Sitting in the negotiating chair can be painful when you have little to offer.
Almost ten years ago, I was in a boardroom in Shanghai trying to close an insurance deal on behalf of a FTSE company.
Other than a lot of lipstick to deploy on pigs that were widely available anywhere, our side had very little to offer beyond brand cachet. The other side had the technology, product and network.
During the negotiation, they took out their laptops and put on screen the live pricing of our weak pigs next to other, non-exclusive pigs with better lipstick on them, to show our side how bad our offer was.
It was a tough gig. We had been outvalued and outplayed.
In China, deal-making and contract writing in the boardroom are often treated as a market, openly and without apology. It can be frustrating at first, but once you understand the dynamic, it opens a lot of opportunities.
For that reason, western consultants have often found it hard to get a solid footing there. Deal rooms in China don't accept fog or things that can't be measured easily, and I had to admire it.
The AI deal room
This is precisely the stance buyers are now starting to take when dealing with consultants everywhere.
Consultants have been supermarketified.
Buyers will put the live pricing, value or outcome on the screen, ask why your pig costs three times more than the AI one next to it, and expect a straight answer.
The first big oops
I recently gave a talk at a consultancy. The exam question was "in the context of AI, what does our business look like next and what changes do we need to make?" I have been asked the same question several times now. Here is my view, starting with the biggest oversight.
The industry is restructuring itself around an assumption that is already out of date.
Young people are not waiting around to be redeployed. They're doing it for themselves at speed.
They have time, motivation, and native fluency with the AI tools, and they are using it. The 22-year-old who spent two years building agents in their bedroom has more practical AI capability than the mid-career manager firms are hiring at a premium. They have already leapt ahead but they don't know it yet because their pay packet hasn't caught up.
It's not the AI tools you should be worrying about. It's the people who know how to use them and show you up in the boardroom. I've been showing companies how to offer their customers software it would take months to build - and I wouldn't say I'm at genius level of this game.
Executives and employees at capacity cannot find the time to learn the tools.
Judgment needs apprenticeship, but tool fluency doesn't. The juniors are already ahead, and the firms that cut graduate intake will be paying premium rates to bring them back in three years.
Outcomes, not time
For thirty years consulting has been protected by the same opacity that China's boardrooms challenged.
Buyers had no real way of knowing whether a twelve-week engagement at £400k was good value.
Clients now have much more to benchmark good value on. You will be subject to pricing wars unless you can prove quality. Decks will not cut it. Advice, skin-in-the-game and deliverables are now tied to the outcome to get paid well.
AI is forcing work to be measurable. Once it is measurable, everything underneath shifts at the same time.
The new scorecard
Buyers will compare consultancies on three axes sitting next to each other. Outcome rate. Quality. Price.
Outcome rate is the share of things you said you would deliver that actually landed.
Imagine being rated on TrustPilot for outcome rate.
The AI customer support world is already living this. Vendors publish the difference between deflection, where a ticket ended without a human getting involved, and resolution, where the problem was actually solved.
A platform can show 90% deflection with only 40% resolution. Did the engagement solve the problem or did it just end?
Quality is where the real fight happens.
Anyone can claim a delivery in 30 days. The question is whether the work held up six months later, whether the regulator accepted it, whether the board believed it.
Price becomes the third axis once the first two are visible. When a customer service agent resolves a contained ticket for 46p versus £4 for a human, the cost of delivery collapses. Firms holding old day rates without holding outcome and quality become the easiest to displace.
What it looks like in practice
In pharma regulation, a four-person boutique with a former regulatory affairs director and two analysts can replace what used to be a twenty-person engagement, priced per submission accepted first time and measured on regulator feedback cycles.
In financial services strategy, a former bank strategy director with two analysts running parallel agent-driven market scans delivers a fixed recommendation in four weeks, with a retainer to follow paid only against revenue retention targets.
In sustainability, SEC climate rules demand scope 3 emissions data nobody actually has, and a small team with carbon accounting analysts and a supplier-data agent can deliver audit-grade coverage in 12 weeks priced on percentage achieved.
Why the big firms are exposed
Big firms are slow to flip because the model breaks them from the inside.
