Green on the Slide, Wrong on the Floor

The kickoff where everyone nods

Picture the kickoff. A core system replacement across four plants, eighty million committed, a steering deck that is green on every line. The sponsor opens with the slide about why this time is different. Down the table the plant managers nod. The supervisors nod. The long-tenured operator who has sat through two of these already nods, and then says nothing, because the room is not built for what he would say. Everyone in that room nods, and almost nobody in it believes the date. I have been in that room. The tell is not what gets said. It is the quality of the agreement: fast, frictionless, total. Real readiness is noisier than that. What you are watching is a building that has learned the cheapest response to a transformation is to agree with it out loud and wait for it to pass.

The thing that should bother an executive is that the deck is not lying, exactly. The sponsor really is named on the charter. The portfolio office really does count the initiatives it tracks. The predecessor program really did close green. Each input is true and the sum is fiction, because every number on that slide was produced by someone who had a reason to round it the safe way. The readiness the program gets priced against is an artifact of that rounding. And the literature has been describing the mechanism, in plain language, since before most of these systems were bought.

The one finding that keeps getting reproduced

Start with the most reproduced number in the field, because it is the one most often nodded at and least often acted on. Jeff Hiatt's Prosci benchmark^[1], now twelve editions deep and drawing on thousands of practitioners, finds the same thing every cycle: active, visible sponsorship is the single largest contributor to whether a change succeeds, named roughly four times more often than anything else. Put a number on it and it is brutal. Projects with extremely effective sponsors met their objectives 79 percent of the time. With extremely ineffective sponsors, 27 percent. Same model, same methods, same change office. The variable that triples your odds is whether one executive actually shows up.

Here is the part the deck hides. Prosci's own research names the trap directly: sponsorship gets delegated to the project manager, and the charter keeps reading green while the lever quietly goes slack. The sponsor's name on the slide tells you nothing, because the thing that moves the 27 to the 79 lives in a diary and in whether the floor has ever seen the person. Hiatt built the ADKAR sequence around the same blunt rule, that you find the first letter where a group is stuck and stop pretending the later ones matter, and the letter that almost never shows up in a survey is Desire. Desire is what people decline to write down. It is the one you only get by asking a specific person, in private, about a specific thing they watched happen.

BCG reached the same floor from the engineering side and I trust it because it refuses to be soft about anything. Sirkin, Keenan, and Jackson studied 225 programs^[2] and built DICE, a pre-launch score of four hard factors that reportedly called the outcome correctly in 88 percent of cases. Score in the win zone and you succeed about 79 percent of the time; land in the woe zone and you are down near 33. What matters for readiness is which two of the four factors the executive floor cannot score. The authors say it outright: senior managers routinely overestimate the commitment of the people who must live with the change, and underestimate the extra effort, because nobody up there has ever measured the day job. Those two numbers do not exist on the slide. They exist only in the heads of the supervisors carrying the load, and the room at kickoff is the last place they will be said.

Kotter gets at the same room from a third angle, and his contribution is to tell you what to watch instead of what to ask. John Kotter argues most transformations die at step one^[3], on false urgency. The deck has the look of motion, the task forces and the rescheduled meetings and the happy talk from the top, and underneath it nobody feels the problem is theirs to fix personally. He gives a behavioral test that cuts through the nodding: announce the work and see whether people actually volunteer for it, because a guiding coalition of delegates and a volunteer army that never materializes both predict failure no matter what the readiness survey scored. The nod is complacency wearing the costume of agreement. The volunteer who steps forward unprompted is the only signal that the story landed, and it is a thing you observe on the floor, not a box anyone ticks.

The load nobody is counting per person

Now the second failure, the one that has gotten structurally worse and that I think most readiness work still misses. The deck counts initiatives. People absorb change per person, and the two numbers have come apart violently. Gartner's change-fatigue research^[4] puts hard figures on it: the average employee went from about two planned enterprise changes a year to ten, while their willingness to support change fell from 74 percent in 2016 to 38 percent in 2022. Read those two lines crossing and you understand why the nodding is so cheap now. The same teams have been asked to absorb five times the change and have quietly decided to stop spending real belief on any of it. Prosci finds 73 percent of organizations already at or past saturation^[11], the point where change disruption exceeds change capacity. Saturation is not a portfolio problem you can read off the PMO inventory. It is a per-person problem, and the only people who hold the lived count are the middle managers watching the same five names get claimed by three project leads at once.

McKinsey frames the upside of getting this right, and here I half agree. Scott Keller and Bill Schaninger's organizational-health work^[5] argues that managing the health of the organization, not just the performance targets, roughly doubles the odds of a transformation succeeding, from near 30 percent to near 80. The Organizational Health Index is the most benchmarked instrument of its kind, and I do not doubt the correlation. Where I stop nodding is the same place I stop nodding at every survey: a whole-organization instrument locates the sick unit and the weak practice, and it cannot tell you what the charge nurse or the line lead on that unit already knows about why the practice is weak. The wheel says accountability is low. It does not say that the last person who raised a load problem got labeled resistant and that everyone watched it happen. That second sentence is the one that decides the launch, and no index carries it.

