Viewability work that starts with measurement coverage

Measured viewability is the wrong first number. If measurement coverage sits at seventy-one percent, then twenty-nine percent of what you bought has no viewability status at all, and every optimisation you run on the remainder is being decided by a sample that selected itself. Coverage comes first. On in-app accounts, moving coverage from the low seventies to above ninety usually changes the reported viewable rate on its own, sometimes downward, which is the point of doing it in that order. The trade I will not make: Viewable rate can be bought. Sticky footers, 320x50 slots anchored to the screen edge, thirty-second auto-refresh and player dimensions chosen for the metric will all take you to seventy-eight percent, and they take your cost per outcome with them. What I optimise is the cost of a viewable impression that also produced something you care about — an install, a registered event, a subscription start. Where the two objectives disagree I will say so and let you make the call. Every engagement here begins by agreeing which outcome we are attached to, because without one this discipline collapses into buying whatever renders most cheaply. Where the points actually come from: Mediation coverage gaps, first. Then creative format mix, since a 300x250 in feed and a 320x50 pinned to the bottom of a screen are not the same purchase whatever the plan calls them. Then refresh policy — publishers refreshing under thirty seconds inflate impression counts and depress everything downstream, and that is a supply decision, not a creative one. Then video: player dimensions, autoplay and sound state, and whether your VAST is being wrapped a third time by an intermediary that drops the verification node and turns the impression unmeasurable. In-app specifics: Open Measurement SDK presence is not the same as OM SDK working. It has to be initialised by the app, exposed by the mediation adapter, and reachable by the verification script for the whole session. I test with a controlled buy on your own inventory rather than trusting a mediation vendor's compatibility matrix, which describes the version they shipped and not the version a publisher integrated in 2022 and never touched again. Excluded: Connected TV. Reported viewability there is close to one hundred percent by construction, and the real work is attention and completion, which I do not sell. Creative production. I will tell you which formats to stop running and why. Somebody else builds the replacements. Guarantees. I will not commit to a viewable rate, and I would be careful with anyone who does. The number depends on your creative, the publishers you are willing to buy and the standard you chose, and I control none of those. Poor fit: If the mandate arrived as a target percentage with a date attached, this will be an uncomfortable engagement. I will hit it, I will write down what it cost, and that document will exist afterwards. If most of your spend is direct-sold with contracted viewability guarantees, the argument to have is contractual, not technical. If the app team cannot ship an SDK update inside a quarter, half of what I find is unfixable this year. Worth knowing before you start rather than in week five. How it runs: two weeks of reading before any change, then changes in tranches with a fixed measurement window between them, because four simultaneous changes teach you nothing. Reporting is one page — coverage, viewable rate, cost per outcome, and what moved which. English and Mandarin, Singapore hours, and I take the platform calls myself.