Amazon DSP Advertising · Case Studies

What Amazon DSP advertising actually did.
Five real accounts, with the method.

Most DSP case studies are a ROAS screenshot and a logo. These are the five results we get asked about most — anonymized brands from our managed portfolio, each with the situation, what we actually did, the numbers, and the honest caveats. The pattern to watch isn't the ROAS. It's what happens to organic sales when programmatic warms the audience.

Published 2026-08-19 · All brands anonymized · Results are what happened on these accounts, not a promise about yours

13.5% → 27.9%
Organic order share after DSP — still climbing 4 weeks after spend stopped
10 → 17.68
DSP ROAS, month 1 → month 4, at $0.21–0.44 CPCs
80% NTB
New-to-brand share of DSP-attributed sales on a prospecting-led program
4X in 3 weeks
ROAS on a ~$6K test — during Prime Week CPC inflation
~22X
Average ROAS on linkout campaigns — the most underpriced segment we buy

The case studies

Five accounts, five different jobs for DSP.

DSP isn't one tactic. Across these five accounts it played five different roles: organic-rank flywheel, cheap-reach engine, new-customer machine, momentum test, and an escape hatch from renting your own customers. That range is the point.

Organic lift · Incrementality

The DSP campaign that kept working after it stopped

A fitness publishing brand · top-of-funnel video program

The situation

A catalog brand with strong products and an expensive problem: PPC was carrying the whole business. Total ad cost of sales (TACoS) was running at 48% — nearly half of every revenue dollar going back into ads — and pushing sponsored bids harder was producing more spend, not more efficiency. The question the client actually needed answered wasn't "what's the DSP ROAS?" It was "can advertising make the organic business bigger?"

What we did

We built a top-of-funnel online-video program aimed at warming category audiences before they ever searched — not retargeting the already-convinced, but introducing the brand to shoppers earlier in the journey. Then we tracked the two numbers that reveal whether DSP is genuinely adding demand: organic order share and PPC efficiency downstream of the video exposure.

What happened

Organic order share went from 13.5% before DSP to 27.9% after — it doubled. PPC TACoS compressed from 48% to 37% and held, because the warmed audience converted more efficiently on every channel it touched. And the part that answers the incrementality question better than any attribution model: when DSP spend stopped entirely, organic share kept climbing for four more weeks. Re-labeled demand doesn't do that.

"Organic share doubled — and it kept climbing for four weeks after every dollar of DSP spend stopped."The persistence is the proof: paused campaigns can't take credit for growth that continues without them.
Efficiency · Underpriced reach

ROAS 10 → 17.68, on clicks the rest of the category ignored

A pet-treats brand · online-video prospecting and retargeting

The situation

A growing pet brand in a category where sponsored-ad clicks cost $5.00–6.60. At those CPCs, every incremental sponsored dollar was a knife fight, and scale meant accepting worse economics. The brand needed a second growth channel with different physics.

What we did

We moved the reach-buying into DSP online video, where the same shoppers cost $0.21–0.44 per click — inventory most of the category simply wasn't bidding on. The program blended prospecting into relevant pet audiences with retargeting of high-intent visitors, and we let audience data compound month over month instead of judging it on week one.

What happened

Month one closed at a ROAS of 10 — already strong. By month four, as the audiences matured and the retargeting pools deepened, ROAS reached 17.68, with over 12,000 units attributed to the program. Same products, same category, 10–25x cheaper clicks — the arbitrage was simply showing up where the auction wasn't crowded yet.

"We bought the same shopper for a quarter that competitors were paying six dollars to reach."DSP's edge in 2026 is mispriced inventory — it lasts exactly as long as the category stays asleep.
Growth · New-to-brand

80% of DSP sales came from customers the brand had never reached

A medical recovery-device brand · prospecting-led program alongside maxed-out PPC

The situation

An established device brand with sponsored ads already tuned hard — strong TACoS, high conversion, disciplined branded spend. The ceiling wasn't efficiency; it was audience. Nearly everyone searching the category keywords had already seen the brand. Growth had to come from people who weren't searching yet.

What we did

A prospecting-led DSP program: in-market and lifestyle audiences chosen for fit with the product's use case, with the buying anchored to the brand's real customer-acquisition math across channels — the founder knew exactly what a new customer was worth to them elsewhere, so we bought to that number rather than to a vanity ROAS.

What happened

Eighty percent of DSP-attributed sales were new-to-brand — genuinely new customers, not re-labeled repeat purchases — with the program reaching roughly 2.5 ROAS by week three and climbing. For context on how unusual that mix is: across our managed portfolio, DSP prospecting audiences typically run about two-thirds new-to-brand, and that's already the highest NTB concentration of any ad type we buy.

"When sponsored ads are maxed out, the next customer isn't searching for you yet. DSP is how you go meet them."Prospecting ROAS looks modest next to retargeting — because it's doing a harder, more valuable job.
Testing · Momentum

4X ROAS in three weeks on a $6K test — during Prime Week inflation

A mushroom supplement brand · first DSP program, deliberately small

The situation

A supplement brand's first time on DSP, with a founder who was rightly cautious: modest budget, short leash, and a live question three weeks in — "can we pause this?" It's the most common moment in every first DSP engagement, and what the data said mattered more than what we said.

