Good Transformer
AI training, coaching or consultancy: which does your firm need?
Training builds team skills, coaching builds a leader's judgement, and consultancy builds a fix for one process. Name your firm's shortage before you buy.
AI training, AI coaching and AI consultancy fix three different shortages, so the way to choose between them is to name your firm's shortage first.
Training changes what your team can do, coaching changes what a leader decides, and consultancy changes a specific process. Buy the one aimed at the thing your firm is short of, and treat any pitch that skips that question with suspicion.
Good Transformer sells two of these categories, 1-to-1 coaching for leaders and advisory for teams, so weigh our interest as you read. The comparison below still says plainly where each option, including ours, is the wrong buy.
Why firms buy the wrong help first
The buying usually starts with anxiety. AI is suddenly everywhere, a competitor has announced something, and a client has asked a question nobody could answer. The leadership team wants to be seen to act. Help bought in that mood tends to match whichever provider called that week, and it skips the question of what the firm is short of.
The evidence says firms are adopting much faster than they are learning whether any of it pays. The Thomson Reuters Institute's 2026 AI in Professional Services Report surveyed more than 1,500 respondents across 27 countries. It found organisation-wide AI use almost doubled in a year, from 22% to 40%. Only 18% said they knew their organisation was tracking the return on its AI tools.
That gap between spend and measurement is where bad purchases of external help live. A firm that cannot say what worked last quarter cannot say what kind of help it needs next quarter.
What each category buys you
The three categories overlap at the edges, and providers blur them further by selling all three under one label. The core of each is distinct.
AI training: skills across the team
Training is group teaching: workshops, courses or structured sessions that raise what your people can do with AI tools. It is priced per head and it scales. It is the right buy when the skills gap is real and shared, for example a practice where half the staff have never used an AI assistant on real client work.
It is the wrong buy when it is the whole plan. Skills fade unless the work around them changes. We have set out why AI training fails when the workflow stays the same. The approvals, templates and quality checks have to move too, or people drift back within weeks.
AI coaching: judgement at the top
Coaching is 1-to-1 work with a leader, built around their actual decisions, writing, meetings and planning. It is priced per leader and it stays narrow on purpose: the value comes from being built on one leader's real work.
It is the right buy when the shortage sits at the top. Think of decisions about AI queuing in the leadership team, advice nobody feels equipped to judge, or a leader setting direction for other people's AI use without confident use of their own.
It is the wrong buy when the shortage is elsewhere. One leader's judgement, however sharp, does not roll a tool out to forty staff or rebuild an intake process. A firm that needs many hands trained, or one process rebuilt, should not buy coaching first.
AI consultancy: a built answer to a defined problem
Consultancy brings in outside specialists to analyse, design or build against a defined problem. That might be picking the use cases, redesigning client onboarding, or building the document pipeline. It is priced by the day or the project, and it is the right buy when the problem is specific, costed and beyond skills you will ever need in-house.
It is the wrong buy when the brief is vague. "Help us do AI" produces a strategy deck, and a deck with no named owner changes nothing. It also carries the dependency risk: if the consultants leave and nobody in the firm can run what they built, you have rented capability rather than gained it.
The three kinds of help, side by side
The whole comparison compresses into one table, and the row that decides most purchases is where the capability sits when the engagement ends.
| Compare on | AI training | AI coaching | AI consultancy |
|---|---|---|---|
| What you buy | Group sessions that raise team skills | 1-to-1 sessions on a leader's real work | Specialists working a defined problem |
| Who it changes | The team | The leader, then the decisions the leader makes | A process or system |
| Where the capability sits afterwards | In your people, if the workflow changes too | In the leader's own judgement | In the deliverable, and often with the consultant |
| Right buy when | The skills gap is real and shared | Decisions about AI are stuck at the top | The problem is specific, costed and technical |
| Wrong buy when | Bought alone, with workflows untouched | The shortage is team skills or a broken process | The brief is vague or nobody will own the result |
| It worked if | The team still works the new way a quarter later | Direction is set and decisions stopped queuing | The process runs, measured and owned, after handover |
The test that separates the three kinds of help
The test of external AI help is where the capability sits when the engagement ends: in your team, in your leader, or in a deliverable you now depend on.
