Skai pricing — the tiers they publish, and the arithmetic underneath them

What is on the page today, and why that is unusual

Read on 20 August 2026, Skai's pricing page lists five bands, each keyed to annual advertising spend running through the platform. Standard is priced at $114,000 a year for advertisers up to $4M in annual spend. Advanced is $276,000 up to $10M. Enterprise is $504,000 up to $20M. Enterprise Premier is $756,000 up to $35M. Above that, Skai says pricing is tailored. The page also promises "predictable annual pricing and the flexibility to review your commitment after the first three months."

Take a second to notice how rare that is. Skai's nearest neighbours — Pacvue, Quartile, Criteo, Rithum — route every pricing question to a demo booking. None of that is sharp practice; enterprise software has always been sold that way. But it does mean a buyer comparing four vendors normally has one real number and three blanks, and a published list is worth something even if you never buy: it anchors every other quote you receive.

One clarification before we go further, because this page sits in a category where a three-letter word does double duty. Where we say DSP we mean a demand-side platform — the software that buys programmatic display, video and audio. Amazon also runs a Delivery Service Partner programme for parcel-delivery franchises. Different business entirely. We are Dr. DSP, an Amazon demand-side platform product from Full Circle, a full-service Amazon management company with more than $500M in managed revenue across 100+ brands.

The number that matters is not the tier price, it is your position in the band

A tier priced by a spend ceiling is cheapest at the ceiling and dearest at the floor. Take the Standard band. Divide the annual licence by the top of its range and the licence works out at under three cents on every dollar of media. Run a quarter of that spend through the same licence and the same fee is over eleven cents on the dollar — the identical product at roughly four times the effective rate.

So the first thing to do with any banded price list is locate yourself, not admire the headline:

• Find your true annual figure. Not last year's, and not the plan — the run rate implied by your current quarter, multiplied out. Banded pricing punishes optimism.
• Measure your distance from the next boundary. If a good year would push you across a threshold mid-contract, ask now what happens. Does the tier change on renewal, or the moment you cross?
• Ask what counts toward the band. Spend through the platform, or total spend on the retailers it touches? Managed spend and observed spend are not the same quantity, and the difference is the tier.
• Price the alternative honestly. Against a licence at the bottom of a band, a managed percentage of media can be materially cheaper. Against one at the top, it usually is not. Run both.

That is not an argument against Skai. It is an argument for doing the division before the demo, because the division is what turns a number on a page into a number about you.

Read the feature list for where measurement sits

The tier contents tell you as much as the prices. Every band from Standard upward includes access to Skai's publisher network and its AI layer. Advanced adds competitive insights and search term analysis. Enterprise adds expanded read-only publishers, customisable audits and QA, and broader customer-journey intelligence. Incrementality testing appears at Enterprise Premier, alongside white-glove onboarding and custom development.

That placement is worth pausing on, because it is the industry's habit made visible. Optimisation is treated as the core product and proof is treated as an upgrade. It is a defensible commercial decision — incrementality work is genuinely expensive to deliver and needs a scale of spend to be readable. But it does mean that a buyer in the lower bands gets a very capable machine for improving reported performance and no built-in way to check whether the reported performance was real.

This is the argument we make on every page we write, so we will make it briefly here. Last-click attribution cannot prove incrementality — it can only tell you which touch was nearest the purchase. Display is the format where that gap is widest, because retargeting always reports beautifully and often sells nothing that was not already sold. The fixes are old and unglamorous: withhold display from a matched set of regions or ASINs and compare against a control, then reconcile in Amazon Marketing Cloud, where DSP and sponsored-ads events sit in the same clean room and stop each claiming the same order. Amazon's own documentation describes AMC as available free to eligible advertisers, which means the reconciliation is a question of will, not budget.

What a licence price never includes

A published figure is only the first line of the bill. Before you compare it to anything, add the lines that are not on the pricing page:

• The media itself. Obvious, and still the line people forget when they compare a licence to a managed fee that is quoted as a percentage of media.
• The operator. Platforms do not run themselves. Budget a salaried trader, or a fraction of one, and be honest about how many days a month your account will actually get. Software plus a genuinely dedicated operator beats most services. Software plus a busy generalist loses to almost anything.
• Amazon's own floor. Amazon states that its DSP self-service option carries no minimum spend requirement, while its managed service typically requires a minimum investment of $50,000, varying by country. Whether you need that route changes the shape of the whole plan.
• Integration and data work. Clean rooms, feeds and warehouse connections are people-time. Ask who does it and whether it is inside the fee.
• The exit. Annual licences renew. Find the notice period on the day you sign, not the month you want to leave.

One more piece of context a buyer should have: in August 2024 Digiday reported that Criteo was in M&A discussions with Skai. No transaction was announced and Skai continues to sell under its own name and publish its own list. It is worth a change-of-control question in any multi-year contract in this category — not because anything is wrong, but because the retail media software market has consolidated hard and your renewal terms should survive that.

Who Skai suits better than we do

We would point a reader to Skai without hesitation in three situations, and we mean it.

• You buy across search, social and retail from one desk. Skai's history runs back to Kenshoo and paid search; the breadth of publishers under one login is the product, and an Amazon-only specialist cannot match it.
• Your procurement team needs a list price. Some organisations genuinely cannot start a process without a published figure to benchmark. Skai lets them.
• You want to trade in-house. If you have a team whose job is the console, buying the console is the right purchase, and paying a service to do work your own people want to do is a bad trade.

Where we would argue for ourselves: when Amazon is the whole business, when nobody internally has DSP as their actual job title, and when the thing you need is not more optimisation but a defensible answer to whether the display budget is producing incremental sales at all.

How Dr. DSP prices, and the two numbers we bring instead

We publish no price for Dr. DSP, and we would rather say that plainly than dress it up. It is a demo, the first 30 days free, and a price set on the call against your real media budget and scope, month to month, with Orbit — our software suite — included at no extra cost. Every proposal we put in front of a client carries three things attached: the evidence behind it, how it will be measured, and the condition that automatically reverses it. You choose the autonomy level and can change it whenever you like; inventory risk, pricing, new products, new creative and the decision to stop spending on DSP always come to a human.

What we can put on the table in place of a list price is a benchmark. From a direct Amazon DSP API pull covering 30 advertisers in July 2026 — not the whole book — the set delivered 6.04x return on ad spend across 78.4 million impressions, a $4.00 CPM, a blended $1.42 cost per click, $5.49 cost per acquisition on 57,137 attributed purchases, and 20.1% new-to-brand. One month, Amazon-attributed, no guarantee attached to any of it. Hold it against whatever a vendor shows you.

Two honest redirects. If the problem is sponsored ads rather than display — search terms burning budget, bids nobody has touched in a quarter — Dr. PPC is the right door, and it publishes its price. If you would rather buy this capability as an agency engagement with a named team rather than as a product, reMKTR does the same media buying agency-side.