Amazon DSP: When It's Worth Turning On

Amazon says its managed DSP service “typically requires a minimum spend of $50,000.” That one line tells you most of what you need to know about who Amazon DSP was built for. It wasn’t built for a seller doing $15,000 a month with two ASINs. It was built for brands that already have search ads working and want to reach shoppers before and after they search, on Amazon and across the wider internet. On September 29, 2026, Amazon also renamed and folded DSP into a new console, which has confused a lot of sellers. This guide explains what DSP does now, what changed, and the plain test we’d use to decide whether you should touch it yet.

What Amazon DSP actually is (and what it’s called now)

DSP stands for demand-side platform. In plain words, it’s a tool for buying ad space automatically, a few thousand impressions at a time, across lots of places at once. Amazon’s own DSP page lists display, video, audio, and streaming TV ads, shown on Amazon.com, Prime Video, Twitch, Fire TV, and Thursday Night Football, plus thousands of third-party websites and apps.

That’s the first big difference from the ads most sellers know. Sponsored Products, Sponsored Brands, and Sponsored Display mostly live inside Amazon’s own pages. DSP can follow a shopper onto a news site, a recipe app, or a streaming show. And you don’t even have to sell on Amazon to use it. Amazon says DSP is open to advertisers who sell on Amazon and to those who don’t.

The second difference is how you pay. Sponsored Products charges you when someone clicks. DSP is mostly bought by impressions, usually priced per thousand views (CPM). So you pay for the ad being seen, whether or not anyone clicks.

Here’s what changed in late September 2026. At Amazon’s unBoxed event on September 29, Amazon merged the ads console and DSP into one product it calls the Amazon Ads Agent. Trade press that covered the event (AdExchanger and PPC Land) reported these changes:

1. The name “Amazon DSP” is being retired. The programmatic side is now called DVA, short for display, video, and audio.
2. DVA+ is the new default campaign type. It’s an AI-run campaign. You pick a goal, choose your products or brand URL, review creatives Amazon generates, and set a budget. Amazon’s software handles targeting, bidding, and placement.
3. The old DSP controls still exist. They sit under “advanced settings,” where you can set your own audiences, publisher lists, and frequency caps.
4. Full-Funnel Campaigns is a separate new product aimed at smaller advertisers who want to try streaming TV and audio without learning every channel.
5. Availability is limited for now. PPC Land reported DVA+ in an extended closed beta in the US only, with open beta expected in late October 2026 and general availability across 35 markets targeted for Q1 2027. During the closed beta, payment is by invoice only, and Amazon plans card payments from Q1 2027.

So if you’re reading an older guide that walks through “the DSP console,” the screens it describes are changing. The question this post answers hasn’t changed, though. Should a seller at your stage pay to show ads by the thousand, off Amazon and on, to people who aren’t searching for your product right now?

Why most sellers shouldn’t turn it on yet

The short version: DSP spends money on people who weren’t looking for you. That can work very well for a brand that already converts its search traffic. It burns cash for a brand that doesn’t.

I’ll be straight about where this post comes from. We manage Amazon stores doing $200K a month for paying clients, and almost all of that ad work is Sponsored Products, with Sponsored Brands and Sponsored Display added once a product is stable. This post isn’t built on our own DSP spend data, because we don’t have a client case to show you. It’s built from Amazon’s own DSP pages, the September 2026 announcements, and the same math we use on search ads every week. That math is enough to show why DSP is a later step for most of the sellers who read this blog.

Look at the money first. If the managed service usually needs $50,000, that’s off the table for nearly every private label seller under a few hundred thousand dollars a month. Self-service has no published minimum from Amazon, but self-service means you (or an agency with a DSP seat) run it yourselves. Agency guides published in 2026 put a practical starting budget somewhere between $5,000 and $15,000 a month, because smaller budgets don’t give the system enough results to learn from. Amazon hasn’t said whether the new DVA+ campaigns carry any minimum. Until it does, plan as if a real test costs thousands, not hundreds.

Now compare that to what a normal launch spends. On our vegetable chopper launch, the whole PPC budget for the launch period was about $3,000, spent at $50 a day over roughly 60 days. That $50 a day is the floor we use for a launch, because below it a new listing doesn’t get enough impressions to rank. A single month of a modest DSP test can cost more than that entire launch budget. That’s the gap you’re looking at.

