Sponsored Products can only reach people already typing your keywords, and that pool has a size. Amazon DSP buys audiences instead of keywords. What it actually does, where the ads run, what it costs, and the point at which it becomes worth it.
Amazon DSP
There is a point in most successful Amazon accounts where more budget stops producing more sales. You own the top of search on every term that matters, the automatic campaigns have been mined dry, and the next pound buys the same shopper slightly more expensively. Amazon DSP is what comes after that ceiling, because it buys audiences rather than keywords.
The ceiling is built into Sponsored Products and it is not a failure of management. Search advertising can only show your product to somebody already typing your keyword. That pool has a size. Once you have it, you have it.
What is on this page
The ceiling built into search advertising
Sponsored Products is the most efficient advertising most sellers will ever run, because it intercepts people who have already decided they want something like your product. That is also its limit.
Once you hold the top of search on your commercial terms, additional spend goes into three places: worse keywords, higher bids on keywords you already win, and placements that convert less well. All three raise your cost per sale without raising your sales much, which is exactly what the flattening curve above looks like from inside the account.
Before concluding you have hit that ceiling, make sure it is real. Most accounts that feel capped still have keyword coverage gaps, wasted spend sitting on terms that never convert, or listings that cannot close the traffic they already get. Work through the search term report first. If cutting waste and harvesting winners still leaves you flat, the ceiling is genuine.
What Amazon DSP actually is
Amazon’s demand side platform buys display, video and audio inventory using Amazon’s shopping data as the targeting layer. The important difference is the unit of targeting: search buys keywords, DSP buys audiences.
Those audiences are built from behaviour Amazon can observe. People who viewed your product and did not buy. People who bought a competitor in the last thirty days. People Amazon can see are in market for your category this week. People who bought from you a year ago and are due to run out. None of those are things you can target with a keyword, because none of those people are necessarily searching at all.
Search finds the hand already reaching for the shelf. DSP decides who walks down the aisle in the first place. Both are worth having, and the order matters: an account with unexploited search opportunity should finish that work before paying more to create demand it cannot yet capture cheaply.
Where the ads actually run
Three broad groups, and the mix matters more than sellers expect.
On Amazon itself. Detail pages, search results pages and elsewhere across the store. This is the most directly commercial inventory, and it includes the defensive placements on your own listings.
Across Amazon’s own media estate. Prime Video, Fire TV, Fire tablets, Twitch, IMDb and live sport. This is where DSP reaches people who are not shopping at that moment at all, which is the entire point of it.
Across third party web and apps. A very large amount of ordinary inventory bought through Amazon Publisher Direct and exchanges, still targeted with Amazon purchase behaviour rather than generic web data. This is usually where the volume sits, and where careful placement management earns its keep.
Each group behaves differently on cost and on conversion, which is why a single blended return figure for a DSP campaign tells you so little.
Not sure whether you have actually hit the search ceiling?
Send us the reports. We will tell you whether DSP is the answer or whether search still has room.
Three things it does that search cannot
It brings genuinely new customers. DSP is judged on new to brand purchases, meaning buyers who had not bought from you in the previous twelve months. On the accounts we run it, that share sits between seventy and eighty percent of everything the channel produces. Search rarely produces many, because people searching your category often already know the brands in it.
It defends what you have built. Competitors are running conquesting campaigns on your detail pages right now. Retargeting your own product viewers is the cheapest defence available, and it is not a placement Sponsored Products can buy.
It makes your search spend work harder. A shopper who has already seen you converts on search at a measurably better rate. Some of DSP’s return therefore shows up inside your Sponsored Products numbers rather than in the DSP report, which is the single most misunderstood thing about the channel.
The attribution trap. If DSP introduces a shopper and search closes them, the DSP report shows spend with modest return and the search report shows a cheap sale. Judge the two channels separately on their own reports and you will conclude DSP is failing and search is brilliant. Both conclusions are wrong.
What it costs and how the money works
DSP is bought on impressions rather than clicks, usually quoted as a cost per thousand. That changes the shape of everything: you are paying for reach and frequency, and the sale arrives later, sometimes considerably later.
Two costs sit on top of media. There is usually a management fee, whether from Amazon’s managed service or an agency running a seat. And there is creative, since display and video need actual assets, which is a real line item that sellers coming from search often forget entirely.
Minimums are the other consideration. Amazon’s managed service has historically required a substantial commitment, while agencies with a seat can run smaller budgets. Those thresholds move, so check the current position rather than a figure from a blog post. What does not move is the underlying constraint: a budget too small to give the audiences volume to learn from will underperform no matter who is running it.
