Media quality is a performance variable

If you buy programmatic on price, some of your cheapest impressions are your most expensive ones.

On certain Toyota and Lexus campaigns I ran, invalid traffic approached ~20%. That is budget spent on traffic that was never a real shopper. With tighter inventory controls and verification, those problem campaigns came down to roughly ~3%.

The fix is simple to say: treat media quality as part of the price, not a compliance report. Block weak inventory, buy cleaner supply, and read quality next to cost every time you optimize.

If you don’t know your invalid-traffic rate by campaign, book a 30-minute call. That is usually the first place I look.

Cheap is easy to see. Waste isn’t.

A low CPM looks like a good deal on every dashboard. Invalid traffic still produces impressions, and sometimes clicks. So poor inventory doesn’t look like a problem. It looks like efficiency, and it wins budget for exactly that reason.

That is why I treat fraud, viewability, inventory quality, and brand safety as economic variables. An impression nobody sees isn’t cheap at any price.

Where this came from

I led programmatic media and measurement for Toyota and Lexus as Associate Marketing Director at interTrend Communications. It was a mature program. Nothing was fundamentally broken. The job was to build a system that could keep finding the next 5%, and media quality was one of the places that 5% was hiding.

What it looks like in practice

• Verification feeds decisions. IAS and DoubleVerify controls, Nielsen and other validation, invalid-traffic monitoring, and brand-safety controls went into the same optimization process as audiences and bids.
• Inventory is chosen, not accepted. Inventory and app blocking, exclusion strategy, and inventory QA were routine work. PMP and curated inventory came out roughly 30–40% better than weaker open exchange on viewability or fraud.
• Quality sits next to cost. Viewability ran at approximately 78–85%+. Read beside CPM and CPC, it shows when a cheaper line item isn’t actually cheaper.
• The numbers have to be trusted. CM360 / Floodlight implementation and QA, consistent naming, and manual checks where automated reporting couldn’t be trusted are what make the rest readable.

Four questions for your own buys

• Is verification data in the same view as cost and conversions, or in a separate report nobody opens?
• Do you know which campaigns carry most of your invalid traffic, or only the account average?
• When a line item gets cheaper, do you check whether its viewability or fraud rate moved too?
• Do you compare private or curated deals with open exchange on quality, not only on price?

This isn’t specific to automotive. Any program that buys at scale is also buying a quality mix, whether anyone measures it or not.

If you want a second set of eyes on yours, book a 30-minute call.

The figures here are approximate historical ranges across different campaigns and windows, not one controlled test, and the ~20% → ~3% result applies to problem campaigns only. This was agency work. The full context is in the Toyota and Lexus case study.