Display Advertising

Display advertising places visual creative into space a publisher sells, and most of that space is now transacted programmatically. A page loads, an ad request goes to an exchange carrying signals about the placement and the visitor, buyers bid within milliseconds, and the winning creative renders. Targeting is assembled from audience lists, contextual signals about the page itself, and retargeting pools of people who already visited you. Nobody in that flow asked to see the ad, which shapes everything about how it performs.

Efficiency in display comes from three things: the quality of the placement, the share of the audience inside it you actually want, and how often each person sees the creative. Frequency is the lever most often mishandled, since too little registers with nobody and too much pays repeatedly to irritate someone. Cheap inventory is cheap for a reason, usually because attention there is low or the ad renders where nobody looks. Excluding weak placements raises unit cost while improving what the budget buys.

Display earns its place at two points: introducing a category to an audience that is not searching, and returning to people who started something and stopped. The second is where measurement becomes straightforward. A retargeting pool built from visitors who opened a scorecard quiz but never finished gives the banner one specific job, and completion of that quiz gives it a countable outcome. Suppression lists matter as much as targeting here, because paying to reach existing customers is pure waste.

Display suffers from a measurement problem in both directions at once. Last-click reporting undercounts assisted influence, while view-through attribution overcounts it by crediting impressions nobody consciously saw. Add ad blockers, weak viewability on some inventory and bot traffic, and small differences between placements become hard to trust. Display also cannot manufacture demand that does not exist: where a category is genuinely unfamiliar, a banner alone rarely creates enough interest to justify the spend by itself.

Judge display on reach, viewability and what the audience does afterwards rather than on click-through rate. The numbers worth watching are viewable impression rate, unique reach against the intended audience, frequency per person, and the rate at which retargeted visitors return and complete the action they abandoned. Cost per completed quiz from a retargeting pool is the clearest single figure available to you.

For upper-funnel display, use indirect evidence carefully. Watch branded search volume, direct traffic and the size of your retargeting pool during and after a flight, comparing exposed regions or audiences with unexposed ones wherever the platform allows it. Treat view-through conversions as directional only, and never add them into the same total as click-based conversions when reporting results.

The predictable failure is running display against the same conversion targets as search and cutting it when the numbers disappoint. Display audiences were not looking for you, so a comparison based on last-click conversions is decided before it begins. Give display its own objective, whether that is qualified reach, returning abandoners or growth in branded search, and hold it to that objective instead of a metric borrowed from another channel entirely.

The second is letting placements run unreviewed. Open networks include auto-refreshing pages, made-for-advertising sites and app inventory where impressions accumulate without any attention behind them. Pull the placement report on a schedule, exclude anything with high impressions and no downstream action, and keep that list current. Add a suppression audience of existing customers and recent converters so budget is not spent reaching people who already converted.