Nobody could see who needed to know

A pricing change gets approved on a Thursday. Finance modelled it, the CRO signed off, and the reasoning is sound. It is a good decision, made by the people who should be making it.

The teams that need it are harder to enumerate than the teams that made it. Sales has quotes in flight at the old number. Customer Success has renewal conversations scheduled against pricing that changed this morning. Marketing has a page and a paid campaign pointing at a figure that is now wrong. Partnerships has a reseller agreement with the old tiers written into it. Support has a macro. Somebody in RevOps has to change the CPQ rules before any of it matters.

Nobody in the Thursday meeting produced that list. Not because they were careless, but because the list does not exist anywhere. It lives in the heads of about nine people, none of whom were in the room, and it is different for every decision.

A third of cross-cutting decisions land well

McKinsey surveyed 1,259 executives on how decisions actually get made. When they narrowed to the decisions that cut across more than one team, things like pricing, sales and operations planning, and new product launches, only 34% of respondents said those decisions were both high quality and timely (Three keys to faster, better decisions).

Two thirds, in other words, are late or wrong or both, and that is on the category of decision where the money is. Nobody convenes a cross-functional review to pick a font.

The explanation McKinsey gives is one line long and easy to skim past: leaders cannot see who needs to be involved. Not that they involve the wrong people, or too many people, or too few. That they cannot see. The information required to answer "who is downstream of this" is not available to the person making the call, at the moment they make it.

The same body of research puts a number on the waste. Sixty-one percent of executives say at least half the time they spend making decisions is used ineffectively, and for a typical Fortune 500 company that adds up to something on the order of 530,000 days of managers’ time a year, roughly $250 million in wages (Decision making in the age of urgency).

That figure deserves a caveat we would rather state than have you catch. It measures decision-making inefficiency in general: too many meetings, unclear ownership, rework. It is not a measurement of decisions failing to reach the teams they affect. It tells you the category is expensive. It does not tell you what this specific failure costs, and we will not pretend otherwise.

The organisation chart is the wrong map

The obvious fix is to invite more people. Widen the distribution list, add a standing cross-functional sync, write a RACI. Most companies have tried some version of all three.

They fail for the same underlying reason. Each one is a guess about who is downstream, made in advance, applied uniformly, and then left to go stale. The reseller agreement matters for the pricing decision and not for the launch date. The support macro matters for both. The partner channel matters for neither until the quarter you sign a partner. A standing invite list encodes yesterday’s dependency graph and keeps encoding it long after it has moved.

And they have a cost that shows up immediately. Invite everyone who might be affected and you have built a meeting nobody can decide anything in, which is the other half of that 61%. So somebody shrinks the room again. Speed comes back, the original problem comes back with it, and the company goes round that loop for another year.

Who is downstream of this decision is a question about the decision, not about the org chart.

Two pricing decisions, same people, same room, same week, can have completely different blast radii depending on what happens to be in flight. No fixed structure captures that. It has to be worked out per decision, from what the company is committed to at that moment.

Twenty-eight points sit between the two

PMI has the closest thing to a measured version of the upside. In their 2013 study of project performance, organisations with highly effective communications met their original goals 80% of the time. Organisations with minimally effective communications came in at 52% (The Essential Role of Communications).

The same report is where the number people usually quote comes from: of every $1 billion spent on projects, $135 million is at risk, and $75 million of that traces to ineffective communications. Ineffective communication is the primary cause in about one third of project failures.

So this is not a soft problem that resists measurement. It has been measured, repeatedly, for over a decade. It stayed unfixed because every fix on offer was a process fix, and process does not scale to a dependency graph that changes every week.

The list has to be derived, not remembered

What is missing is not discipline and not another meeting. It is a record of which commitments are currently standing on which decisions, maintained as a side effect of work people already do, so that when a decision changes the set of affected teams can be derived rather than remembered.

That is the thing we are building. A decision goes into a shared memory with who made it, when, why, and what it replaced. The commitments already in flight, the quotes, the campaigns, the renewal plans, are linked to the decisions they depend on. When one of those decisions moves, the set of teams standing on it is a query, not a recollection, and each of them gets the change in their own language in the tool they already use.

We have not proven this out at scale, and it would be dishonest to imply we had. The public research establishes that cross-cutting decisions fail at a measurable rate and that communication effectiveness moves project outcomes by 28 points. It does not establish that our particular mechanism closes that gap. That is the thing we are testing, and we will publish what we find, including if it disappoints us.

In the meantime, the diagnostic is cheap. Take the last decision your company made that crossed three teams. Ask who was told, and when, and how you would know if someone had not been. If the answer is a person’s name rather than a mechanism, you already have the finding.