If a vendor's proposal is on your desk this week, this is the test to run before you sign. Five questions, out loud. Three honest “no”s and you don't build, you've just saved a year. That's the whole method. The rest of this is the why, so you can run it without us.
You know the meeting. The plan is funded, or nearly. Someone has shown you something slicker than what you have. The vendor is keen, the deck is good, and everyone in the room is nodding because nodding is easier than being the person who slows it down. The decision is basically made; you're looking for a sanity check you can tell yourself you did.
That's the moment this is for. Not because the build is always wrong, sometimes it's exactly right, but because that room has no one in it whose job is to argue the other side. Here are the five questions that do.
The five
1. Is it costing you real money this quarter? Not “could it pay off eventually”, is it bleeding now: lost hours, double entry, a customer you can name. “This quarter” is the load-bearing phrase. A cost that's always twelve months away is usually a story, not a cost.
2. Has something you already pay for actually failed at this? Most “we need a new system” is a tool you own, configured by nobody, and never turned on. Before you buy the replacement, make someone prove the thing you have can't do it. Out loud. With the actual feature open.
3. Would you still want it at double the price and double the timeline? Because that is roughly what you'll get. If the answer is still a fast yes, it's probably real. If it wobbles, the appeal was the brochure, not the need.
4. Can you name the one person whose day this makes easier, on day one? A specific human, a specific task, the first week. If the benefit is “the business will be more scalable,” nobody is waiting for this and nobody will adopt it.
5. Will it still matter in eighteen months? If it's a patch on a problem you'll have outgrown or re-org'd around by then, the cheapest fix is the right one, duct tape is a legitimate engineering decision when the thing is temporary.
Why three no's is the line
One “no” is noise, every real project has a weak spot. Two is a conversation worth having before you commit. Three is the system telling you the want is louder than the need: the cost isn't biting, you haven't ruled out what you own, and you can't name who it's for. At three, the expensive path is almost always solving a configuration, training or process problem with a purchase. Buying is just the most visible way to look like you're addressing it.
Note what this test refuses to do: scare you. There's no “competitors are pulling ahead,” no urgency you didn't bring yourself. That's deliberate. Manufactured urgency is how the want gets sold to you as a need, so the test simply doesn't supply any.
You don't need us to run this. Take it into the room, ask the five out loud, count the no's, and you can defend the call to your board without mentioning our name. That's the point of writing it down.
If you got three no's: don't build this quarter. Write down what would have to change for it to become a yes, and revisit when it does, not before. If you got fewer than three: you have a real one. Now the question stops being “should we” and becomes “what's the smallest version that closes the gap”, which is a different, much cheaper conversation.