Amazon DSP Just Became the Ad Channel Nobody Can Ignore

Amazon DSP used to be easy for smaller ecommerce teams to ignore.

Sponsored Products were simpler. Sponsored Brands were easier to explain. Search placements felt closer to the sale. DSP sounded like something for enterprise brands, agencies, and teams with enough budget to tolerate fuzzy measurement.

That assumption is getting weaker.

In Q1 2026, Skai reported that Amazon DSP CPCs fell 53% year over year to $0.86, making DSP cheaper per click than Sponsored Products for the first time in its dataset. Retail media spend still grew 27% year over year. Clicks grew even faster. That combination deserves attention.

It does not mean every brand should dump budget into DSP tomorrow. It does mean the old habit of treating DSP as optional background noise may be out of date.

Amazon is no longer only a search marketplace. It is a media ecosystem. Shoppers research, browse, compare, stream, read reviews, watch content, abandon carts, return later, and buy across surfaces. DSP is one of the ways brands can influence that larger journey.

The question is not whether Amazon DSP is powerful. The question is whether your brand is ready to use it without confusing cheap clicks for profitable growth.

Why DSP matters now

Sponsored Products capture demand that already exists. A shopper types a query, sees a product ad, clicks, compares, and buys or does not buy. That is useful, but it mostly happens late in the journey.

DSP can reach shoppers before or after that search moment. It can retarget people who viewed your product but did not buy. It can reach in-market audiences. It can support launches. It can run display and video. It can extend beyond Amazon using Amazon shopping signals.

That broader reach used to come with a simple objection: it was more expensive and harder to measure. If clicks are becoming cheaper while inventory and targeting improve, the objection changes. DSP may still be harder to measure, but it may no longer be easy to dismiss on cost alone.

This shift matters because Amazon search is crowded. In many categories, Sponsored Products auctions are mature, margin is thin, and everyone is bidding on the same obvious terms. A brand that only buys bottom-funnel clicks may end up fighting the most expensive part of the market.

DSP gives brands another option: create familiarity before the search, recover demand after the view, and test audiences that do not fit neatly into keyword campaigns.

Do not use DSP as a dumping ground for leftover budget

The worst DSP campaigns start with a leftover-budget mindset.

The team has money remaining. Sponsored Products are capped or expensive. Someone suggests DSP because it sounds sophisticated. A broad audience gets built. Creative is repurposed from another channel. Reporting comes back with impressions, clicks, and attributed sales. Nobody knows whether the campaign actually changed behavior.

That is not a DSP strategy. That is a spend disposal plan.

DSP needs a specific job before money moves. A retargeting campaign should not be judged the same way as a connected TV awareness test. A launch campaign should not be judged like branded defense. A competitor conquesting campaign should not be expected to produce the same ROAS as cart abandonment retargeting.

Before launching, define the job in one sentence. For example: recover shoppers who viewed the hero ASIN in the last 14 days but did not buy. Or introduce a new product line to category buyers before Prime Day. Or defend high-margin replenishment SKUs from competitors. If the team cannot write the job clearly, the campaign is not ready.

DSP is strongest when the product page is already strong

Cheaper DSP clicks will not rescue weak merchandising.

If your listing has poor images, thin reviews, vague bullets, unclear variants, weak pricing, slow delivery, or unsupported claims, DSP will only expose that weakness to more people. The click may be cheaper, but the wasted visit is still wasted.

This is why DSP planning should start with the retail shelf. The ASIN needs to be conversion-ready before you widen the audience.

Check the product detail page. Can a shopper understand the product in ten seconds? Does the hero image work at thumbnail size? Do reviews support the promise in the ad? Does the listing explain who should buy and who should not? Are shipping and return expectations clear? If the ad says the product is built for travel, does the page show travel use cases?

The same issue appears in AI commerce. Product data and content clarity determine whether a shopper or assistant can understand the product. That is why the product-data argument in Your Product Feed Is the New SEO, and Yours Is Probably Failing applies to media teams too.

Retargeting is the first DSP test for many brands

For many ecommerce brands, the simplest DSP starting point is retargeting.

These shoppers already showed intent. They viewed a product, visited a brand store, added to cart, or interacted with a category. The campaign does not need to explain the entire problem from scratch. It needs to bring the shopper back with a sharper reason to finish.

