behind it. Platforms don’t refund percentages. They refund claims, and a claim needs per-click evidence: IP, click ID, campaign, and a reason.

What I built

A sub-2KB beacon that classifies every visit as it lands — automation flags, datacenter and proxy IPs, and input physics that catch bots driving real Chrome. Three honest buckets instead of one number, wasted spend priced at your real CPCs as a range, and one button that exports a Click Quality Form-ready CSV. Agencies can put their own name on the report.

12
bot signals checked on every visit — automation flags, input physics, datacenter IPs, behaviour.
5
ad platforms auto-detected from their click IDs: Google, Meta, X, TikTok, Bing.
60 days
the dispute window platforms typically accept — retained in full on every paid plan.
Demo
90 seconds, no audio needed
Screen recording coming
A 90-second walkthrough is being recorded.
Until then, the live site is one click up.
What I'd do differently

I shipped detection before I shipped the export. For three weeks people could see their bot rate and do nothing about it, and the feedback was all the same question: “so how do I get the money back?” The claim CSV should have been in the first release.