There’s also a math problem that has nothing to do with minimums. Most of the useful DSP audiences for a small seller are retargeting audiences: people who viewed your listing and didn’t buy. That pool is only as big as your traffic. Say your listing gets 5,000 sessions a month and converts at 10%. About 4,500 people left without buying. If you show each of them 5 ads over the month, that’s 22,500 impressions. At an example CPM of $6, that’s about $135. You can’t spend $5,000 a month retargeting 4,500 people without either hitting them dozens of times or widening the audience to strangers. And once you’re buying strangers, you’re doing brand awareness, which is the hardest spend in all of Amazon advertising to measure.

How to tell if you’re ready for DSP

Before any DSP talk, a brand should pass a simple readiness check. These are the signals we’d want to see on an account before we’d even suggest a test. None of them is an Amazon rule. They’re operator judgment, and you can check every one of them yourself in Seller Central and your ad reports.

1. Sponsored Products is profitable and stable. We treat a product as healthy on paper when ACOS sits around 20% and TACOS lands in the 10% to 15% range, with a normal returns rate. If you’re still fighting 40% to 60% ACOS on search ads, DSP will only add a second leak.
2. TACOS is steady or falling over 3 months. That tells you paid traffic is turning into organic sales. If TACOS is climbing, your ads are carrying the product, and DSP won’t fix that.
3. The listing converts. If the shoppers who search for your product and click don’t buy, the shoppers DSP brings from other websites won’t buy either. They’re colder than search traffic, not warmer.
4. Ratings sit at 4.5 stars or higher. Retargeting sends people back to the same reviews they already saw. If those reviews made them leave once, they’ll likely make them leave again.
5. You sell more than one product, or a product people rebuy. Consumables and multi-SKU brands can earn back the cost of a new customer across several orders. A single one-off product has one shot to pay for the impression.
6. Your ad spend is big enough that a test is a small slice of it. A rough rule we use: if a DSP test budget would be more than 10% to 15% of your total monthly ad spend, you’re early. That’s judgment, not a benchmark, but it keeps a test from eating the budget that’s already working.
7. You have creative. DSP and DVA run display, video, and audio. Amazon can now generate creatives for you, but you should still have product video and clean lifestyle images worth showing to a stranger.

If you tick 6 or 7 of these, a small, well-measured test can make sense. If you tick 3 or fewer, every dollar you’d put into DSP will do more work inside Sponsored Products or on your listing.

How a first DSP test should run, step by step

If you’ve passed the readiness check, the worst thing you can do is turn on a big, broad campaign and judge it a month later by the headline ROAS number. A first test should be small, narrow, and built so you can tell whether it added sales you wouldn’t have made anyway.

Step 1: Pick one goal, not three

DSP can chase awareness, consideration, or sales. A first test for a private label seller should usually chase one thing: bringing back shoppers who were already close to buying. Under DVA+, that roughly lines up with Performance+, the side Amazon describes as aimed at sales and conversions. Brand+ is the awareness and video side. Mixing both goals in one test makes the results impossible to read.

Step 2: Start with the warmest audiences

The audiences that tend to pay back first are the ones closest to a purchase:

• Detail page viewers who didn’t buy, usually over the last 14 to 30 days.
• Shoppers who added to cart and left.
• Past buyers of a consumable, timed to when they’d run out. If your product lasts about 30 days, a reminder around day 25 to 35 is the idea.
• Viewers of close competitor products, but only once the first three are working.

Cold “lifestyle” or broad in-market audiences come last, if at all. That’s where the money disappears fastest.

Step 3: Cap how often people see you

Frequency is how many times one person sees your ad in a set period. With small audiences, an uncapped campaign will keep hitting the same few thousand people. The advanced settings let you set a frequency cap. For a first retargeting test, a few impressions per person per day is plenty. If you use the AI-run DVA+ setup, check what controls it gives you, because you’re handing those decisions to Amazon’s software.

Step 4: Set a fixed budget and a fixed window

Pick a total test budget you’re fine losing, and a window of 4 to 8 weeks. Write both down before you start. DSP reporting counts sales from people who saw an ad, not only people who clicked one, so results tend to look better as the weeks go on. Deciding the length in advance stops you from either quitting in week one or running forever because “it’s still learning.”

Step 5: Measure the whole brand, not just the campaign

This step separates a real test from a vanity report. Campaign ROAS will always look good on retargeting, because many of those shoppers would have come back anyway. So check three things at the brand level:

1. TACOS before, during, and after the test. If total sales didn’t move and total ad spend went up, TACOS rises and DSP didn’t add much.
2. New-to-brand numbers. Amazon reports how many orders came from first-time buyers of your brand. Retargeting your own past buyers looks great and adds little.
3. Organic sales and rank on your main keywords. If DSP is bringing people back to search for you, you should see it here too.

