What an AI Readiness Assessment Should Cost a Small Company

For a company under 100 people, published benchmarks put AI readiness assessment cost at roughly $2,000 to $8,000 for a narrow engagement, rising toward $15,000 as scope grows, and boutique firms in some markets charge SGD $15,000 to $35,000 for the full-report version. Below about $2,000 you are usually not buying an assessment at all but a sales call for a follow-on build. That is the market. My advice sits inside it: a small company should pay a fixed price in the low thousands, get days of work rather than weeks, and insist the deliverable is implemented change (a pilot, a policy, trained people) rather than a thick report. Here is how to read the price spread and buy well.

What are you actually buying?

Strip the branding and every readiness assessment sells the same four things: a map of where your team actually spends time and where AI already sneaks in, a shortlist of use cases ranked by payoff and risk, a defensible tool choice with data-handling terms your lawyer would sign, and a plan your team will actually follow. The packaging varies wildly. At the top end, Pertama Partners' pricing breakdown describes the full ritual for sub-100-employee companies: five to eight stakeholder interviews, a 25-to-35-page report, a readiness scorecard, and a 12-to-18-month roadmap, at SGD $15,000 to $35,000 over two to three weeks.

Ask yourself honestly whether a 30-person company needs a 30-page report and an 18-month roadmap. In my experience the answer is almost always no. You need the first ninety days done right, and the paper ages badly because AI tooling changes faster than roadmaps do.

What do the published benchmarks say?

Aries Consulting Group's pricing guide is the most usable public reference: $2,000 to $8,000 for a narrow small-business engagement, $5,000 to $15,000 for mid-market, $15,000 to $50,000 and up for enterprise. Their description of the SMB tier is worth quoting: "A 1-2 week engagement, typically scoped to a single workflow or single dimension. Light scorecard, narrow recommendations, short walkthrough." They also name the trap at the bottom of the market: offerings under $2,000 are typically "sales calls, diagnostic theater designed to anchor a follow-on build."

The same guide lists what moves a price within a tier: how many tools are in your stack, how many people get interviewed, regulatory exposure, and whether build-versus-buy economics need modeling. Those are legitimate cost drivers. What is not a legitimate cost driver is report thickness.

Why is there a 10x spread for the same promise?

Three reasons. Consultant overhead: a firm with partners, project managers, and a template library bills for all three. Deliverable weight: interviews and long reports consume days that do not change anything in your business. And incentive structure: an assessment priced as a loss leader exists to sell the implementation, which quietly biases the recommendations toward whatever the assessor builds.

The spread is not fraud, it is packaging. But a small company pays the enterprise ritual premium without getting enterprise value from it, because the value of an assessment at your size is concentrated in a handful of decisions: which three workflows, which one tool, what policy, who gets trained. Those decisions fit in days.

How should a 10-100 person company buy this?

Four rules. Pay fixed price against a written scope, never day rates with an open end. Demand that implementation is inside the engagement: a working pilot on one real workflow beats any roadmap page. Check that the tool recommendations come with data-handling terms you can defend, since much of what an assessment surfaces is the unapproved usage I described in Shadow ChatGPT: Your Team Already Uses AI. Now What?. And expect the policy and training to be part of the package, because they are the cheapest risk reduction in the whole exercise.

One boundary worth knowing: a readiness assessment is for deciding where AI should go. If you already have an AI feature or agent in production and the question is whether it works, that is a different engagement with different methods, closer to the audit I described in What an AI Agent Audit Covers, and What It Should Cost. The long version of that production-side story is my explainer, The Agent Observability Gap.

In the interest of showing my own math: my AI Adoption Sprint is the implementation-weighted version of this for companies of 10-100 people, 2 days + follow-up at EUR 1,800 fixed, deliberately priced at the floor of the published SMB range because it skips the report ritual and keeps the decisions.

FAQ

Is a free AI assessment worth taking? Take it for what it is: a sales conversation with homework. You will learn something, and the assessor will learn how to sell to you. Just do not mistake it for independent advice, and never hand over sensitive data for one.

How long should it take? For under 100 people: days of consultant effort, delivered inside one to two calendar weeks. Longer timelines at this size usually mean interview bloat or report writing.

Report or implementation, if we must choose? Implementation. A pilot that works is evidence; a report is an opinion with a cover page. The exception is when you need the document itself for a board or a regulator.

When does the expensive version make sense? When you genuinely have enterprise properties: many systems, regulated data, multiple locations, procurement that requires formal artifacts. At that point the $15,000-plus tier is buying rigor you will actually use.