Why the floor has learned to lie up

The reason the cynicism is so consistent is that it is rational, and the research says so plainly. Reichers, Wanous, and Austin^[6] defined cynicism about organizational change as a learned response built from failed prior changes, poor communication, and a lack of any real say in the matter. It is not a personality flaw in the long-tenured operator. It is data about history. He is pattern matching this program to the last three, and he is usually right, because the organization keeps repeating the decay mechanism that killed them. The EY and Oxford Saïd study^[7] sized the cost of fumbling those moments: handle the emotional turning points of a transformation well and outcomes improve sharply, handle them badly and the workforce carries 3.5 times the negative emotion into the next change. That is scar tissue forming in real time, and it is exactly what the veteran in the kickoff is protecting when he nods instead of speaking.

Katzenbach found the lever that actually moves this, and it is not in HR. Jon Katzenbach and the Critical Few^[8] argue you ride the existing culture rather than fight it: find the three or four authentic informal leaders the workforce already treats as credible, the people they walk over to when something confusing comes down from above, and learn from them which behaviors actually get rewarded and punished here against what the posters say. The diagnostic that matters before launch is the gap between the espoused reward system and the real one, because if the behavior the transformation needs is the behavior the floor quietly punishes, no communications budget on earth fixes it. The informal leaders are the only source of truth for that gap, and they are invisible to the org chart by definition. Appoint your champions off the chart and you appoint people with no credibility to carry the thing.

And the metrics that keep these programs looking green are gamed in ways everyone near the work knows about. Training completion gets counted as adoption. Logins get counted as usage. The workarounds literature^[9] documents how systematic this is on the floor: Koppel and colleagues catalogued fifteen distinct types of nursing workaround to a single barcode medication system, and the follow-on research shows these persist for a decade after go-live, almost never logged, because documenting a workaround invites an audit of the person who wrote it down. The new system automates the fiction the official process map records, the dashboard stays green on the gamed number, and the predecessor program that closed green five years ago is, on the floor, quietly running both the new reports and the old ones it was supposed to retire. The closure report said success. The closure report was written to close.

The number that is not actually true

I want to be careful with one number, because it is the one most often quoted and it is the weakest. Mark Hughes audited the famous claim^[10] that 70 percent of change initiatives fail, traced it to its five most-cited sources, and found no valid empirical basis behind any of them. The honest version is narrower and more useful: large surveys put self-reported transformation success near 30 percent, value capture is usually partial, and the defensible predictors are the ones already on this page. BCG's pre-launch zones. Prosci's sponsorship spread. McKinsey's health effect. Gartner's collapse in willingness against the quintupling of load. Quoting the broken 70 to a room of operators is the fastest way to lose them, because the experienced ones know the figure is folklore, and the moment you reach for it they stop trusting everything after it. The real indictment does not need the inflation. It is that every predictor that matters is measurable before the money is spent, and organizations measure none of them, because the only channel they have built for asking is the kickoff room where the answer is a nod.

Pull the threads together and they converge on one mechanism, which is the thing I actually believe. The information that predicts whether a transformation will land does not live in the deck. It lives in the heads of the people doing the work, and it stays there because every channel built to collect it punishes honesty. Tell the steering committee your team is past saturation and you get read as resistant. Write the real reason on a readiness survey and it climbs to your manager with your name on it. Name the five-year-old failure this program rhymes with and you mark yourself in front of the sponsor who ran it. So the supervisor nods, the manager rounds the lived load down to the tracked count, the count averages into a status color, and the color reaches the board as green. Each step is locally rational. The sum is a decision-maker who is structurally the last person in the building to know the one thing the floor knew at kickoff.

The grim joke is that the truth is not missing. Every named expert above is, underneath the framework, describing the same recovery move: get a neutral party to ask a specific person about a specific recent thing they watched happen, somewhere the answer cannot be used against them. Prosci's facilitated assessment instead of the emailed form, which their own research finds nine times more effective. Katzenbach's listening sessions with the informal leaders. The candid interview that surfaces a workaround no audit ever will. None of it is exotic. The reason organizations launch blind is not that readiness is unknowable. It is that it is unasked, because the one room they ask it in is the room the floor learned long ago to nod through.

Which points, finally, at the kind of instrument this calls for. Not another engagement survey, which finds the sick unit and stops there. Not another steering deck, which collects the safe color by design. Something closer to what the canon has described for thirty years and few have run at scale: a neutral, confidential conversation, anchored to a real recent shift rather than a feeling on demand, run for enough people that no answer traces to one person, and read back at the altitude where each one actually knows something, with the sponsor's account set against what the floor reports and the contradictions treated as the finding. That is the kind of instrument we are building at Latent Variables. The literature already told us which people hold the answer and exactly why they keep it. The open problem was only ever reaching them before the date is locked.