Run that test on any proposal and the categories stop blurring. Training aims the capability at your people. Coaching aims it at the person who sets direction. Consultancy aims it at a piece of work. That last is right for problems you should not staff permanently, and wrong for judgement you will need every quarter.
There is a reason no purchase escapes this question. As Ethan Mollick puts it in Making AI Work, "Nobody has special information about how to best use AI at your company, or a playbook for how to integrate it into your organization."
He argues that gains for individual workers do not turn into gains for the firm without organisational change. No category of external help arrives holding your firm's playbook, so every engagement is really a decision about who writes it and who keeps it.
Which to buy first
Our advisory view, offered as judgement rather than data, is this: name the shortage in one sentence. If the sentence names two shortages, fix judgement before skills, and skills before builds. A leader who can judge AI advice makes every later purchase cheaper, because vague briefs and unmeasured rollouts are what expensive help feeds on.
The counter-cases are real. A process that is bleeding money now justifies consultancy first. A firm where nobody below the board has touched an AI tool justifies training first, provided the workflow changes with it. Where the shortage is at the top, that judgement work is what we do 1-to-1 through AI Lessons for Leaders, and where it is firm-wide, an advisory engagement is the better shape.
Sequenced well, the three purchases reinforce each other. A leader with working judgement writes sharper training briefs. A trained team gives consultants real counterparts. A consultant's build survives because people can run it.
How we built this comparison
The categories and criteria come from our advisory practice with professional services firms, and from the published guidance this page sits alongside. That guidance includes our tests for choosing an AI consultant and the five questions we ask before another AI tool.
The adoption and measurement figures are from the Thomson Reuters Institute's 2026 report, linked above. The organisational-gains argument is Ethan Mollick's, also linked above.
The pages currently ranking for this comparison are mostly published by providers of one of the three services, and they tend to conclude in favour of the category the publisher sells. That is the gap this page fills, and the reason we disclose our own interest at the top.
Common questions
What is the difference between AI training and AI consultancy?
Training transfers skills to your people through teaching; consultancy applies outside specialists to a defined problem and hands over a result. After training, your team should work differently. After consultancy, a process should work differently. If a proposal cannot say which of those two changes you are paying for, it has not decided what it is selling.
Does a small firm need external AI help at all?
Often not at first. A firm that has not tried AI seriously on its own real work learns more from a structured internal trial than from any purchase. Set a simple measure for the trial, time saved or quality gained. External help earns its fee when the trial surfaces a gap the firm cannot close alone.
How does a firm know whether external AI help worked?
Decide the measure before signing, not after. Pick the work the help is supposed to change, baseline it, and compare a quarter later. We set out the method in our guide to whether your AI is paying for itself. The category determines what you measure: team behaviour for training, decision flow for coaching, process performance for consultancy.
Use this with your team
These five steps fit into one leadership meeting, and none needs a supplier in the room.
1. Write your shortage in one sentence: skills across the team, judgement at the top, or one broken process. Make no purchase until the sentence exists.
2. Run the capability test on any live proposal: ask where the capability sits when the engagement ends, and reject any answer that amounts to "with us".
3. Ask each provider where their own category is the wrong buy. A seller who cannot name one is describing a product, and their answer tells you more than their references will.
4. Fix the measure before the contract: what changes, from what baseline, checked when.
5. Diarise a review for one quarter after the engagement ends. Judge the purchase then, when the enthusiasm has worn off and only the working habits remain.
Name the shortage this week, before the next provider calls. The firms that buy well decided what they were short of before anyone was selling. If the gap you name sits at the top of the firm, book a discovery call and test our judgement against it.