Amazon DSP used to be a walled garden for 8-figure brands and agencies. In 2026 that changed: Amazon widened self-service access and a layer of lower-floor tools opened the door to mid-size sellers. The hype that followed is dangerous, because most "best Amazon DSP" roundups rank platforms by ROAS, the one metric that hides whether DSP actually made you money. This roundup scores five platforms on net profit and minimum-spend reality, and states plainly the spend floor below which DSP burns cash.
In this article
What is Amazon DSP and how did it change in 2026?
Amazon DSP is Amazon's programmatic platform for buying display, video, and audio ads that reach shoppers on and off Amazon, across the Amazon home page, Fire TV, Twitch, Kindle, and third-party exchanges. Where Sponsored Products only appear inside Amazon search results, DSP reaches audiences before and after they search, and its core strength is retargeting shoppers who viewed but did not buy. Where PPC captures existing demand, DSP builds and re-engages it.
The 2026 shift is access. Amazon expanded self-service DSP, and a set of lower-floor self-service partners and tools made it reachable for mid-size sellers rather than only 8-figure brands with managed-service contracts. Amazon's own DSP product page now markets self-service alongside managed-service, which is the structural change driving the wave of seller interest.
What is the real minimum spend?
The real minimum spend is the number vendors bury, and it is the single most important figure in this decision. There is no hard, enforced public minimum on Amazon's self-service DSP tools as of 2026verified May 2026, but the practical floor for measurable results is roughly $10,000 to $15,000 per month sustained for at least 90 days. Managed self-service through an agency partner can start near $5,000 per month plus the agency fee; traditional direct managed-service historically wanted around $35,000 per month.
The reason the floor exists is mechanical: below it, your reach frequency is too thin and audience overlap too high to drive measurable incremental lift. A solo seller running DSP part-time at $4K/month will underperform a $9K/month managed campaign run full-time, not because the software is worse, but because the spend cannot buy enough frequency to move behavior. This is why the entry minimum is the wrong number to optimize, the right number is the spend at which your audiences are dense enough to convert.
The DSP platform matrix
The matrix scores each platform on the three axes that decide net-profit outcomes: realistic self-serve spend floor, whether it surfaces net-profit reporting (not just ROAS), and its best-fit spend tier. Figures are 2026 snapshots and negotiate at volume.
| Platform | Self-serve floor | Net-profit reporting | Best-fit spend tier | Standout strength |
|---|---|---|---|---|
| Hector | ~$5-10K/mo | Partial | $5-25K/mo | Low-floor self-serve entry |
| Sellics (Perpetua) | ~$10K/mo | Yes | $10-50K/mo | Profit-centric heritage |
| Perpetua | ~$15K/mo | Yes | $25-100K/mo | Agentic goal-based optimization |
| Quartile | ~$25K/mo | Partial | $40K+/mo | ML audience modeling |
| Pacvue | Custom (~$25K+) | Yes | 8-figure / agency | Multi-retail-media breadth |
The winner row is Hector specifically because it lowers the self-serve floor closest to where mid-size sellers actually operate, not because it is the most powerful platform. At the top end, Pacvue is unmatched on breadth but irrelevant to anyone under 8 figures. The matrix's real lesson: match the platform to your sustainable spend, because a powerful platform you cannot feed enough spend is worse than a simpler one you can.
Why score on net profit, not ROAS?
Net profit is the only metric that tells you whether DSP made you money; ROAS can look spectacular while you lose cash. A 5x ROAS sounds great until you subtract product cost, Amazon fees, and the fact that many of those "attributed" sales would have happened anyway, the incrementality problem. DSP is especially prone to ROAS illusion because retargeting claims credit for shoppers who were already going to buy.
The platforms with profit-centric heritage (Sellics, now part of Perpetua, and Perpetua itself) surface net-profit and incrementality views rather than headline ROAS, which is why they score higher on the axis that matters. When you evaluate a DSP platform, the question is not "what ROAS can it hit" but "can it show me the incremental net profit after product cost and fees, and does that number stay positive." If a platform only reports ROAS, it is optimizing the metric that hides your losses. For the deduction side of ad spend, our friends at CeoCult cover how DSP and PPC spend is treated under Schedule C for sellers and LLCs.
The five platforms, briefly
Hector is the realistic 2026 entry point, a self-service DSP tool with a floor low enough for mid-size sellers, partial net-profit reporting, and a manageable learning curve. Sellics (absorbed into Perpetua) carries a profit-first heritage and remains a strong choice for sellers who want net-profit dashboards from day one. Perpetua brings its agentic, goal-based optimization to DSP, you declare a target and it manages structure, which justifies its higher floor for brands at $25K+/month. See our Perpetua review for the full platform picture.
Quartile applies its ML audience modeling to DSP and fits brands at $40K+/month that want the model to drive audience selection. Pacvue is the enterprise retail-media platform, spanning Amazon DSP plus Walmart, Instacart, and more, and is the right answer only for 8-figure operations and agencies. For where these same vendors sit on the search-ads side, our best Amazon PPC software roundup ranks them on cost-as-percent of ad spend.
When DSP burns cash (the honest anti-recommendation)
Three situations where DSP loses money in 2026, regardless of which platform you choose:
- Spend under ~$10K/month. Reach frequency is too thin to move behavior; you pay for impressions that never compound into conversions. Master Sponsored Ads first.
- Sponsored Ads not yet optimized. If your Sponsored Products and Brands still have obvious waste, DSP is reaching audiences you have not converted on the cheaper, higher-intent surface. Fix search first.
- Measuring success by ROAS. A campaign optimizing for ROAS will happily retarget shoppers who were already buying, showing a great number while adding zero incremental profit. Demand net-profit and incrementality reporting before you scale spend.
Get the seller profit-calculator template
The spreadsheet template we use to compute true net profit after product cost, Amazon fees, and ad spend, so you can tell whether DSP actually paid. Free, no fluff.
Bottom line: DSP is additive, not a starting move
Amazon DSP got more accessible in 2026, but accessible is not the same as advisable. The decision rule is simple: only enter DSP once you can sustain roughly $10K+/month for 90 days and you have already optimized Sponsored Products and Brands so DSP adds incremental reach rather than duplicating cheaper search demand. At that point, pick the platform that matches your sustainable spend, Hector to enter, Sellics or Perpetua for profit-centric optimization, Pacvue only at 8-figure scale.
Above all, score on net profit and incrementality, not ROAS. The platforms that surface those numbers are the ones that will keep DSP profitable as you scale; the ones that only report ROAS are optimizing the metric that hides your losses. DSP is a powerful additive layer for brands with scale, and a cash incinerator for everyone who reaches for it too early.