What I'd say is that if I was going to host something for you, I'd put a lot of margin on it. That's to cover you calling me up and asking me stupid questions and generally complaining about things which I have no control over. It also insulates me from my costs going up and down mid-contract. If you don't want that level of service, then sure, go use someone else to host it for you. You can call up their crappy customer support when the site goes down due to some mistake on their part.
Okay, but, the problem isn't that we're some cluless "head-in-ass" style company, I'm in charge of IT, and I'm more then competient enough to standup the service. I told them that, and offered the server to them. I didn't want them hosting it, they insisted, but, couldn't explain why it was required to be hosteed by them.
I have worked for a 'big' consultancy, and yes, they charged me out at a lot of money - many times more than I was costing them. The customer could absolutely have gone to the open market and hired a contractor with similar skills for a fraction of the money - BUT they'd have had to spend the time doing that (and hiring people can take a lot of time - especially if you're hiring for a role you personally don't understand). Instead, when the consultancy tried to field a couple of flunkies, the customer just said "no thanks" and a few days later, some new people showed up. When the customer said "we need a person", the consultancy found such a person and put them on the project. Likewise, when I left the project (and the consultancy), the consultancy found someone to replace me. All that convenience costs money, and whilst you can definitely argue that the consultancies over charge, they *do* provide a service.
That's a fine argument, that you can get your turn over covered quickly, but is the quality present? Going back to the example of the training, there was no need for them to host it. They rambled on about security and privacy, but got 99% of what they said wrong, or, wrong by association. On top of that, you can't recommend a product that costs X when another is free, without giving a very good defense, can of course they couldn't. Why couldn't I use Moodle? Why? Oh, because their people didn't know to set it up or configure it, or, stand up a Linux server and administer it. So because they're functionally retarded, I should pay them a lot of money to host something for me?
Funny enough they were showing me a demo of the platform after ranting about how serious they take security, and privacy, but had it hosted on a public URL. During the demo I ran scans against the server and it was not configured correctly, I would have failed a high school student for doing that poor of a job. They got really upset about that, and claimed I had no right to run a scan on their infrastructure, but, either put up or shutup.
FWIW, my own view is that the government should use a lot more small consultancies instead of always using the big ones. That is, the small consultancy I now work through would quote day rates far lower than the big one I used to work at, and they have a much higher concentration of good people than the big place had. They can't realistically deliver on a multi-million project, but they could absolutely have done what my previous employer did on that particular customer project. The government could have saved, likely, at least half of the money it cost them to get as far as they did before I left.
That's a fair argument, again, but it's all about what you get out of it. Lets assume it takes one day to setup a server and Moodle. It takes one day to setup a course, badges, users, and other associated work, and, two days to write the training. That's four days of work, we'll round up to five, to make it a week. Accenture would charge at least $25k+, you'd pay a senior employee $1500 / week (after tax). In this example is your output 83% better in a qualitative fashion? Lets assume the weekly rate was $10k, at what point is the output going to justify the cost?
As far as DOGE and the like, there's definitely room to question the consultancies quotes for big projects. The Oracle engagement in Birmingham City Council, for example is a place where the consultancies need to be hauled over the coals a lot more than they are/have been. No project should ever cost >10 times the quote, so either the quote was wrong, or the project was wrong - and the consultancy should have called out which it was at the start.
Perfect example, no argument.
Lastly, AI will indeed knock out a chunk of work which is currently highly lucrative for the consultancies. The rest of their work could be accelerated by AI, and a lot of the long reports they write will be largely AI generated in future. That ought to result in a reduced rate to the customer - we'll see if that happens I guess.
Yes, AI is going to quickly change the consultancy market, and well AI can't replace true skill, talent, and knowledge, as a consultant, you need to be able to explain why you rate is worth it. I'm going to go back to: What do I need to pay another company for, to get me Cybersecurity training? In my experience, a lot of consultants aren't worth the paper of the invoice.