Retargeting also exposes whether the product page is leaking demand. If many shoppers view and leave, the issue may be price, reviews, delivery promise, missing comparison information, or unclear fit. A DSP retargeting campaign can recover some of that demand, but the better move is to fix the reason people hesitate.

Segment retargeting windows carefully. A shopper who viewed yesterday is not the same as a shopper who viewed 30 days ago. A cart abandoner is not the same as a casual product viewer. A repeat buyer of a consumable is not the same as a first-time browser.

The shorter and more intent-heavy the audience, the more performance-oriented the campaign can be. The broader and colder the audience, the more realistic the measurement needs to be.

DSP can support launches, but it cannot create product-market fit

New product launches are a tempting DSP use case.

The logic makes sense. Search volume may be limited because the product is new. Sponsored Products may struggle if the listing has no review base. DSP can introduce the product to relevant category shoppers and build early awareness.

That can work, but only if the launch has a clear offer and a clear reason to exist.

DSP cannot fix a product nobody understands. It cannot create urgency from a generic launch. It cannot compensate for a listing that fails to explain why the new product is better than the existing options.

Before using DSP for a launch, write the buying argument. What changed in the category? What problem does the new product solve? Which competitor or alternative is it replacing? What proof exists? What audience has the strongest reason to care?

If those answers are strong, DSP can help the product get seen by the right people. If those answers are weak, DSP will only make the confusion more expensive.

Streaming and offsite inventory change the creative requirement

Amazon DSP is not only product-grid retargeting. It can reach shoppers across display, video, and streaming inventory. Amazon describes DSP as a way to programmatically buy ads on and off Amazon using Amazon audiences.

That changes the creative bar.

A Sponsored Products ad can rely heavily on the product image, price, reviews, and search intent. A DSP video or offsite display placement has to create context quickly. The shopper may not be actively searching. The creative needs to remind them of the problem, show the product, and make the next step obvious.

This is where many marketplace-first brands struggle. They have optimized listing images but not broader advertising assets. They can win a search click but cannot explain the product in a video, display unit, or streaming environment.

If DSP is becoming cheaper, brands should use the window to build creative muscles, not just buy more impressions. Stronger creative will matter more as retail media moves into streaming, creator content, and offsite audience extension.

Measurement needs to separate defense from growth

DSP measurement can get messy because Amazon has strong attribution incentives. The platform can show attributed sales, but that does not automatically prove incrementality.

A retargeting campaign may claim sales from people who would have returned anyway. A branded audience may look efficient because the brand already had demand. A broad awareness campaign may look weak on last-click ROAS while still increasing branded search later.

The solution is not to reject platform reporting. The solution is to label campaign jobs clearly.

Defense protects existing demand. Growth creates or expands demand. Retargeting recovers demand. Launch campaigns introduce new products. Each should have different expectations.

For defense, efficiency matters. For growth, new-to-brand share, branded search lift, detail page view lift, and downstream conversion may matter more. For retargeting, frequency and recency discipline matter. For launch support, early learning may be as valuable as immediate sales.

If every DSP campaign is judged by the same ROAS target, teams will overfund the safest audiences and underfund learning.

Margins still decide whether DSP is worth it

Cheap clicks do not override unit economics.

A campaign can have attractive CPCs and still lose money if the product has low gross margin, high return rates, expensive shipping, heavy discounts, or weak repeat purchase.

Before scaling DSP, calculate contribution margin by SKU. Include marketplace fees, fulfillment, returns, ad spend, discounts, and expected repeat purchase. If a SKU cannot support paid acquisition, do not use cheaper DSP as an excuse to push it harder.

Some products should be advertised because they acquire valuable repeat buyers. Some should be advertised because they pull customers into a larger bundle. Some should be protected because they are profitable and high intent. Others should not receive paid media at all.

DSP makes audience buying more tempting. Margin math keeps that temptation under control.

Prime Day makes DSP more important and more dangerous

Prime Day arriving in June creates a compressed retail media window. Brands will raise budgets, competitors will bid more aggressively, and shoppers will compare deals across categories.