What we did

We sized the test small on purpose — roughly $6K over three weeks — and structured it to produce a readable signal fast: tight audiences, clean creative, and honest measurement against the noisiest possible backdrop, because the test happened to run straight through Prime Week, when CPCs inflate across the entire platform.

What happened

ROAS quadrupled within three weeks despite the event-driven CPC surge. Just as important was what the test demonstrated about momentum: DSP audiences take time to build, and the learning compounds — pausing at week three would have forfeited both the $6K of learning and the trajectory. The brand kept the program on.

"Three weeks and six thousand dollars was enough to know — even in the most expensive week of the quarter."A well-built small test beats a big vague one. Size the test to the decision, not the ambition.
Own the customer · Linkouts

~22X ROAS on linkouts — the segment almost nobody is bidding on

A supplements brand (and a portfolio-wide pattern) · DSP linkout campaigns to owned channels

The situation

The complaint we hear from every DTC-first founder: Amazon owns the customer relationship. No emails, no remarketing list, no LTV — every repeat purchase has to be re-bought with ads. For brands whose store customers are worth multiples of a one-off Amazon order, that's the single most expensive fact about the channel.

What we did

DSP linkout campaigns — programmatic placements that drive shoppers to the brand's own site instead of the marketplace listing. It's a corner of DSP most buyers ignore, which means thin competition and quiet auctions. We treated it as an owned-audience acquisition channel: the sale lands where the brand keeps the customer, the email, and the lifetime value.

What happened

Linkout campaigns have averaged roughly 22X ROAS — the best-performing segment we buy, by a wide margin. The mechanism is no mystery: barely contested inventory priced accordingly, pointed at the one destination where a customer is worth their full LTV instead of a single order. It will not stay this quiet forever.

"The cheapest customer on Amazon is the one you send to your own store."Linkouts are the answer to 'Amazon owns my customer' — and right now they're priced like nobody knows.

How we read results

The honest caveats — read these before you extrapolate.

These are real accounts, anonymized, and they are illustrations — not promises. Every category, catalog, and margin structure is different; a result that took three weeks on one account can take three months on another, and some programs we recommend against entirely (if your ad budget is under roughly $10–15K/month, DSP is usually not where it should go — we'll say so).

ROAS is the weakest number on this page, on purpose. Attributed ROAS flatters retargeting and undersells prospecting. The results we weight most are the structural ones: organic order share, new-to-brand percentage, branded search growth, and PPC efficiency downstream of DSP exposure. That's why the fitness-publishing case leads this page — a doubling of organic share that persisted after spend stopped is harder evidence than any attribution screenshot.

Incrementality gets measured, not asserted. Where clients want proof, we build it: Amazon Marketing Cloud reconciliation so sponsored ads and DSP stop double-counting each other, holdout designs, and de-duped comparison against internal revenue. If a vendor won't show you the method before the number, discount the number.

Questions

What brands ask before starting DSP.

How long does Amazon DSP advertising take to work?

Expect an early signal within about three weeks and compounding results over months. The $6K test above hit 4X ROAS in three weeks — during Prime Week. The pet-treats program went from ROAS 10 to 17.68 between month one and month four as audiences matured. The one reliable way to lose is pausing early: DSP builds momentum, and a pause forfeits the audience learning you just paid for.

How should I evaluate DSP performance beyond ROAS?

Watch organic order share, branded search volume, new-to-brand percentage, and whether your PPC gets more efficient as DSP warms the audience. ROAS alone misjudges DSP in both directions — it flatters retargeting and undersells prospecting. The strongest single indicator in our portfolio: organic share that rises with DSP and keeps rising when spend stops.

What's the minimum budget for a managed DSP program?

Roughly $10–15K/month for a sustained program that produces readable signal and room to optimize. Below that, the same dollars usually work harder in sponsored ads or other channels — and we'll tell you that rather than run a program too small to succeed. Structured tests can start smaller, as the $6K case shows, but tests are for making a decision, not for running a channel.

Is DSP incremental, or would those sales have happened anyway?

The right question — and the reason we lead with the persistence evidence. When organic order share doubles and continues climbing for four weeks after all DSP spend stops, that growth cannot be re-labeled demand. For formal measurement we use AMC reconciliation, holdout designs, and de-duped revenue comparison. Method first, then numbers.

What does reMKTR charge?

A commission on managed spend that scales down as your spend scales up — no retainers, no creative fees, no long-term contracts, and spend throttled to your ROAS goals so exposure is limited to what's working. The Growth Leak Audit below is free and shows you what a program would look like on your account before you commit anything.

Free · No call required to claim it

Find out what your account is leaking first.

Before anyone should sell you DSP, you deserve to know where your current spend is bleeding. The Growth Leak Audit reviews your ads, organic share, and fees the way we reviewed the five accounts above — and hands you the findings whether or not you ever work with us.

Claim Your Free Growth Leak Audit → No retainers. No creative fees. No long-term contracts.