Discover more from Coté's Wunderkammer
How AI could make management consulting better, improve corporate strategy, and create new revenue for boutique consultancies
Plus, a sad clown cleans the dishes.
This is some really good thinking and actual tactics from Andrew on how big consultancies could use AI. His team works at EY, so they’re not just arm-chair speculating on a Saturday morning while watching Hilda like I’m about to do.
The AI is your team of summer interns
I’m eager to see how much better the responses Andrew got would be with things like the Bing AI: being able to pull in contemporary knowledge and put in links to citations would be great.
The next massive step of awesomeness will be when you can load in your own data. In consulting, with this AI addition, you would still do all the round of interviewing, gather the PowerPoints and memos (if you’re lucky enough to be working with a client that actually writes memos instead of just doing strategy-by-SmartArt), even email threads and meeting transcripts, and then whatever market/SWOT analysis you can get from analyst/consultancy/academic PDFs (you can see how Andrew tunes his prompts with that - but imagine if you could actually load your own PDFs that were locked behind, for example, the Gartner/Forrester/IDC/etc. paywalls).
The humans still have work to do here - a lot of very valuable work. The human element of curating and gathering that information and then massaging it into a helpful collection is incredibly valuable.
Once the humans have curated and, I don’t know, “shaped” all that, you can go to the AI and kind of (I don’t know if this the right term to metaphorically use) Monte Carlo the strategy: “give me ten possible strategies with detailed happy path/risks and downsides analysis.” To put it in a jokey way: “generate the 500 slides we put in the ‘appendix’ of the five slide readout.”
The Head-Meat Work
And then you’d tell the AI to do things like “now, for scenario 3 and 8, tell me how to convince the LoB GMs to agree to putting the strategies in place. Do the horse-trading/haggling, three each of zero-sum and win/win. Next, explain how this strategy aligns to the board’s goals. Finally, drawing on what you know about the 15 financial analyst that cover our company, give me a way of explaining how the strategy will contribute to our FY24 +12% EBITA y/y goal.”
More focus on real work, less focus on shitwork
Far from threatening big consultancy firms, I think this would improve their value. They could truly be long term collaborators with their customers.
The consultants (esp. the senior partners who are doing all those steak dinners with the SVPs and board members - and, again, I mean this in a good way) could continually feed in the “headspace” and undocumented nuances of businesses. They could feed the AI client sentiment.
So much of how businesses actually run is just, you know, what humans think and their world view: you can throw up all the Think Cell charts and McKinsey titles you want, but even a half decent executive knows that there’s selection bias: you could get your slide-drones to stay up until 3am making three more charts that contradict the chart on the screen, or that show an equally, if better, strategy.
All Mekko’s and Minto’s aside, what matters is driving to an actual decision and agreeing on execution and how success/failure will be determined. This is why so many Big Meetings never seem to get past the cover slide. (And, also because these slides have been “socialized” to all the execs participants at the table and their staff behind them on the walls for two to four weeks [God help you if more!] ahead of time.)
A lot of that is something that a good AI could do at a “C-“ level. But then, the team and the senior leadership team have to be like “yeah, so now, what are we going to do?”
Put another way, it would force the consultant and the client to do more valuable work, to think about more important things than all that shitwork of gathering information and putting it into charts and Mintos (the work done by my reoccurring caricature of “the summer interns” on the podcast).
And there’s so much more, for example, having an AI simulate questions for the pre-meetings and The Big Meeting.
Solving the Corporate Prisoner’s Dilemma
If you’re a cynical person, you can imagine creating an AI-driven analysis that aligns to each SVP’s zero-sum positions and be like “look, we all know that if each of you - each of us - came up with a strategy, it’d mysteriously align with your BU and annual bonus plans. Who knew?! So rather than waste time on that, we’ve generated the best case for each of your BUs and added in counter arguments from your zero-sum opponents (that is, the other BU heads). So, we’re just going to slide all that bullshit into the trash and work on a strategy for the whole company instead of each of us fucking around with Prisoner’s Dilemma fuck each other over fuckery. If your plan matches one of these selfish now-garbage plans, come up with a new plan or have a really God damned good case for taking it our of trash can.”
You start with a baseline of “probably bad, selfish strategies” and work from there.
