Grocery Shopping and IT Projects?

Someone asked me recently what I thought about the subject of "why IT projects fail", or maybe it was "how can you tell if an IT project is going to fail" or something like that.  I paused for about 0.0000001 seconds and replied: "buying eggs before knowing what to cook."  I got the usual "what a dumbass" look, which I expected.

Let me explain.

Whenever you sit in on a IT project planning (or project status) meeting, and someone asks a question like "What are we doing?", that's when you should finish swallowing your gulp of cold shitty coffee and perk up.  This is the moment when you can tell if the PM (and probably the entire PMO organization in most cases), has a clue of what they're doing.

Response 1: "We are trying to make this widget do this function in accordance with the requirement."

Response 2: "We are trying to find a way to make X do Y, without exceeding Z costs or time."

Response 3: "Are you asking about the entire project goal, or just this one task?"

If the response is not #3, get up, walk (or wheel) to your desk, pack your shit and find another place of employment.  If that's not feasible, find the nearest bar and drink up.

Of all the IT projects I've rubbed-up against in the past 30 years (yes, I'm *that* old), those that failed, or ended up being much less than they promised, EVERY single one of them fell into the mindset of responses 1 or 2.  In other words, they become fixated on the trees, while the forest is being cut down and hauled away.  They remain oblivious and charge onward, to the next pay check.

I was trying to think of some concrete examples, without risking the embarrassment or anger of anyone implicated along the way, so I decided on this one, which happens to have been a real-world situation (in other words: it really happened):

Business Manager: "I want a kick-ass storage system to support our growing kick-ass company!"
IT Manager: "OK!  What's the budget for this?"
Business Manager: "Spec it out and I'll tell you if we can afford it."
IT Manager: (gives weird smirk, as if smelling someone's fart. Then leaves)

A week later:

IT Manager: "Here you go.  100,000 Petabytes and dedicated T-1 lines between every single SSD in the multiple-redundant cube interconnect fabric!"
Business Manager: "How much?"
IT Manager: "$40 bazillion dollars."
Business Manager: "Too much.  Make it work within $10,000."

Another week later:

IT Manager: "Here you go.  $9,999 solution which will do 100,000 Petabytes, but it will be slow and unreliable."
Business Manager: "Too crappy, make it quicker, but come down on the capacity if you need to."

Can you see where this is going? (or went?).

The problem here is that the "PROBLEM" itself was not clearly defined, so the "SOLUTION" cannot be clearly defined either.  You can't cut a key without knowing what lock it's made for.

Buying eggs before you know what you're going to cook for dinner.  Maybe you don't even need eggs.