Full-Funnel Media for a Brand Selling in Two Places

Why two separate media plans waste money

The default setup for a brand selling in two places is a separate plan for Amazon (Sponsored Products, Sponsored Brands, Amazon-destination DSP) and another for the brand's own site (Meta, Google, email). Different people run them, different tools measure them, and nobody has a shared view of the customer.

That structure duplicates prospecting spend. The same shopper is reached twice, through two unrelated systems, each unaware of the other. It also throws away what Amazon DSP does best: an audience built from shopping signals that is useful for targeting whichever checkout the click lands on. Amazon's own guide to demand-side platforms describes DSP as a way to reach audiences on and off Amazon, which is why the audience does not have to end at an Amazon detail page.

A funnel mapped to destination, not to channel

Think of the funnel as three stages, each with a different sensitivity to where the shopper ends up.

• Awareness (streaming, video, broad display): destination matters little, because the goal is reach and recall, not a same-session sale. Amazon-owned inventory and off-Amazon exchange inventory both work, and either can carry link-out or Amazon-destination creative.
• Consideration (retargeting cart-viewers and abandoners, sponsored ads on non-branded search): destination starts to matter. An Amazon cart-abandoner is usually best retargeted back to Amazon, where friction is lowest for that shopper. A shopper who abandoned your own checkout is best retargeted back there.
• Purchase (branded search, high-intent retargeting): route to whichever destination that shopper's prior behavior points to, rather than applying one blanket rule to the account.

The rule underneath all three: route by the shopper's history, not by which team owns the budget.

A worked budget example across the funnel

For example, take a $40,000 monthly full-funnel budget:

• $15,000 to awareness (37.5%): streaming and display, split across Amazon-owned and off-Amazon exchange inventory, with creative pointed at whichever destination fits the message.
• $15,000 to consideration (37.5%): retargeting split by where the original interest came from. Amazon cart-abandoners go back to Amazon, site cart-abandoners go back to the site.
• $10,000 to purchase (25%): branded search on Amazon plus branded retargeting on the site.

This split is an illustration, not a benchmark or a recommendation for your brand. The point is the discipline: the unifying metric is not one blended ROAS across all $40,000. It is whether each stage's spend is routed to the destination where that audience segment has the lowest-friction path to buy.

How streaming fits this funnel shape

Streaming and connected-TV inventory is the placement most likely to be judged unfairly if you measure it like the rest of the funnel. It sits at the very top, reaching people who have not yet decided which destination, if either, they will buy from.

Judge it on new-to-brand reach and on how many of those shoppers later show up in consideration-stage audiences, not on purchase-stage return. Setting streaming its own target before launch, separate from retargeting, avoids cutting it for failing a test it was never built to pass. For context on what blended DSP looks like once all placements are counted, across our measured portfolio of 30 advertisers in July 2026, reMKTR saw a 6.04x return on ad spend, and 20.1% of attributed purchases came from shoppers new to the brand.

What a shared measurement view requires

You do not need a single unified dashboard on day one. You need three pieces of visibility that most two-team structures lack:

• Shared audience definitions. Both teams see each other's audiences at least at summary level, so a retargeting pool built on the Amazon side is not quietly duplicated on the DTC side.
• Shared underlying data. Both teams can reach the same performance data, even if they report on it separately, so a cross-channel question gets answered instead of debated from two incomplete pictures.
• A named owner for reconciliation. Without one, the shared view slides back into two silos within a quarter.

Amazon Marketing Cloud covers the Amazon half of this well, with path-to-conversion queries, overlap analysis and new-to-brand tracking across DSP and sponsored ads (see Amazon's Amazon Marketing Cloud page). What it cannot do alone is see your own site's conversions. Those come in through first-party upload or get reconciled against your site's analytics, with a deliberate bridge between the two.

A checklist for the two-place funnel

Run through these before the next budget meeting:

• Can you list every retargeting audience on both sides, with its source signal?
• Is there a named person who owns the overlap between the Amazon and DTC pools?
• Does each funnel stage have its own success metric, rather than one blended ROAS?
• Is streaming judged on reach and new-to-brand, not last-click?
• Are site conversions brought into the Amazon view (or the reverse) on a set schedule?
• Does creative at each stage point to the destination that fits that shopper?

Two or more "no" answers usually mean the funnel is really two plans sharing a logo. If your Amazon side is mostly search, our guide to advertising on Amazon covers how sponsored ads and DSP divide the work.

When the two channels start competing for the same budget

If leadership is debating whether the next incremental dollar should go to Amazon or to DTC media, that is usually the wrong question. The better one is which audience segment, at which funnel stage, has the clearest path to conversion right now, whatever platform it ultimately routes through.

Answering that takes the shared view described above, not a debate about which channel the company should prioritize as a matter of identity. The common mistake is letting organizational structure dictate the media plan: an "Amazon team" and a "DTC team" with separate budgets, creative and reporting, each optimizing its own slice. Good people on both sides can produce good individual results while the same shoppers get retargeted twice. Unifying the plan does not require cutting either team's budget. It requires seeing which shoppers each team is actually reaching.