Hacker News Reader

Comment by notlukesky | Hacker News Reader

My favorite Accenture story is when they charged Hertz $32 million for a website that never went live. Wonder if they will perform better with a smaller headcount.

https://www.henricodolfing.com/2019/10/case-study-hertz-acce...?

View on HN

That's with an American privately owned company. Now you can imagine their deals with corrupt governments in banana republics.

Unless you count the UK as a banana republic (well maybe these days...a turnip monarchy perhaps?) they did manage to fleece the NHS quite successfully:

https://www.digitalhealth.net/2021/05/nhs-digital-denies-con...

I worked with some Accenture people on a gov.uk project not a lot of years ago -- they were hand installing redhat 6 machines that we'd have to request via spreadsheet....

I wonder how this ended ? But I think it is still in the courts. Not that I am a big fan of Hertz, but I hope the win in order to send a message.

Things like this are way too common these days and not only with software, I have seen it myself may times. If I were a CEO (I far from it), I would make sure that the contract states exactly what is needed plus cost recovery if delayed more than 6 months.

edit: typos

Consultancies like Accenture have a vast army of contract lawyers who'll be on the look out for clauses like that, ensuring that they don't end up out of pocket because a project overran or didn't meet expectations.

That initially sounds incredibly shady, however it's not infrequent for projects to fail for reasons outside the contractor's control. Maybe the client can't make up their mind as to what they want, or internal politics mean funding gets cut for the project and it's terminated halfway through. Big contractors absolutely will sign a contract with clauses around deliverables and timelines, but they'll also refer every little deviation from what's listed in that contract to a change control committee to ensure it's accounted for, and doesn't come back to bite them in the ass.

Mostly you don't want to run a project like that, because as you work through the implementation you find variations from what was originally documented, or new edge cases that need dealing with. In many cases you're better off taking the risk of overspend or overrun than ending up with some software that perfectly fulfils a set of requirements that don't reflect reality.

It would be interesting to understand how things went so wrong from Accenture's perspective. I've no doubt Accenture (and every large consulting firm) does have some of top technologists in many different domains, but that's not necessarily who is working on your project. Typically you'd try to balance those high end resources with cheaper offshore talent, and that offshore talent is a dice roll. It's up to Accenture to manage the offshore team and provide enough oversight to detect and correct problems before they make it to the client. This sounds like they were working directly with an offshore team without the onshore resources providing the appropriate oversight. That or they got a dud for the onshore oversight who didn't or couldn't appropriately manage the offshore deliverables.

Consulting companies like that come in two flavors. Both types start with 'you want something we will build it'. When it comes time to do it though you get two wildly different types. One is you will specify every little detail. Think number of pixels from upper left of the screen a control will go. Oh you want it to move with the window and reflow correctly? Oh that's a material change of contract (i.e. more money for them and crap for you). The others seem to have a clue and will just build the right thing in the first place (rare). They both snag you on the back end with support. It is like a box of chocolates you never know what you are going to get.

Having used this particular one in the past. You want to have your details in order before you build anything (waterfall). But at that point you might as well make it yourself in house. But you are already committed to it (sunk cost).

It is great for middle managers who want to still 'code', but not really, but want to be architects. They get a 'cheap' team and maybe look like they are still doing things. I personally got the leftovers of that on a few projects when the manager realized it was basically a double time job to manage his own team and the offshore one too. The manager will get bored with it or overwhelmed by it and realize 'oh wait I need 3 of my devs and 2 PMs to manage this bad boy'.

If you think you have the 'they will build the right thing' kind of guys you need to test for that. Even then a 'good shop' can turn into something else later on as they switch people in and out on you.

Your first method sounds like what is necessary if you're working primarily with offshore consulting. It's primarily staff augmentation, and you've got to own and specify every last detail like you said. This is often where I see consulting engagements go astray. Companies take on staff aug resources without the necessarily skills to get good work out of them.

> Even then a 'good shop' can turn into something else later on as they switch people in and out on you.

YES! Those super talented consultants are in high demand for new deals. They often lead the charge on the proposal, initial project work and early conversations with the client, but may rotate off to focus on landing a new opportunity.

If you’re dealing with consultants then put in the contract that the team that sold you the deal will be the team that delivers it or you walk. It’s not rocket surgery.

I worked on this project as a Lead FE Developer. Management was a shit show and they were dreaming with scope. Our first "epic set" was supposed to be delivered sprint 1, wasn't completed until like 5 months later. We brought it up immediately and were shut down at every step of leadership. Eventually complained enough that I personally got rotated off, but it's to this day the worst project I've ever been on.

To echo thoughtpalette (we were 2/4 onshore FE leads) - There was a lot of miscommunication between Accenture and Hertz in terms of requirements and sizing up the existing architecture. There were a lot of bad assumptions made, inadequate experience with the chosen FE stack and inexperienced offshore developers. Was an absolute clusterfuck.

Reply on news.ycombinator.com