Buildings, partner ladders, utilisation targets all sit on hours. Moving to outcomes means writing down the value of the practice that pays for everything else, so they delay, and they write decks about agentic AI while small senior teams are deploying it.
Small and mid-sized firms can change the price sheet on Monday and start selling outcomes by Wednesday. That window probably lasts 18 months before the big firms copy enough of it to compete.
The pricing trap
Rushing at this before understanding the AI supply chain however is risky.
Outcome economics work today because model pricing is low and venture-subsidised. Frontier labs are nowhere near charging the true cost of compute, and if model prices rise three to five times over the next few years, which is plausible, the maths flips.
The consultant who priced an outcome at £50k against £8k of assumed compute is suddenly looking at £30k of compute and a loss. The client who fully outsourced a function, fired the team that used to do it, and built around a single vendor stack has nowhere to go when the renewal arrives at triple the price.
Smart sellers are pricing with margin, indexing contracts to model costs, and avoiding single-vendor lock-in. Smart buyers are keeping a thin internal team that remembers how the work used to be done.
(Index your contracts to model costs << we'll do another post on this soon.)
The upside
The downside arguments dominate the conversation, but once you have agentified your business properly the upside is significant.
The marginal cost of running the next engagement drops. A two-person boutique that used to handle three clients can now handle ten. An eight-week engagement compresses to three.
Diagnostic work that used to take a fortnight is done in two days, which means the partner can spend their week on the thing only they can do, which is judgement, creativity and the conversations that actually move the deal forward.
So the picture is not just smaller firms doing cheaper work. It is small firms doing far more work, faster, with higher margins per partner than the old pyramid ever produced.
That is the real opportunity, and the firms that focus only on cutting costs to defend their existing book will miss it entirely.
The new shape
The pyramid is how consulting has always been built. A wide base of juniors doing the research and the decks, a layer of managers running the engagements, a thin top of partners selling the work and holding the client relationships.
The economics depend on the base being big, because that's where the margin gets made.
Big firms are moving to a diamond. I think they are dead in the water for this.
The diamond has fewer juniors at the base because AI does the research and the slides, a fat middle of mid-career managers orchestrating the agents and bridging domain with technology, and a thin top of partners holding the diagnosis.
How many mid-career, full-time employed people do you know who have built an agent? How many have agentified their business?
The diamond is the least disruptive answer, which is why the big firms like it.
It also brings every problem that comes with middle-manager-heavy organisations, including slow decisions, politics, coordination cost, and each manager needing a reason to exist.
Boutiques skip the middle entirely - and just focus on the outcome.
The next consultant is a senior operator, AI fluent, fixed outcomes. That person owns an audience, plugs into a partner network with a productised offer and uses AI native juniors well.
I think niche, small operators and boutiques are in for a great ride if they get their tooling and mindset right.
What you can't fake
The work is still spotting the real problem underneath the one the client thinks they have.
The other thing you can't fake is network. The people they can call to solve a problem at 9pm on a Tuesday, the peers who'll tell them what's actually working, the audience that trusts them enough to hire them without an RFP.
Community is now a working asset rather than a soft one.
Where this leaves us
If you're a buyer, you've never had more leverage. If you're a seller, you've never had more upside, but you have to actually flip the model rather than describe flipping it.
The new consultancy model
- Outcome pricing kills the day rate. Paid for results, not hours. Zendesk, Intercom and Decagon are already there, and 70% of buyers reject per-seat pricing.
- Senior-only beats the pyramid. Boutiques deliver 5-10x faster at lower cost, and partners are walking out of the Big Four to do it.
- Run the function, don't advise on it. Agentic AI makes managed services high-margin. Operate the process rather than recommending changes to it.
- Tight loops, not slideware. AI work is iterative. The phased "assess, recommend, implement" sequence is dead.
- Community keeps you relevant. Trust, judgement and access are the scarce inputs when delivery commoditises. Curated peer rooms, partner ecosystems and an owned audience.
How many partners do you know who are doing that?
Book The AI Stress Test workshop with Dan
If you don't know where to begin with AI - or how to think about it clearly - book an AI stress test with me to cut through the noise and find the right place to start. In a one-day workshop, I help you work out what AI actually means for your business, where it will pay back fastest, and how to think about the choices in front of you. One day. Clarity you can act on. Book it NOW.