If you can, hold back part of the audience as a control group that sees no DSP ads. Then compare. That’s the cleanest way to see what the ads actually added.

Step 6: Decide, then scale slowly

At the end of the window, you have three honest answers. It added sales at a TACOS you can live with, so you raise the budget in steps of 20% to 30%. It brought in new customers at about what they cost, so you keep it small and only on the warmest audiences. Or it added nothing beyond what you’d have sold anyway, so you switch it off and put that money back into search ads and the listing. All three are useful results. The only bad result is not knowing which one you got.

Common DSP mistakes

Most DSP stories that end badly end badly for one of a handful of reasons. None of them is about the platform being “bad.” They’re about turning it on at the wrong time or reading it the wrong way.

Using DSP to rescue a weak product. This is the big one. We managed a generic wood-therapy massage tool for a year. Every new tactic we tried, from PPC keyword work to ASIN targeting to listing refreshes, lifted sales for a while, then sales slid back. Competitors selling the same physical item undercut on price, and shoppers bought the cheapest one. More reach wouldn’t have changed that. DSP would have shown the product to more people who’d then pick the cheaper copy. If a product can’t win the click and the sale in search, showing it on Twitch won’t save it.

Ignoring returns. On our vegetable chopper launch, the ad numbers looked fine for a while. ACOS around 20% and TACOS in the 10% to 15% range. But the whole category had top sellers rating below 4.5 stars, and returns ate the margin. DSP brings in colder buyers who are often more likely to be disappointed. If your category runs a high returns rate, every extra sale DSP buys can cost you more after the return.

Judging it by campaign ROAS alone. Retargeting your own detail page viewers and past buyers will show a high return on ad spend, because those people were already warm. Some would have bought anyway. If you only look at the DSP report, you’ll overpay for sales you’d have made for free.

Running broad audiences with no frequency cap. This is the fastest way to spend a month’s budget in a week. Broad prospecting plus a small brand plus no cap means a lot of impressions shown to people who’ll never remember you.

Paying for a managed minimum you can’t feed. If someone offers to run DSP for you on a $50,000 managed plan while your Sponsored Products ACOS is still out of control, that’s backwards. Fix the search ads first. The managed minimum assumes a brand that already has the volume to use it.

Forgetting the attribution window when comparing ad types. Sponsored Products reports seller sales on a 7-day window. Sponsored Brands and Sponsored Display use 14 days, and DSP counts views as well as clicks. Lining them up side by side without adjusting makes the display and DSP numbers look better than they are.

Treating the new AI campaign as set-and-forget. DVA+ hands targeting and bidding to Amazon’s software, and Amazon’s announced results (up to a 68% better cost per acquisition and 34% higher ROAS) come from Amazon’s own internal data on Brand+ and Performance+ campaigns. Those may be real, but they’re not your account. Read your own brand-level numbers before you trust anyone’s averages.

Amazon DSP vs Sponsored Display: which one fits you

For most of the sellers reading this, the real choice isn’t “DSP or nothing.” It’s DSP or Sponsored Display, which already gives you a smaller version of the same retargeting idea inside the normal ads console.

Sponsored Display is self-serve, has no minimum, and runs on a daily budget you set. It can retarget people who viewed your product, show your ad on chosen product and category pages, and pay per click or per thousand viewable impressions. For sellers, it needs Brand Registry. DSP goes wider: more audience types, off-Amazon placements, streaming TV and audio, and finer controls like frequency caps and publisher lists. That reach costs money and attention. If you want the full rundown of the self-serve side, our Sponsored Brands vs Sponsored Display guide covers setup and the order we switch them on.

Here’s how we’d sequence it. Sponsored Products first, until the product is profitable and stable. Then Sponsored Brands and Sponsored Display to defend your brand and bring back viewers. Only after those are working, and only if the readiness check passes, a small DSP or DVA test on the warmest audiences. Skipping steps doesn’t speed anything up. It just moves the losses higher up the funnel, where they’re harder to see.

There’s also a cheaper outside option some brands use before DSP: running their own ads on Meta and sending that traffic to the Amazon listing. A method our team uses is checking the Meta Ads Library first to see how many competitors are already advertising on your main keywords there. A keyword that’s crowded and expensive on Amazon can be wide open on Meta. It’s a different tool with its own risks, but for a small brand it’s a much smaller bet than a DSP budget.