DSP can help warm audiences before the event, retarget viewers during the event, and recover shoppers after the event. But it can also burn budget if the offer is weak or inventory is not ready. The Prime Day operating plan in Prime Day Is Coming Early should come before the media plan.

Do not use DSP to drive demand into products that may stock out too early. Do not advertise discounts that destroy contribution margin. Do not run broad campaigns without a clear post-event retention plan.

Retail media can amplify a strong Prime Day strategy. It cannot make a bad one safe.

A practical DSP testing plan

Start with one or two hero SKUs. Choose products with strong reviews, healthy inventory, enough margin, and a clear buying argument.

Run a retargeting test first. Separate detail page viewers, cart abandoners, and brand-store visitors. Keep windows tight enough that the audience still has intent.

Next, test one category audience. Use creative that explains the product's strongest use case, not generic brand awareness.

Then test one launch or seasonal angle if relevant. Tie the campaign to a specific event, problem, or deadline.

Measure by contribution margin, new-to-brand share, detail page view quality, and downstream sales. Keep notes on audience, creative, frequency, and landing experience.

After 30 to 45 days, decide what earned more budget. Do not scale every campaign. Scale the campaign with a clear job and evidence that it is doing that job.

What smaller brands should do before calling an agency

The agency question comes up quickly with DSP because the platform can feel less familiar than search advertising. Some brands should absolutely use a specialist. But a specialist cannot replace internal clarity. If the merchant does not know which products have margin, which audiences matter, which offers are protected, and which outcomes define success, the agency will inherit confusion and turn it into media spend.

Before hiring help, build a short internal brief. List the SKUs that can support paid acquisition. List the products that should never be pushed because they create returns, support tickets, or stockout risk. List the audiences already proven in Sponsored Products, Amazon Posts, brand-store analytics, email, and paid social. List the seasonal moments when the product has a real reason to be considered. This document does not need to be elaborate. It needs to prevent the first DSP test from becoming a guessing exercise.

Then decide what work should stay inside the company. Product positioning, margin rules, inventory constraints, and customer objections should not be fully outsourced. Campaign setup, audience planning, creative production, reporting, and optimization can be supported by outside help. The cleanest DSP relationships happen when the brand owns the business logic and the partner owns the media execution.

This matters more when costs fall. Cheaper clicks attract more casual testing, and casual testing creates vague briefs. Vague briefs usually become broad audiences, generic creative, and reports that nobody trusts. A little planning before the first campaign saves the team from months of argument later.

Use DSP to learn, not just to spend

The best DSP test produces more than a sales number. It tells the brand which audience responds, which promise gets attention, which creative format explains the product, and which hesitation blocks purchase. That learning should feed back into the Amazon listing, the brand store, the website, the email program, and the product roadmap.

For example, if retargeted shoppers respond to comparison creative, the product page may need a clearer comparison table. If category buyers click but do not buy, the offer may be too generic or the proof may be too thin. If video performs better than static display, the product may require demonstration before a shopper understands its value. If cart abandoners return only when the discount is deep, the base price or shipping promise may be doing more damage than the media team wants to admit.

This is the difference between performance marketing and operating feedback. A narrow team looks at DSP and asks whether the campaign hit target ROAS. A better team asks what the campaign revealed about demand quality, product clarity, price resistance, and channel fit. DSP can be a paid research tool when the test is designed that way.

That does not mean every lesson is statistically perfect. Retail media data is messy. Attribution is imperfect. Audiences overlap. But imperfect data can still improve decisions when the campaign has a defined hypothesis. If the hypothesis is clear, the team can decide whether the signal is strong enough to change the next move.

The final discipline is to write down what the brand will not do. Do not use DSP to hide weak listings. Do not use broad awareness when the product story is unclear. Do not scale because CPC looks cheap while contribution margin is unknown. These negative rules keep the testing program from drifting into spend for spend's sake.

The bottom line

Amazon DSP is becoming harder to ignore because the economics are changing. If CPCs fall while retail media inventory expands, brands have a chance to test audiences and formats that were previously too expensive or too vague.

But DSP is not a shortcut. It rewards brands with clear products, strong listings, disciplined measurement, and honest margin math.

The brands that win will not be the ones that spend because DSP got cheaper. They will be the ones that know exactly what each DSP dollar is supposed to do.