# How AI Is Changing the Consulting Industry (and Why the Retainer Wins)

Published: 2026-07-22T13:00:00.000Z · Updated: 2026-08-01T13:24:53.000Z · Author: Rod Amora · Canonical URL: https://rodamora.com/blog/how-ai-is-changing-the-consulting-industry-and-why-the-retainer-wins/

> AI won't kill your consulting firm. Your client's ChatGPT might. The billing model decides who keeps AI's efficiency gain: hourly firms hand it back as smaller invoices, retainer firms keep it as margin, as long as the client still feels they get more than they pay for.

McKinsey's headcount is [down about 5,000 from its peak](https://www.thedailyupside.com/finance/are-mckinsey-layoffs-a-sign-the-consulting-industry-is-in-trouble/?ref=g.rodamora.com), with more cuts planned. PwC shed [5,600 people in its 2025 financial year](https://finance.yahoo.com/news/consulting-had-huge-change-2025-223001069.html?ref=g.rodamora.com). A Deloitte executive showed a company town hall a chart in which hourly consulting work shrinks to a ["sliver" of the market by 2035](https://www.wsj.com/cfo-journal/inside-consultants-messy-shift-from-hourly-billing-7bd9b802?ref=g.rodamora.com). It reads like an industry dying. It's an industry repricing. AI is collapsing the cost of delivering consulting work, and the billing model decides who keeps the difference: a firm that bills hours hands the gain to the client as a smaller invoice, while a firm on a monthly retainer keeps it. [58% of clients expect consulting prices to fall](https://sourceglobalresearch.com/blog-post/our-most-viewed-charts-of-2024-how-have-things-changed?ref=g.rodamora.com); they can feel the repricing coming. And the same AI that fattens a retainer now sits in the client's browser. I've watched clients cancel consulting contracts because they decided ChatGPT gives them a good enough consultant for free.

**The short version:** the AI shift in consulting is a repricing of delivery cost, not a collapse in demand. Client adoption of generative AI hit 84% in late 2024 and firms keep buying advice. What's breaking is the hour as the unit of value, and a retainer only holds while the client still feels they get more than they pay for.

## What is actually happening to the consulting industry?

The pain in the headlines is big-firm pain, and it has a specific shape: the pyramid is dying. The traditional model sells a partner's judgment and bills an army of junior hours underneath it. AI now does most of what that junior army did. Research, first drafts, data analysis, slide production. When the hours evaporate, so does the thing the invoice was built on. McKinsey's headcount went from [45,000 at its peak to around 40,000](https://www.caseinterviewhub.com/post/consulting-crisis-mckinsey-bcg-bain?ref=g.rodamora.com). Bain, Deloitte, KPMG, and Accenture all cut staff in 2025.

![Drawing of a pyramid with a finished cobalt-edged apex floating above a base crumbling into sketch lines, tiny figures below.](https://rodamora-uploads-347628392068-sa-east-1-an.s3.sa-east-1.amazonaws.com/illustrations/2026-07-31-ai-changing-consulting-industry/fig-illus-1.jpg "The pyramid loses the hours it was built on")

Watch what the big firms are running toward, and the diagnosis gets clearer. McKinsey has moved [more than 30% of its fees to outcome-based pricing](https://www.wsj.com/cfo-journal/inside-consultants-messy-shift-from-hourly-billing-7bd9b802?ref=g.rodamora.com). PwC says it wants managed services at [20 to 25% of its advisory revenue, KPMG at 15 to 20% of US consulting](https://news.bloombergtax.com/financial-accounting/big-four-firms-embrace-ai-to-revamp-corporate-service-offerings?ref=g.rodamora.com). Those are stated targets, not current revenue. But the direction is unanimous: away from the hour, as fast as their partnerships allow.

Demand is not the problem. [84% of clients were using generative AI by Q4 2024](https://sourceglobalresearch.com/blog-post/our-most-viewed-charts-of-2024-how-have-things-changed?ref=g.rodamora.com), and they still hire consultants to make sense of it. The work exists. The pricing container is breaking.

If you run a $1M to $20M firm, you never had a pyramid to lose. So the question is what this repricing does to a firm shaped like yours.

## Why does the billing model decide who wins?

When you bill hours and AI cuts the hours, the invoice shrinks and the client captures the gain. When you charge a fixed monthly fee and AI cuts your delivery cost, the fee holds and [you capture the gain](https://rodamora.com/blog/what-roi-does-ai-actually-produce-in-a-service-firm/?ref=g.rodamora.com). Firm size doesn't decide who wins here. The contract does.

Clients are telling us the same thing, in a way that looks contradictory at first. [87% say they would pay more for consulting delivered with AI tools, while 58% expect prices to fall](https://sourceglobalresearch.com/blog-post/our-most-viewed-charts-of-2024-how-have-things-changed?ref=g.rodamora.com). Both are true at once because price is detaching from labor. Clients will pay for what AI-enabled delivery produces. They won't keep paying for the hours it used to take.

Billing model

Who keeps the AI gain

Your exposure

Hourly / time and materials

The client. Fewer hours, smaller invoice

Highest. Every efficiency you gain discounts your own revenue

Fixed monthly retainer

The firm. Fee holds while cost to deliver falls

Low on price, high on churn if the client stops feeling the surplus

Outcome-based

Shared. Fee tied to a result, not to effort

Low on price, high on scoping and measurement discipline

I've watched hundreds of service firms adopt AI through a franchise network's delivery data, and the monthly-fee consultancies are the quiet winners so far. They produce more with less effort and fewer people, and they keep charging the same. Most small firms already have the right shape for this: a few seniors, a thin bench, one junior serving many projects, and retainers as the default contract. The pyramid problem the big firms are bleeding from was never your problem. Which makes it strange that some of these retainer firms are losing clients anyway.

