Programmatic Display Buying, From Budget to Delivered Impression

The buying process, in the order it actually happens

A programmatic display buy starts with a goal, not a platform — reach, consideration, or conversion — because that goal decides most of what follows. Next comes budget and flight dates, then targeting (who the campaign should reach), then creative (what they'll see), and finally the deal type: open exchange for maximum reach and flexibility, private marketplace for curated, higher-quality inventory at a premium, or programmatic guaranteed for a fixed volume at a fixed price on a specific publisher. Only after all of that is set does the DSP start bidding — and once it does, the buying process doesn't stop, because pacing, optimization, and reporting keep running for the length of the flight, adjusting where the budget goes based on what's actually converting.

The order matters more than it looks. Teams new to programmatic buying often reverse it — picking creative first because it's the most tangible, visible piece of the process — which leads to building assets before knowing which sizes, formats, and deal types the campaign actually needs. Setting the goal and deal type first means the creative brief can be written against real constraints from the start, instead of getting redesigned once the targeting and inventory decisions are finally made.

The decision most buying guides skip: setting a different target per inventory type

The single biggest mistake in programmatic display buying isn't in the setup steps above — it's what happens after launch, when every inventory type in the plan gets judged against the same return target. Streaming and connected-TV inventory is the clearest example. In a 31-day pull across 27 advertisers in our own Amazon DSP book, streaming inventory returned 0.77x — the worst return-on-spend number on the entire report — while delivering the highest new-to-brand rate of any inventory type, 56.3%. Streaming CPMs ran $15.51 against a $3.71 portfolio blend. If that line item is measured against the same target as retargeting, it gets cut every time, and cutting it removes the part of the plan that was finding first-time buyers no other inventory type reached.

Our standing practice is to set streaming a return target around 1.0 going in, and judge it on video completion rate and branded-search lift instead of ROAS, because Amazon's own attribution structurally undercounts streaming: shoppable features only fire for logged-in Prime members, so a real share of streaming-driven demand never gets attributed at all. Two operational details matter here specifically: Prime Video ad inventory and free ad-supported channel inventory behave differently and shouldn't sit in one line item together, and completion rate tells you whether the creative is landing well before the sales data can be trusted.

A worked example: setting the right target before the budget goes out

Say a $40,000 monthly display budget splits three ways: $15,000 to open-exchange retargeting, $15,000 to open-exchange prospecting, and $10,000 to streaming. Setting a single 4x ROAS target across all three and reviewing weekly, the streaming line will look like the obvious cut candidate within the first reporting cycle — it's structurally never going to hit 4x given the attribution gap and the CPM described above. Setting three separate targets going in — 6x+ for retargeting, 3x for prospecting, and roughly 1.0x plus a completion-rate and branded-search check for streaming — means each line gets evaluated against what it's actually capable of producing, and the decision to cut or scale each one is based on whether it's meeting its own bar, not a bar built for a different inventory type entirely.

The other place buying goes wrong: choosing a deal type by default

A closely related mistake is running every buy through the open exchange by default, because it's the fastest to set up, without asking whether the campaign's goal actually needs a private marketplace's curated inventory or a guaranteed deal's fixed placement. Open exchange is the right default for reach and testing. It's the wrong default for a brand-sensitive launch or a campaign built around one specific, high-value publisher relationship — those need the control a PMP or PG deal provides, even at a higher CPM.

What to check before scaling or cutting any line item

Check whether the line item has its own target, matched to what that inventory type can realistically produce, before comparing it to any other line in the plan. Check attribution structure second — an inventory type with a known measurement gap, like streaming, needs a secondary metric alongside ROAS, not instead of it. Check the deal type third — a weak result on the open exchange might improve on a curated PMP at a higher floor price, which is a different fix than a targeting or creative change.

Where Dr. DSP fits

Dr. DSP is Amazon DSP run as a managed product by Full Circle, which has managed more than $500M in revenue across 100+ brands. Setting a separate, honest target for each inventory type before the budget goes out is a small process decision that keeps a plan from quietly cutting its best new-customer channel because it was judged against the wrong number. A reader who never buys anything from us should still leave this page knowing that one ROAS target for an entire programmatic display plan is a planning shortcut, not a best practice.