Acorn vs the field — a scorecard, then four short head-to-heads

Set the terms before the bake-off

Three definitions, because getting them wrong is how these evaluations go sideways.

Amazon DSP here means the Demand-Side Platform — Amazon's programmatic system for buying display, video and audio inventory on and off Amazon, with Amazon shopping signal behind the targeting. It is not the Delivery Service Partner programme, which is about parcel vans and has nothing to do with advertising.

Acorn here means Acorn-i, at acorn-i.com. The word collides four ways in search results — Acorn by New Engen, Acorns the investing app, and Acorn TV all outrank the one you want. Acorn-i describes itself on its own site as "A Technology Led Ecommerce Agency" and "Experts in Amazon and ecommerce", London-based, running an analytics application called Ignite and publishing a 94% client retention claim beyond twelve months. Their site does not name Amazon DSP.

The candidates are not the same species. Acorn-i is a service. Skai and Quartile are platforms with service wrapped around them. The Trade Desk is a platform, and one aimed at a different job entirely. A scorecard that treats them as interchangeable will hand you a winner you cannot actually buy.

So the useful structure is: score everybody on questions that apply regardless of species, then run the head-to-heads to see where each is genuinely strongest. That is the shape of the rest of this page.

The seven questions to score every candidate on

Give each candidate a mark out of five on each. Score them yourself rather than accepting the vendor's own framing, and score us on the same sheet.

• Whose seat is it? Does the Amazon DSP advertiser seat belong to you or to them? This decides what happens on the day you leave, and it is the single most commonly skipped question in the category.
• What is the cadence? How often does a human look at the account and change something — daily, weekly, or when a report is due? Ask for the last four weeks of change logs, not a description of the process.
• Who designs and computes the measurement? If the party being measured also designs the test, that is worth knowing. Ask whether the holdout or matched control is written into the statement of work, with a named methodology, and who runs the query.
• What shape is the fee? Fixed licence, percentage of working media, banded tier, or inside the media rate. And is your budget inclusive or exclusive of it?
• Where does escalation go? Name the person who decides when a campaign should be paused, and how fast they can be reached.
• What does exit look like? Notice period, term, renewal, and what data leaves with you.
• How honest is the sales conversation? Does anyone tell you where they are not the right answer? The candidate who concedes something unprompted is usually the one telling the truth about the rest.

Seven questions, five marks each, thirty-five available. In our experience the spread between candidates on questions one, three and six is far wider than the spread on price, and price is what most shortlists are actually decided on.

One procedural note that improves the results more than the questions themselves: score each candidate immediately after its call, before the next one. Scoring all of them at the end of a two-week process reliably rewards whoever presented last and whoever was most charming, which are not properties of the work. And write one sentence per score explaining the mark. If you cannot write the sentence, you did not get an answer — you got a reassurance, and the difference is the whole point of the exercise.

Acorn-i vs Quartile — a service against a platform with scale published

Quartile is the most directly comparable name on this list because it does publish its scale, even though it publishes no price. On quartile.com, read 20 August 2026, it describes itself as "The World's Largest Retail Media Optimization Platform" and states $2B+ in annual retail ad spend managed, $25B+ in annual managed sales, 5,300+ customers across 32+ countries. It explicitly names Amazon DSP alongside Amazon sponsored ads, Walmart, Instacart, Criteo, Google, Meta and Microsoft.

Those four published numbers permit arithmetic nobody seems to do with them. $2B of ad spend across 5,300 customers averages roughly $377,000 of annual media per customer — about $31,000 a month. And $25B of managed sales against $2B of managed spend implies a blended 12.5x across their whole published book, or an 8% advertising cost of sales.

Both figures are useful and neither is a criticism. The first tells you where you sit relative to their typical customer, which is the most decision-relevant thing on their homepage and is not stated anywhere as such. If you spend $8,000 a month you are well below their average and should ask directly what service level that buys; if you spend $300,000 you are an order of magnitude above it and have leverage. The second is a company-wide blend across many channels and many account types, so it is not a forecast for you — but it is a published, self-reported figure that they have chosen to stand behind.

Where Quartile wins: breadth. If you sell across Amazon, Walmart, Instacart and DTC and want one optimisation layer across all of it with an account team attached, that is exactly what they are built for. Where Acorn-i wins: service depth in one place. A London agency with content production, insight and media under one roof is a different offer from an optimisation platform, and for a brand whose bottleneck is execution rather than bid maths it is often the better one.

Acorn-i vs Skai — the only one that shows its whole price

Skai publishes four exact annual figures against four exact media-spend bands, in US dollars, read on skai.io on 20 August 2026: Standard $114,000 up to $4M of annual media, Advanced $276,000 up to $10M, Enterprise $504,000 up to $20M, Enterprise Premier $756,000 up to $35M, with a custom tier above that. Nobody else on this page shows that much.

The arithmetic worth doing on it is not the discount curve — it is what the same tier costs you depending on where you sit inside the band. Standard covers up to $4M. At $4M it works out at 2.85% of media. At $2M it is 5.70%. At $1M it is 11.40%. Same product, same invoice, and an effective rate that quadruples as you move down inside a single band.