You could also use this create a new line of business going downmarket to clients that previously could not afford the consultancy. If you remove the costs and time of doing all that shitwork (the client has to do this, sending a bunch of stuff to you…you could even use an AI to run the interviews that gather the actual documents and all the stuff that’s only in people’s heads, similar to this), and set the expectation that the output is, you know, what the client could afford, you could probably put together a week long SMB strategy package that relies on a shit-ton of AI work, then maybe two days work with one analyst and a partner to come up with recommendations, then a half day readout, and the chance for followup questions for a quarter or something. I feel like you could bundle that up into a $250,000 package and add in more “features” from there. (Maybe half a million max - you’d have to market-test the reaction to “quarter of a million” versus “half a million” across various industries and clients, then tune backwards to COGS.)
What would be key in the down market here would be the industry expertise that the big consultancy brings. “We do about 30 of these a year, plus the more intense work with the big firms. So we can bring an outside perspective, industry knowledge (not client secrets though, of course), blah blah <smiles and takes out Amex to pay for steak dinner>”
And, so, the math. I don’t know the rates, overhead, and T&E of consultant firms. But, once you knew what your were doing, if we assume 3 to 5 days work by a mid-level analyst, 2 days work by a partner (inclusive of the 1/2 readout), 1/2 day for arts and crafts from the graphics department, and 1 day work for follow-up by email through the next quarter, I feel like the numbers would work well. And I’m guessing these are normal human days - 6 hours of actual work once you subtract out lunch and farting around in the office, filing expenses, and laughing at senior partner’s jokes in the hallway. I don’t know, adjust the package rate as needed. You could also sell a subscription where you revisit it every quarter based on results.
You can imagine the reverse: a boutique consultant firm could sell upmarket (or whatever market, really) by using an AI to do all that work. For example, my old pals at RedMonk would could never do all that shitwork, or even be able to synthesize all the internal lore and culture of a client (all the unspoken stuff - there must be a name for this in consulting? “What they really think” and “how they actually operate,” etc.). The consulting work they do is great, but it doesn’t scaling that kind of thing is hard. Worse, when business is going great, you reach a hard ceiling because you simply don’t have enough analysts to do the work across all the clients.
So the five analysts at RedMonk could get the AI to do all that shitwork and come up with recommendations. Just one of them could do it for an even cheaper package price. (They spent some time coming up with their programming languages product with similar scale-by-automating logic.)
The competitive advantage/USP/whatever would be RedMonk’s taste and head-meat knowledge applied to the curating of the input data, prompting, tuning, analysis, and ongoing feedback/re-analysis. That is, they would apply their artists' sense-making to the projects. The highest compliment they’d get from a client would be “you can just smell RedMonk’s finger-prints all over this work, can’t you? I mean, can’t you just smell that Maine lobster?”
Back when I worked at RedMonk and then 451, I did several quick consults (an hour or half day consult with about a day of prep and “shower time” thinking) and several larger consults (like, two months of work and then a memo/readout) on the topic of “developer strategy.”
It was impossible to do much more than “here’s a bunch of things that seem obvious to me that you’re not doing and industry best practices applied to your portfolio.” As one person, you can’t pull in (and I mean this in a good way) a bunch of numbers and charts. You also can’t take into account all that company’s lore and hidden “how to actually works” stuff. And you certainly can’t run ten scenarios and do that Monte Carlo like stuff.
AI could improve the work, but also create new revenue
In summary, consultancies - big and small - need to start thinking about new business strategies and GTM strategy with AI, instead of fearing it. It’ll be fun to see what Andrew and friends hack-out.
If you like a mix of strategy-think and computers, typos, plus the occasional commentary on breakfast tacos, you should subscribe to my daily-ish newsletter.
You’ll like my three booklets on IT strategy and tactics, all based on real world case studies from large organizations:
Changing Mindsets (O’Reilly) – this book goes over what it means to think in terms of products instead of projects, how to motivate people to change the way they work, and how managers can shift their mindset regarding their daily tasks, goals, and staff.
The Business Bottleneck, free with leadgen (O’Reilly) – once IT becomes agile and all DevOps-y, the bottleneck moves outside of IT to The Business. This book goes over some things the business side should do to break that bottleneck and get on to creating software-driven business models.
Monolithic Transformation, free with leadgen (O’Reilly) collects together the stories and successful tactics large organizations are using to get better at software.
And, hey, we at VMware Tanzu do a pretty bang-up job putting your cloud native strategy in place - you know, agile development, kubernetes, and all “digital transformation” stuff. Check us out! Talk with us! (Email me - I guess? - and I’ll find the right person to talk to?)
My bacon is literarily burning (I’m making weekend breakfast tacos), so I don’t have time to put in links and waste book stuff. We’re not over 500 subscribers, so good job, all y’all.
I mean, if you’re liking these newsletters, I’d sure appreciate you doing some WoM marketing for me and telling other people to subscribe. Especially if you’re one of those super-spreaders.
See you next episode!