How to tell whether you are ready
Four questions, and you want a yes to all four.
Have you finished the search work? Full keyword coverage, waste cut, harvesting running, top of search held on your commercial terms. If not, that work is cheaper and faster than DSP.
Do your listings convert? DSP sends colder traffic than search, so a page that struggles to close warm shoppers will struggle badly with cold ones. Check it on a phone first, using the mobile checks.
Can you fund eight to twelve weeks? DSP judged at week three looks like a failure in almost every account, because the cost is immediate and much of the return is not.
Do you know your customer lifetime value? This is the one people skip, and it decides everything. DSP buys new customers. If you do not know what a new customer is worth over a year, you cannot say whether the price was good.
If the honest answer to any of those is no, fix that first. It will cost less and the DSP campaign will work better when you get to it.
Search buys the sale you were going to get slightly sooner. DSP buys the customer you were not going to get at all. The second one is harder to measure and worth more.
Already running DSP and unsure whether it is working?
Send the reports. We will read them properly and tell you what the numbers actually say.
Measuring it without kidding yourself
Leading with ROAS makes a well run DSP look like a poorly run search campaign, because you are compressing a demand creation channel into a demand capture metric.
Ask for new to brand sales as a share of total, new to brand cost measured against lifetime value rather than an ACOS target, detail page views and branded search volume over the period, and search performance on the same terms while DSP runs. The full case for that reporting, with two real accounts read properly, is in new to brand, the only DSP metric worth reporting.
One practical habit: agree the measurement before the campaign starts, in writing. Every argument we have ever seen about DSP performance is really an argument about which numbers count, and it is far easier to settle that in week zero than in week nine.
Five ways DSP money gets wasted
Running it before search is finished. Paying to create demand while cheaper demand sits uncaptured.
Judging it at three weeks. All of the cost, almost none of the payoff, and then the campaign gets switched off just as the audiences start working.
Only retargeting. Retargeting is the cheapest and easiest audience, so budgets drift towards it. If new to brand share is low, you are mostly buying people who were already coming.
Ignoring frequency. Showing the same creative to the same person twenty times a week wastes money and irritates the audience you paid to reach.
Treating creative as an afterthought. On search, your listing is the creative. On DSP, the ad has to earn the click on its own, and weak assets waste good targeting.
Handled properly, DSP is the channel that grows an account once search has stopped being able to. Handled as an extension of search, with search metrics and search patience, it usually gets switched off before it was ever given the chance to work.
Wondering if DSP is right for your account?
Twenty minutes on a call and we will tell you honestly, including if the answer is not yet.
Frequently asked questions
It is Amazon’s demand side platform, which buys display, video and audio advertising using Amazon’s shopping data to decide who sees it. Sponsored Products targets what somebody typed. DSP targets who somebody is: people who viewed your product, bought a competitor, or are in market for your category this week.
No. DSP is open to advertisers who do not sell on Amazon at all, since the value is the shopping data rather than the storefront. In practice most sellers using it do sell on Amazon, because the retargeting and defensive audiences are the easiest place to start.
Managed service through Amazon has historically carried a substantial minimum, often quoted around the tens of thousands, while agencies with a seat can run much smaller budgets. Minimums move, so check current terms, and be aware that a budget too small to give the audiences anything to learn from will underperform regardless of who runs it.
They do different jobs, so the comparison does not really work. Sponsored Products captures existing demand cheaply and should almost always be maxed out first. DSP creates demand and defends what you have built. Running DSP while your search campaigns still have room is spending on the harder problem before the easy one is finished.
On Amazon’s own detail and search pages, across its streaming and device estate including Prime Video, Fire TV, Fire tablets, Twitch and IMDb, and across a large amount of ordinary web and app inventory bought through Amazon Publisher Direct and third party exchanges, all still targeted with Amazon shopping behaviour.
On new to brand sales as a share of the total, on what a new to brand customer costs against lifetime value, on detail page views and branded search movement, and on whether your search campaigns convert better while it runs. Leading with ROAS alone makes a well run DSP look like a poor search campaign.
Give it at least eight to twelve weeks. The audiences need volume to learn, the creative usually needs at least one iteration, and a meaningful share of the return arrives later through search and repeat purchase rather than immediately in the DSP report.
Through Amazon’s own managed service or through an agency seat, mostly. Self service access exists but is limited, and the platform rewards experience: audience construction, frequency management and creative rotation all have real craft in them and mistakes there are expensive.
Send the right reports. Get a straight answer back.
We read the account properly and write back with where the money is going, what it is buying, and the three things we would change first. You keep it either way.
No obligation · no card · a person replies, usually the same working day