That is not a flaw in Skai's pricing; it is what "up to" means, and they have published the numbers that let you work it out, which almost nobody else has. It is a flaw in how buyers read banded pricing. The rule it gives you is short: a fixed licence rewards the top of a band and punishes the bottom of it, and a percentage of media does neither. Work out which end of which band you are on before you decide that a flat fee feels safer.

One more thing on that page worth carrying into every conversation on this list: incrementality testing appears on Skai's grid from Enterprise Premier upward. Read as a category fact rather than a swipe, that means proving advertising caused the sale can be a priced tier rather than a default inclusion. Ask each candidate where it sits for them.

Where Skai wins: published, defensible pricing and a mature multi-publisher platform for a team with a trader in-house. Where Acorn-i wins: you are buying people, not a licence, and there is nobody in-house to drive a console.

Acorn-i vs The Trade Desk — usually a category error, with one honest number

This head-to-head is on the list because people put it there, not because it is a real choice. The Trade Desk describes itself on its own site as "The leading independent DSP built for data-driven marketers". It is buy-side, independent, strongest in connected TV and the open internet, and it is not the route to Amazon retail signal or to Amazon Marketing Cloud. Comparing it to a London Amazon agency is comparing a distribution channel to a team.

What makes it worth a section anyway is that The Trade Desk is publicly traded, which means it publishes numbers that let you do something no private vendor permits: compute a take rate from primary filings.

From their own fiscal 2025 results release: revenue of $2.896 billion against gross spend of $13.4 billion. Divide one by the other and the company-wide implied take is 21.6% of the money that flowed across the platform. The same release reports customer retention "over 95% during the year, as it has for the past twelve consecutive years".

Two cautions on that 21.6%, because a number this quotable is exactly where errors live. It is a company average across every client, including the largest agency holding companies, who negotiate rates that individual advertisers do not get. And gross spend includes media that is not the platform fee's basis in any simple way. So it is not a quote and should never be presented as one. What it is, legitimately, is a public reference point for what a large independent platform earns per dollar transacted — and there is no equivalent figure available for any private vendor on this page, including us.

Where The Trade Desk wins: reach beyond any one retailer, at scale, with a retention record it publishes. Where Acorn-i wins: if the question is Amazon, this comparison was never the right one.

Acorn-i vs Dr. DSP — where we win, and where we do not

Ours is the last head-to-head so you can score it against the same seven questions.

Where we do not win. Acorn-i is a full-service commerce agency with content production, consumer insight and multi-market coverage. We do not make your product photography, we do not run your DTC site, and if your real bottleneck is that your listings and creative are not good enough yet, a media partner is the wrong first purchase. Acorn-i publishes a 94% twelve-month retention claim; we have not published an equivalent figure and will not invent one. And on price transparency, Dr. DSP publishes no number at all, so Skai is plainly ahead of us there.

Where we do win. Cadence and measurement. Dr. DSP is Amazon DSP run as a managed product: Fable 5 reads the account and works it daily, supervised by operators from Full Circle, which has managed more than $500M in revenue across 100+ brands. You choose the autonomy level — every change waiting on your click, routine changes automatic with the larger ones queued, or fully autonomous inside agreed guardrails. Every proposed change carries three things before it runs: the evidence behind it, a measurement plan, and a rollback trigger.

And the measurement is designed before the money moves. Last-click attribution cannot prove incrementality — it only sees the conversions it already claimed. A holdout or a matched control can, and we reconcile DSP against sponsored ads in Amazon Marketing Cloud so the two stop taking credit for each other. That is the same argument several serious competitors now make, and we would rather name that than pretend it is ours alone. The distinguishing question is the one on the scorecard: who designs the test, who computes it, and is it in the statement of work.

The book behind that: across 30 advertisers in July 2026 our Amazon DSP API pull showed 6.04x return on ad spend, 78.4 million impressions at a $4.00 CPM and a blended $1.42 cost per click, with a $5.49 cost per acquisition across 57,137 attributed purchases, 20.1% of them new to the brand. One slice of one month, scoped exactly as measured.

Score the sheet, then check the contract

When the seven questions are marked, one document decides more of your downside than the scores do. Quartile publishes theirs, so use it as the worked example even if they are not your pick: a one-year initial term auto-renewing "for successive one-year periods", sixty days' written notice of non-renewal, fees "due in advance of Services rendered" and no refund on termination, a sixty-day Evaluation Period at the start in which you can terminate from inside the product, and month-to-month available only as a right Quartile "may, in its sole discretion, grant certain Clients".

Draw that on a calendar and the shape becomes obvious. Days 1–60 are reversible. Days 61–305 are committed. Days 306–365 are your notice window. If you also want a sixty-day overlap with a new partner, your real decision date is somewhere around day 245 — eight months into a twelve-month relationship. Diary it the week you sign, on any vendor.

Then ask every candidate, ours included: is the seat mine, who computes the incrementality test, is my media budget inclusive or exclusive of your fee, what notice do I get before a fee change, what data leaves with me, and what happens to my rate if you are acquired. That last one is not hypothetical in this category — Amazon DSP tooling has been consolidating for four years.

Two places to go next if this page did not settle it. If the fee shape rather than the shortlist is what you are stuck on, our page on Acorn-i cost and the four fee architectures works through the arithmetic. If you have already decided to leave and the question is what to move to, the four kinds of Acorn-i replacement covers the handover.

Dr. DSP is a product of Full Circle, and Orbit — the full software suite — is included with it at no extra cost.