Case study: an anti-fraud SDK killed click injection and saved 14% of budget

Case study: an anti-fraud SDK killed click injection and saved 14% of budget

What it is: A forex app-install campaign had suspiciously high install-to-FTD drop-off. An anti-fraud layer (timestamp-based click-injection detection) revealed publishers hijacking last-click attribution.

Best for: Mobile buyers running broker-app installs through multiple sub-publishers.

What was done:
— Analyzed click-to-install time distribution across 18,000 installs
— Flagged installs where click landed under 10 seconds before install completion
— Cross-checked the worst sub-publishers against FTD rate

Outcome:
— 14% of paid installs showed click-injection signatures
— Three sub-publishers accounted for 80% of the fraud; they were cut
— Reallocating that 14% to clean sources lifted true FTD volume 11%

Pros:
— Time-distribution analysis needs no SDK if you have raw timestamps
— Cutting fraud sources improved attribution accuracy network-wide

Cons:
— Sophisticated fraud randomizes timing to beat the 10-second rule
— False positives on slow devices/networks need a manual review pass

Who should skip this: Web-only forex funnels — click injection is a mobile-attribution problem.

Verdict: Watch the click-to-install clock.