Traditional Consulting: Becoming Just Less Average
June 2, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I have an AI bias. My personal experience with different smart software systems is what I would call “mostly average.” Somewhere along the line, I learned about research that demonstrated the obvious. Run the same query set against a group of AI models. Run the queries every couple of weeks. At the end of six months. Look at the outputs. The person who explained this research to me wanted me to tell him what the test showed. I said, “Good enough similar outputs.” He looked at me and asked, “How did you know? Did you read the study?” I said, “The methods produce a good enough because that’s how the math works.”
An AI-centric management consulting firm learns a painful lesson. The work was just good enough but not good enough to prevent immediate termination. Thanks, MidJourney. You are indeed good enough.
I encountered this “good enough” when we worked on some projects for a then-famous enterprise search system based on Bayesian statistics. Train up the system, ask queries, and watch the precision and recall diminish over time. Re-train the system and one can observe the same degradation. Why? The answer is that Bayesian and Bayesian-related methods “drift.” Users have novel problems, but the model is not aware of the user’s context and how the words or other symbols operate in the unknown context. Therefore, the outputs have to converge. This is the engine of drift.
I thought about this “drift” and the convergence to the mean when I read “How AI Threatens the Giants of Consulting.” (Heads up: The story sits behind the weird orange newspaper’s paywall. Don’t complain to me. Just subscribe.) The article is interesting, but it did not touch upon this “good enough” angle, nor did it explain where the “value” comes from when hiring a blue-chip consulting firm. That’s okay. To an outside observer, one sees AI replacing a certain type of employee at a blue chip consulting firm. The obvious impact is staffing, billing, and how the work is completed.
The write up makes some good points; for example:
- Blue chip firms developed a system to scale to handle big projects. The write up does not explain what scale means in a large, global study on a volatile topic. Let me add some color: Low tier and mid tier consulting firms struggle, even with AI, to handle certain types of projects. AI would not be particularly helpful in the study of world economic change I worked on years ago. Consultants had to interview individuals in specific institutions and use that information to inform other components of the research project spanning a dozen countries. AI can do some things, but it cannot do this type of knowledge work. Maybe some day? But today to pull this original research off, one has to work for a blue chip outfit.
- AI makes it seem that anyone can set up, market, operate, and avoid litigation by become a next-generation consultant. Okay, maybe. Let me put in my two cents: Setting up an AI centric consulting firm or boutique and staying in business are two different things.
- Today’s consulting firm clients demand fee linked to outcomes, not just hours on time sheets. My observation: That’s true for certain engagements and certain types of firms. However, blue chip firms once enjoyed client repeat business in the 80 to 85 percent range. This means that the blue chip firms capture a client and can bill equivalent fees and present the invoice any way the client requires. A job for a Fortune 50 CEO may be a single line item: “Consulting services for April 2026, $2.4 million.” A government job uses pay grades, hours, engineering change orders or scope changes to generate an invoice in line with he funds appropriated for a project. For a very large client of a blue chip firm, a project can be buried in another on-going project for that firm and described as “research.” Billing has been flexible for as long as the industry has existed.
- AI enabled firms can be fleeter of foot and more like a Cirque du Soleil act. My view: If an engagement requires speed and agility, find a firm that has specific strengths. Blue chip consulting firms, even when AI enabled, have specific, proprietary systems and methods. Sure, a blue chip consultant can jump to another firm, but that transplant has to learn the specific systems and methods at the different consulting firms. Most of the consulting firm’s recipes contain quite specific ingredients in appropriate proportions. Most of this stuff is not taught in MBA school. Feel, style, and selection of ingredients are unique to each firm. That’s why clients of blue chip firms lock in on one or two blue chip firms for the majority of their high end consulting work. The firm delivers the dish the client likes without surprises.
What the article suggests is that the world of blue chip consulting will be roiled yet again. I don’t agree. Blue chip consulting firms did not become blue chip by being “good enough.” Delivering a report anyone can do with a prompt says more about the client’s expectations and the firm doing the work than relevant information about a blue chip firm.
The blue chip consultants have been affected by the corrosive effect of online information and modern technology. Modern business produces winners and losers. That means that firms disappear, get acquired, or decide to merge. The same forces apply to Web search and to the US auto industry.
There are what I call azure chip consulting firms or mid tier outfits. LinkedIn runs ads when one of the better known mid tier firms says it wants to hire an expert in technology X or business sector Y. These firms take work that is too small to be of interest to the blue chip firms. In Manhattan one can hire a research firm to do a study. The firm does okay work, and its clients usually don’t know the difference between what a blue chip firm would have delivered and what the mid tier outfit delivered. Mid tier consultancies have always been cheaper.
There are outfits like Gerson Lehrman-type outfits who “rent” former blue chip consultants to companies working on a budget. These organizations have been around for years and have not killed the blue chip consulting firms. My hunch is that these rental agencies will be able to deliver more value because the individual experts they rent will avail themselves of AI to make a one hour Zoom meeting more useful.
The idea that AI will light the fire of lower cost, AI consulting firms is accurate. However, most of these will not survive. That’s because start ups, regardless of business sector, fail at a rate that would make moms, friends, and clueless family offices sick to their stomachs.
I think the article is interesting. I am not sure that the angle of attack is matched to how the management consulting business operates and has operated for more than a century. I think Booz, Allen fired up in 1917 or so with a project for Sears (may it rest in peace).
AI is a tool. Management consultants are knowledge value professionals who use tools. A tool provides something of use to a skilled user. By itself, the tool does not do much of anything. It’s the know how that produces a hierarchy with the best performers at the top and the wanna-bes at the bottom. AI is good enough right now for some things. I am not convinced it is good enough for context sensitive, fast changing, dynamic things.
Stephen E Arnold, June 2, 2026
Comments
Got something to say?
-
- Subscribe to Beyond Search
Feature archive
News archive
-
