Implementation Neglect: Why Analytically Perfect Decisions Fail in the Hallway

You’re rolling out a new expense and approval policy across the company. Finance built it, legal signed off, the numbers work, and the leadership team approved it in twenty minutes. You announce it Monday. By Wednesday the regional sales managers are in a side channel comparing notes, because the new approval step adds two days to anything they buy in the field, and nobody asked them what that does to a deal in motion. They don’t argue with you in the meeting. They just slow-walk it. Some keep using the old workaround, some escalate to their VP, and within a month the policy you shipped is half-adopted, openly resented, and back on your desk to rework.

Nothing in the policy was wrong. The math held, the controls were tighter, the rollout plan was clean. What you got was resistance, and the resistance had a source you could have named on day one. The people who have to live inside the policy found out about it at the same time everyone else did. That single fact, more than any flaw in the design, is what decided whether it moved.

The evidence

This is the most reliably documented failure mode in the study of organisational decisions, and the cause runs deeper than the analysis. Across two decades of tracking real decisions inside companies, researchers found that more than half of them failed, and the leading cause was that the decision never got implemented, because the people who had to carry it out had been left out of making it.

The pattern has a shape. The most common way organisations decide is for a leader or a small group to land on a preferred answer and then push it onto everyone else to execute. That approach fails roughly six times out of ten. The approaches that succeed are the ones that pull affected people in early, before the answer is locked, and they fail at less than half that rate. What separated the winners from the losers was the process, not the quality of the analysis.

Underneath that sits something more specific about how people decide whether to back a call that wasn’t theirs. Whether someone commits to a decision depends less on whether the outcome favours them and more on whether they felt the process was fair. And fairness here has parts you can name: were affected people actually engaged before the call, was the reasoning explained, and was it made clear what the decision means for each of them. When those are present, even people who lost the argument will execute the result. When they’re missing, resistance shows up, and it tends to show up as compliance on the surface and drag underneath.

The part that should change how you operate is this. People given a real voice in a process judge the outcome as more legitimate and go along with it more willingly, even when the final call went against them. This is procedural justice, and the load-bearing variable is the experience of having been heard. Being shut out of the process registers as a kind of injustice that has nothing to do with whether the decision was right. That is why a sound policy lands badly: the sales managers were reacting to being handed the approval step, not to the step itself.

How it works

Map every significant decision as a web of people who are affected by it, hold power over it, or both. Ignore that web and you move the hard part from the planning stage, where you can still absorb it, into the rollout, where it comes back as delay and quiet refusal.

The three-list move makes the web visible. Write who the decision changes, who can block or override it, and who you haven’t actually spoken to. The danger sits in the overlap, and especially in one corner of it: a person who is heavily affected, holds real power, and has not been consulted. That’s motivated resistance from someone with the standing to make it stick.

A pre-decision conversation gathers information, but its real work is legitimacy. Each conversation with an affected person before you commit does two jobs at once. It surfaces a constraint you couldn’t see from where you sit, and it gives that person the experience of having been heard, which is what later turns into commitment. Ten minutes with the head of field sales before the policy is final buys you more real adoption than a month of persuading them after they’ve been blindsided.

The voice you spend before the decision buys commitment that no amount of persuasion can recover once someone has already been surprised.

That asymmetry is the whole game. Input given before a call shapes the call and binds the person to it. The same input demanded after the call arrives as a challenge to it, and now you’re defending instead of deciding.

How to use it

Before you finalise any rollout that changes how other people work, spend, or get measured, write the three lists. Who does this touch, who can stop it, and who haven’t I actually talked to. On the expense policy that’s the field sales managers, the regional finance partners, the office managers who process the receipts, and the one VP whose team lives closest to the old workaround. Anyone sitting on two of those lists gets a conversation before you commit, not an announcement after. That mapping takes about three minutes and it’s the cheapest insurance you’ll buy all quarter.

Make the conversation short and honest. “We’re tightening the approval flow. It would add a step to field purchases. What breaks if we do this, and what would make it workable?” Ten minutes. You’re not asking permission and you’re not promising to do whatever they say. You’re giving them a real chance to put their constraint on the table while you can still bend the design around it. Half the time they hand you a fix you’d never have found, like a higher auto-approval threshold for anything under a deal-critical amount. The other half, they feel the difference between being consulted and being managed, and that feeling is what carries the rollout.

Then handle the hard case directly. Sometimes you talk to the powerful affected person and they still don’t want it, and you’re going ahead anyway. Voice was never a promise to obey. So close the loop the way fair process demands: tell them what you heard, tell them what you decided and why, and tell them plainly what it means for their team. “I heard the two-day delay is the real problem. We’re keeping the approval step, here’s the reasoning, and here’s the carve-out we built for live deals.” Someone told that, even when they lost, will usually execute. Someone left to infer that they lost and were never going to be heard will not, and they have the power to prove it.

One more habit pays off later. Write the affected-party map down next to the decision itself. When a rollout dies, the post-mortem almost never asks who wasn’t consulted, because by then the wreckage looks like an execution problem. The note you made at the time is what lets you trace the resistance back to the gap that caused it.

Why it matters

The default model of deciding in most organisations is analytical. Define the problem, gather the data, weigh the options, choose. The buried assumption is that a correct analysis carries its own execution, that being right is enough to make a thing happen. It isn’t, and it fails to be so consistently that it’s the single largest source of decision failure on record.

A decision is a social event before it’s a logical one. It changes who does what, who controls what, and who has to absorb the cost of the change. The analysis settles whether the direction is correct, and the process settles whether anyone actually walks in that direction. You can be completely right about the first and lose on the second without ever knowing the second was where the fight was.

So the affected-party map makes your policy movable, by spending a little legitimacy up front instead of a lot of authority later. A flawless rollout that nobody runs has failed, however clean the spreadsheet. The decent one that the people affected actually commit to is the one that ships.