## Why would a client cancel a consulting contract over ChatGPT?

I've seen clients cancel consulting contracts and decide to use AI chats as their consultant from then on. Not threaten it in a negotiation. Do it.

The pattern that leads there is predictable. An engagement matures. The deliverables settle into a rhythm, the hard early problems are solved, and the consultant starts coasting on the relationship. Meanwhile the client has been asking ChatGPT their questions on the side, and the answers come back in seconds, written well, and free. The day the client can't tell your monthly deliverable from a chat window is the day the retainer dies. It just takes a few more billing cycles for the cancellation email to arrive.

A retainer survives on perceived surplus. The client keeps paying the fixed fee because they feel they receive more than it costs. Nobody renews a retainer out of fairness; they renew because canceling feels like losing something. AI attacks that feeling from the client's side of the table, one good-enough answer at a time.

## How do consulting firms defend the retainer?

The firms I've watched hold their clients are over-delivering far beyond scope, at AI's marginal cost. Market research the client never asked for. Data analysis on the client's own numbers. Competitor breakdowns, social strategy notes, reading lists, content suggestions. All of it [AI-produced and human-validated](https://rodamora.com/blog/what-is-an-ai-native-service-business/?ref=g.rodamora.com), and none of it possible at the old cost of delivery. A workup that used to take a junior a week now costs an afternoon of agent runs plus an hour of a senior's review. AI made surplus cheap, and surplus is exactly what a retainer runs on.

The targeting matters more than the volume. When you talk to clients, read their data, and pay attention, you start to capture their pain points before they say them out loud. The move is to answer those pains before ChatGPT does. Forecast where the client will hurt next quarter and put the analysis on their desk before they ask. Over-deliver where they need it most, not where it's easiest to generate.

The counterexample is the firm that gets this backwards. A monthly dump of generic AI reports doesn't read as surplus. It reads as noise, and worse, it teaches the client that your extras are just ChatGPT with your logo on top. That accelerates the exact substitution you're defending against. Untargeted over-delivery is how a firm demonstrates its own replaceability.

Done right, this is the same move as [redirecting freed hours into retention](https://rodamora.com/blog/the-extra-time-ai-buys-you-is-already-gone/?ref=g.rodamora.com), applied to the retainer. The hours AI frees don't disappear into slack; they get spent making the surplus visible again. One note: steer away from AI slop. It should never reach a client.

## What happens when the validation step gets skipped?

Human validation is the license to over-deliver. The extras are cheap because AI produces them. They are safe because a person who knows the client checks them before they ship. Skip the second half and the economics invert.

![Drawing of a machine feeding shapes down a belt, a blue figure inspecting them, a broken structure past the belt's end.](https://rodamora-uploads-347628392068-sa-east-1-an.s3.sa-east-1.amazonaws.com/illustrations/2026-07-31-ai-changing-consulting-industry/fig-illus-2.jpg "Validation is the license to over-deliver")

Deloitte Australia learned this in public. In October 2025 it [refunded part of a $440,000 government report](https://www.theguardian.com/australia-news/2025/oct/06/deloitte-to-pay-money-back-to-albanese-government-after-using-ai-in-440000-report?ref=g.rodamora.com) after the report shipped with AI-hallucinated citations, including references to academic papers that don't exist. If a Big Four firm with a methodology department can ship fake citations, so can your one junior with a Friday deadline.

The pressure is structural, and it grows with the strategy in this article. GPTZero's CEO [warned in the same WSJ piece](https://www.wsj.com/cfo-journal/inside-consultants-messy-shift-from-hourly-billing-7bd9b802?ref=g.rodamora.com) that fixed-fee models push firms to produce more output and cut fact-checking corners. Over-delivery raises output further. That's why the validation gate is part of the play itself, not hygiene you add later. Every extra carries your name, and one hallucinated chart can cost more trust than a year of surplus built.

## Will AI replace consultants?

AI replaces consulting hours, not consulting judgment. The production layer of the work, research, first drafts, data analysis, is already moving to AI, which is why firms that sell hours are shrinking. Demand for advice is holding: [84% of clients use generative AI](https://sourceglobalresearch.com/blog-post/our-most-viewed-charts-of-2024-how-have-things-changed?ref=g.rodamora.com) and they still pay consultants to make sense of it, with 87% willing to pay more for AI-enabled delivery. The consultants who get replaced are the ones whose monthly output a client can't tell apart from a chat window. Keep supplying judgment the client can feel, and AI becomes the cheapest capacity your firm has ever added.

## What should a $1M to $20M firm do now?

Three moves, in order of how fast they pay back.

**Audit your billing mix.** Every hourly engagement is a pipe that leaks your AI gain to the client. Where the relationship allows it, move maturing hourly work onto a fixed monthly scope. You don't need to convert the whole book; you need the engagements where AI is already cutting your delivery time.

**Build a pain list per client and run a cadence against it.** Pull it from your calls, their data, and what their industry is about to go through. Once a month, ship one piece of over-delivery aimed at the top of that list. The test for each piece: would the client have asked ChatGPT this question next quarter? Get there first.

**Put a [named validation gate](https://rodamora.com/blog/delegate-the-inputs-own-the-outputs/?ref=g.rodamora.com) on every extra.** One person who knows the client reads every AI-produced deliverable before it leaves the building. Named, not "the team." Unvalidated surplus is a liability wearing a gift's clothing.

On the [Delivery Model Ladder](https://rodamora.com/delivery-model-ladder/?ref=g.rodamora.com), this is a Stage 1 to Stage 2 move. Stage 1 captures the efficiency; Stage 2 redesigns the workflow around what the efficiency makes possible. The retainer defense is that redesign, applied to the oldest question in consulting: why should the client keep paying you next month?