Even if they bill $500/hr, and they billed 24 hours a day, that would come out to $4.38m / year for each consultant. That's a 11 member team billing 24 hours a day, all year round, for two years straight.
And if they billed more realistic hours, said team would blow up by many multiples. But of course, billed hours is not the only thing consulting firms will charge.
EDIT: For comparison, the website www.yr.no/en, has I believe 10 - 12 devs working. Maybe they've grown since the past years.
To my knowledge rather consulting firms are great at selling the necessity of lots of consultants consultant days:
Just let the customer talk very openly about their wishes for the project, and you immediately get an insane scope explosion for the project, i.e. it "needs" an insane amount of consultants over many years to implement all these wishes.
To increase the bill, every highly qualified consultant that is necessary for the project "needs" a lot of support personnel (senior consultants) so that the senior consultant can 100 % concentrate on their work (otherwise the customer would pay insane hourly rates for highly qualified experts to do "grunt work" - no customer would "want" that). This way, you sell a huge number of senior consultants (this is rather some low rank) to the customer.
And, by the way: since of cause many consultants you sell to the customer shall be highly qualified experts in their discipline, and the project trivially consists of a lot of disciplines, the number of subject-matter experts that can be sold to the customer can be increased by a lot. In some ordinary software project, you would simply use a small team of good generalists (jacks of all trades, master of none) who can do most things in the project, but of cause, as a consulting company, you rather sell the customer "some of the greatest experts that money can buy" (without mentioning that these are insanely expensive and not really needed for the project).
That's how you do it; scamming or billing unrealistic hours is for amateurs.
"Oh yeah, we can do that!" Boom, there's a team...somewhere...working on it. It's a line on an on-site project manager's status report.
You can't just code the website, zip the code and mail it to the client. They have many stakeholders like this person needs to be able to show this that persin needs to be able to access this etc because they are running a business or service with than many people. Then you will have requirements like blind people should be able to use that and someone should be able to monitor all that. For each complication you will use specialized tools and do integration, i.e. Adobe will sell you one thing Oracle will sell you another thing and you will have to have people overseeing all these integrations and requirements etc.
That's why you have thousands of employees in tech companies with seemingly a simple product.
This is the crux of the issue. If you have outsourced software engineering competency, yet one of your core missions is maintaining a large pile of software, then this is the inevitable result.
I would imagine the margins on that project to be astronomical.
Surely someone can request to see where this went? Even the original figure of $4.1m is insane.
This was Accenture and Deloitte - not some backyard dev shop.
But the problem is with the assumption that the website needs a full overhaul. So often a full overhaul is where projects go to balloon in cost by 20x. An outside agency sells the leadership on a big picture full of fluff about "modernization" without any connection to real improvements.
A better approach would be to determine the most important weaknesses of the existing website, and incrementally improve them. But big organizations struggle with this. Government agencies are probably even worse than big corporations, but big corporations are terrible too.
(It was slightly weird that the old website didn't support https -- but on the other hand, I can't really think of a realistic case where that mattered. And I reckon they could have sorted it out for closer to $0m than $100m.)
Downvote me if you want. But I just built a small business website for a relative in about 5 minutes using Vercel's v0. All I did was upload the logo design, gave it some details about the business and it spit out a fantastic professional looking website in about 1 minute. Made some changes to it and pressed a button to publish with a custom domain and it went live. The entire process took 5 minutes.
I'm sure I can make a weather website with a map for $96k.
- That "weather website" has to serve all of Australia.
- It's got to be usable on big screen desktops, tablets, smartphones.
- It has to have an uptime of what I estimate to be 99.99%. As the article says, farmers will pitchfork you if you can't tell them when rain will hit their fields.
- It has to be slinging dynamic image data to (about) every visitor.
- The data comes from somewhere. You're lucky if they have that under control already. Probably not.
I came up with these aspects, not knowing anything about what the "Bureau of Meteorology" actually needs in a website. It's just common sense speculation.