Who Amazon DSP is not for (and what to do instead)

The queue note for this post said to be plain about who DSP isn’t for, so here it is in one place. DSP, or DVA as it’s now called, is not for you yet if any of these describe your account:

• You’re launching your first product. Every launch dollar belongs in Sponsored Products. Our launch floor is $50 a day on search ads for about 60 days, and that’s where the ranking happens.
• Your search ads aren’t profitable. If ACOS is far above what your margin can carry and TACOS keeps climbing, fix keyword targeting, bids, and the listing first.
• Your product is generic. If 5 or more sellers offer the same item and buyers choose on price, more reach just sends more people to compare prices.
• Your ratings are under 4.5 stars or returns are high. Retargeting sends shoppers back to the same reviews that made them leave.
• Your total monthly ad spend is a few thousand dollars. A real DSP test would swallow most of it.
• You can’t read brand-level reports yet. If TACOS, new-to-brand, and organic rank aren’t numbers you already check, you won’t be able to tell whether DSP worked.

What should you do instead? For most sellers in that list, the honest answer is less exciting than a new ad platform. Tighten Sponsored Products until you know your numbers. Get TACOS under control. Work through the 3 ad types in order, and add Sponsored Display retargeting before you think about programmatic. If you want the full order by budget, our Amazon PPC guide by ad spend lays out what to run at each level from $0 to $10K a month, and our ACOS vs TACOS guide shows how to set the targets you’ll need before any DSP test means anything.

Don’t worry if DSP feels out of reach right now. It’s supposed to be. Plenty of brands doing healthy sales never run it, and that’s fine. It’s a tool for when search is already working, not a shortcut to get search working.

Learning to run Amazon ads yourself

DSP sits at the far end of Amazon advertising. Everything before it, the part that actually decides whether a product makes money, is learnable: keyword research, campaign structure, bids, placements, ACOS and TACOS targets, and when to add each ad type.

That’s what the PPC lessons in our $27 course walk through, using the same methods we run on the stores we manage. It doesn’t cover DSP, because DSP isn’t where a beginner or early-stage seller should be spending. It covers the ads that get a product to the point where DSP might one day be worth a test. If you want that foundation, the $27 Amazon course has 9 modules and 102 lessons, with the full PPC section included.

Frequently Asked Questions

Is Amazon DSP worth it for small sellers?

Usually not yet. DSP buys impressions on and off Amazon, mostly by the thousand, and the managed service typically needs $50,000. A small seller gets more from fixing Sponsored Products and adding Sponsored Display retargeting first, then testing DSP once search ads are profitable and stable.

What is the minimum spend for Amazon DSP?

Amazon says its managed service typically requires a minimum spend of $50,000. Self-service has no published minimum from Amazon, but agency guides in 2026 put practical starting budgets around $5,000 to $15,000 a month. Amazon hasn’t said whether the new DVA+ campaigns carry a minimum.

What happened to Amazon DSP in 2026?

At unBoxed on September 29, 2026, Amazon merged the ads console and DSP into the Amazon Ads Agent and renamed the programmatic side DVA, for display, video, and audio. DVA+ is a new AI-run campaign type, with the older DSP controls kept under advanced settings. It was in a US closed beta, with open beta expected in late October 2026.

What’s the difference between Amazon DSP and Sponsored Display?

Sponsored Display is self-serve inside the normal ads console, with no minimum and a daily budget, and it needs Brand Registry for sellers. DSP reaches further, including third-party sites, apps, Prime Video, and streaming TV, with more audience options and finer controls, and it costs more to test properly.

Can I use Amazon DSP for retargeting?

Yes. Retargeting people who viewed your product or added it to cart without buying is the most common first use. Keep the audience warm, cap how often each person sees the ad, and measure brand-level TACOS and new-to-brand orders so you don’t pay for sales you’d have made anyway.

Do I need to sell on Amazon to use Amazon DSP?

No. Amazon says DSP is open to advertisers who sell on Amazon and to those who don’t. Brands can use it to send traffic to their own website as well as to an Amazon listing.

The Bottom Line

Amazon DSP, now called DVA inside the new Amazon Ads Agent, is a strong tool for a specific kind of brand: one whose search ads already make money, whose listing converts, whose ratings hold above 4.5 stars, and whose ad budget can fund a proper test without starving what’s working. For that brand, a narrow retargeting test with a frequency cap, a fixed window, and brand-level measurement is a sensible next step.

For everyone else, it’s a distraction with a high minimum. The managed service typically starts at $50,000, and even self-service needs thousands a month to give you a readable result. Most sellers will make more by tightening Sponsored Products, controlling TACOS, and adding Sponsored Display retargeting first. Get search working, and DSP becomes a real option later. Skip that part, and DSP just shows a weak offer to more people. If you want to start with the ads that matter now, the free guide below is a good next read.