Amazon DSP Agency Reviews and What They Are Worth

Why this particular review corpus is so thin

Software accumulates reviews easily. A tool has thousands of users, a self-service signup, an in-app prompt and a low-stakes opinion to give. Agencies accumulate reviews slowly, and DSP agencies more slowly still, for three structural reasons.

• The buyer population is small. Programmatic display is a later purchase than sponsored ads. Far fewer brands have ever hired a DSP partner than have hired an Amazon PPC agency, so there is simply less to read.
• Display is rarely the whole contract. Most reviews are written about an engagement that included listings, search, creative and reporting. The reviewer describes responsiveness and account management because that is what they experienced day to day, not audience architecture.
• Nobody reviews the thing that matters. The question worth answering is whether display added sales that would not otherwise have happened. That takes a designed test, and a brand that never ran one has nothing to report either way.

None of this means reviews are useless. It means they answer a different question than the one you are asking, and reading them as if they answered yours is the mistake.

First, the disambiguation this search needs: Amazon DSP is the demand-side platform, the programmatic advertising product. Amazon Delivery Service Partner is the courier franchise programme. If you are researching delivery routes and vans, no advertising agency review will help you.

Where the reviews actually live, and what each source is worth

Six places carry something, and they are not equally informative.

• B2B service directories. The richest text you will find on agencies, and the most structured — project size, industry, duration. Note that listings are usually built with the agency's cooperation and reviewers are often supplied by the agency, which is disclosed but easy to skim past. These sites also block automated retrieval, so read them in a browser rather than trusting a summary someone else scraped.
• General consumer review platforms. Occasionally carry agency profiles. Volume is low and the reviewers are a mix of clients and, sometimes, former staff. Treat as anecdote.
• Software review sites. Genuinely useful for platforms, but they list agencies inconsistently and the review template is written for products. A five-star rating of an agency on a product template tells you almost nothing about media buying.
• Amazon's own partner directory. Not reviews — a listing of service offerings, expertise levels and marketplaces, with status badges. Verifiable, which puts it above most sources, but it is a credential rather than an opinion.
• Case studies on the agency's own site. Selected by definition, but useful if they carry denominators. A case study that names the scope, the period and the advertiser count is worth ten that name only a multiple.
• Your own network. The highest-signal source by a distance, and the only one where you can ask a follow-up question.

Why a five-star agency review tells you less than you expect

Three biases sit inside almost every positive agency review, and knowing them changes how you read.

The timing bias. Reviews are usually solicited when the relationship is going well, which is early. Onboarding enthusiasm and month-fourteen reality are different things, and the review captures the first one. Look at the date on every review and mentally discount anything written in the first quarter of an engagement.

The author bias. The person writing the review is frequently the person who chose the agency. A critical review is partly a critique of their own decision, so mild dissatisfaction tends to surface as four stars and a warm paragraph rather than as a complaint.

The subject bias. Reviews describe service: responsiveness, reporting, meeting quality, whether the team is pleasant. Those things matter. They are not the same as whether the media buying worked, and an agency can score perfectly on all of them while never having established that display was incremental.

The useful move is to read reviews for texture — how they handle problems, how stable the team is, what happens when results dip — and to get your answer on performance somewhere else entirely.

When a firm has almost no public reviews at all

This happens more than you would expect, and it is not a red flag by itself. A practice area inside a large holding company does not accumulate reviews the way an independent agency does, because the client relationship sits with the group, the contracts are enterprise-scale, and nobody involved is filling in a directory profile. Newer firms have the same problem for the opposite reason.

The workaround is to stop looking for reviews and start looking for artefacts you can verify:

• Named credentials. Amazon runs partner status badges and an annual Partner Awards programme; both are checkable against Amazon's own material rather than the agency's.
• Public speaking and published method. An agency that publishes how it designs a test is exposing itself to correction, which is a form of evidence.
• Team tenure. Public professional profiles will tell you how long the senior people have been there. Continuity is one of the strongest predictors of a good engagement and one of the least discussed.
• Reference calls with a client who left. Ask for one. The response to the request is informative even if the call never happens.

The reference-call script that beats any review

Twenty minutes with a comparable brand outperforms an afternoon of reading. Six questions, in this order, because the order matters — the easy ones first buy you candour on the hard ones.

• "What did the first ninety days actually look like?" Onboarding is where most engagements are quietly decided.
• "Who is on your account each week, and has that changed?" Turnover in the delivery team is the most common cause of a good engagement going flat.
• "What did they get wrong, and how did you find out?" Every partner gets something wrong. The informative part is whether the client heard it from the agency or discovered it themselves.
• "Has display ever been withheld to see what happened?" If the answer is no, nobody in that relationship knows what display contributed.
• "What arrived on an invoice that you were not expecting?" This surfaces the inclusions gap faster than reading a contract does.
• "Would you hire them again for the same brief, and for a different one?" The gap between those two answers is where the honest assessment lives.

How to read the one critical review properly

When an agency has thirty positive reviews and one that is lukewarm, most buyers discount the outlier. That is backwards. The critical review is usually the most informative document in the set, because everything pushing reviews toward the positive — solicitation timing, author bias, social norms — had to be overcome for it to exist at all.

Read it for three things:

• The specific complaint, stripped of tone. "Communication could be better" is noise. "Fees changed at renewal without discussion" is a mechanism, and mechanisms repeat. One genuinely useful contract clause came out of exactly that kind of comment: get written notice of any fee change, with your agreement required rather than a revised schedule appearing on an invoice. Ask for it from any agency, ourselves included.
• Whether the agency replied, and how. A defensive reply tells you how disputes go. A reply that concedes a specific fact and describes a change is the single best signal available on any review platform.
• Whether the complaint is structural or personal. A named individual who has since left is a solved problem. A complaint about billing, scope creep or reporting is about how the firm is run, and firms change that slowly.

Then take the mechanism into your reference calls and ask about it directly. "I read that fees changed at renewal for one client — has that happened to you?" is a far better question than "were you happy with them", and it turns a single anonymous review into something you can verify.

What to ask us, and what our own numbers say

Apply all of the above to reMKTR. We are the Amazon DSP arm of Full Circle, a full-service Amazon management company with $500M+ in managed spend across 100+ brands, and we run 109 live Amazon DSP advertiser seats. We do not publish client names on the DSP side, which means we are asking you to weigh a portfolio figure rather than a logo wall.

Here is that figure with its scope attached: across 30 advertisers in July 2026, 6.04x return on ad spend, 78.4 million impressions at a $4.00 CPM, a blended $1.42 cost per click, and a $5.49 cost per acquisition across 57,137 attributed purchases, of which 20.1% came from shoppers new to the brand. Thirty advertisers is a slice of the book, not all of it, and we say so because a number without a denominator is not evidence.

The claim we would rather be judged on is the method. Last-click attribution cannot prove incrementality and never could; holdouts and matched controls can. We reconcile in Amazon Marketing Cloud so that display and sponsored ads stop double-counting each other. Ask us to describe a specific test, including one that came back weaker than we hoped — that request is a better filter than any star rating, and it works on every agency you are considering.

Two siblings, and when each applies. Dr. DSP — when you would rather evaluate a product with a published scope than evaluate an agency's reputation. Full Circle — when the reviews you should really be reading are about full Amazon account management, because display is not your first problem.