Twelve years chasing non-human traffic teaches you a handful of things that don’t fit on a dashboard. These are the ones I keep coming back to.

• Follow the money before you follow the traffic; the incentive explains the anomaly.
• Every metric is a claim. Ask who benefits if you believe it.
• The cheapest inventory is cheap for a reason you haven’t found yet.
• A defense with a single tell has an expiration date.
• Fraud rarely spikes; it blends. Smooth is suspicious.
• Vendors sell certainty; systems produce probabilities. Trust the ones who admit it.
• Averages hide the exact thing you’re paid to find.
• The absence of evidence is not evidence of absence.
• Seasonality is a fraud calendar; budgets tell the bots when to arrive.
• You cannot optimize your way out of a measurement problem.
• The best control is the one that costs the adversary more than the prize.
• Write it down publicly; the industry improves only when the blind spots are named.

None of these are clever. They are just expensive to learn any other way. If a single one saves you a quarter of chasing the wrong number, it earned its place on the list.