The system and the performer

The work has no face.

A process has no face

If the evidence has been clear for half a century, the obvious question is why every organization still reaches for the person first. The answer is not that managers are careless. It is that one of the two candidate causes in front of them is visible and the other is not.

Walk a floor and you can see people. You can see who arrives early, who looks harried, who asks good questions and who avoids the difficult customer. You know their names, you have appraised them, you have an impression of each of them that took years to form and takes seconds to recall.

Now try to see the process. It has no location. It crosses four departments, three systems and two shifts. Nobody is accountable for the whole of it, so nobody reports on the whole of it. Its worst moments happen in the gaps between the people you can see: in the two days an order sits in a queue that belongs to no one. There is no meeting at which it is present.

This asymmetry decides where the conversation goes. When a result disappoints, the explanations that come to mind are made of the material that was available to the mind, and what was available was people. Nobody chose to look in the wrong place. They looked in the only place that could be looked at.

Four reasons the reflex outlives the argument

Being told that the environment holds the larger gain does not change what a manager does on Monday. Four forces hold the reflex in place, and each of them is rational on its own terms.

The fourth is the most durable, because it is true. People do sometimes cause problems, and when they do, addressing them works quickly and unmistakably. The trouble is that a rule which is right one time in six will feel right every time it is right, and feel like bad luck the other five. Nothing in the ordinary experience of managing corrects it.

There is a fifth reason that belongs to the organization rather than the individual: the people closest to the problem often cannot say what they work around. Reporting that a system forces a daily improvisation is, in many organizations, indistinguishable from admitting you have been doing it wrong. Where speaking up is expensive, the system stays invisible by consent. Pooling 136 independent samples, psychological safety relates to information sharing and to performance across sectors and countries (Frazier et al., 2017); the review of where that literature now stands is Edmondson and Bransby (2023).

What the research calls it

This is one of the most thoroughly studied findings in social psychology, and it has a name. Knowing the name is useful mainly because it stops the conversation being about anyone’s competence.

When people explain someone else’s behavior, they systematically overweight what the person is like and underweight the situation the person was in. The effect was named half a century ago and has been examined ever since. Reviewing what had accumulated, Daniel Gilbert and Patrick Malone gathered it under the more careful label correspondence bias and established something more useful than the effect itself: it is strongest precisely when the situation is hard to observe (Gilbert & Malone, 1995). The name is a coincidence the field has learned to live with; this is not Thomas Gilbert, whose six conditions run through the rest of this series.

That last finding is the one that matters here. The bias is not a general failure to think about context. It is a failure that grows as the context becomes harder to see, which is an exact description of a cross-functional process.

How far this evidence actually reaches

It is worth being exact about how far this evidence reaches, because it does not reach all the way by itself. The attribution studies asked people to explain the behavior of strangers: a student who had read out an essay, a driver who cut somebody up, a candidate in an interview. Nobody has ever sat a management team in front of a delivery report and measured how they explain it.

So the step from that research to your Tuesday meeting is an inference rather than a finding. What carries it is the mechanism, not the setting. In the laboratory and in the meeting the same asymmetry holds: one thing is there to be seen and the other has to be assembled out of records, and the bias grows exactly as the second becomes harder to assemble (Gilbert & Malone, 1995). By that measure a management meeting is close to the worst case there is. The people are in the room, with names and reputations. The process is in four systems and nobody has printed it.

It is a well-supported inference: Rummler and Deming arrived at the same conclusion from inside organizations, by different routes and without the psychology. But it is an inference, and this paper says so rather than dressing it as a result.

The space between the boxes

Rummler put the invisibility in one image that has outlasted every model built on it: the organization chart shows the boxes, and the work happens in the space between them.

Performance can be examined at three levels: the organization, the process that crosses it, and the person doing the work. The first is on every strategy page. The third is on every appraisal. The second, which is where most results are actually made or lost, is on nobody’s report (Rummler & Brache, 2013).

Why the middle level has no owner

The gaps are invisible for a structural reason, not a cultural one. Every function measures its own throughput, and no function measures the wait between two functions.

Zone A reports what it picked. Zone B reports what it picked. Neither reports the two days an order spent between them, because that time belongs to neither and appears in neither system as work. It is visible only to the person waiting for the order and to the customer at the end of the line, and neither of them attends the operations meeting.

An example makes it concrete. Four departments each hold a one-day target for their own step, and each one meets it, week after week. That is four days of work. The promise to the customer is five days. The customer waits nine.

Nobody has done anything wrong and no report shows a problem. The five missing days were spent in the three spaces between the four departments, and a department does not report on a space it does not own. Ask each of the four heads whether their step is on target and every answer is yes, truthfully. Ask where the other five days went and there is nobody to ask, because the question belongs to no job.

That is the pattern, and it is recognizable once you have seen it: every department’s numbers look acceptable while the end-to-end number is unacceptable, and the discussion turns to whether people are pulling together. Deming had the arithmetic for it long before anyone had the diagram: most of what goes wrong belongs to the system, and the system belongs to management (Deming, 2000, p. 270).

The problem with a person’s name

A building-products distributor, 68,000 orders a year, five branches. The figures are illustrative and reconcile from the rates stated with them.

The promise to customers was delivery within five working days. Roughly three orders in ten were arriving late, and the explanation everyone offered was the same: the warehouse had an ownership problem, and specifically it had a supervisor with a reputation. Two branch managers had said so independently. A performance conversation was scheduled.

Nobody had split the number. When somebody did, orders drawn from a single picking zone were running at 4.7 days and orders that needed both zones at 11.2. The split ran straight through the supervisor everyone had named: he ran the multi-zone queue, and every order that arrived late arrived late through him.

What was visible, and what was not

Nothing about this case is subtle once it is written down. That is the point worth sitting with: it was not subtle before either, and it had run for a year.

What was visible: a supervisor, a queue that was always behind, and two branch managers with a consistent impression. What was not visible: a transfer between two zones that nobody had scheduled, nobody timed and nobody owned, and a shift lead in another branch who had quietly solved it and had no reason to think her walk to the far end of the warehouse was worth reporting.

The fix was a scheduled zone-to-zone transfer twice a shift and a timestamp on the handoff, so that the wait would appear in a system for the first time. The timestamp was a configuration change. The performance conversation did not take place.

Both moves belong to the analysis proper: the split is question two and the exemplary performer is question three, taken up in numbers 06 and 07. They appear here only to make the point that the invisible becomes visible faster than anyone expects, once somebody looks.

Five moves that make the work visible

None of these is an analysis. They are what you do in the week before one, and any of them can be done by the person who owns the number, without a budget.

The fifth deserves emphasis because it is the cheapest and the most consistently ignored. Nobody builds a private workaround for fun. Each one exists because the official system fails at a specific point, and the person who built it knows exactly where that point is. All that is required is to ask without making it an accusation, which brings the argument back to why the fifth reason in section 02 matters so much.

A case in which we were the problem

Every case in this series so far ends well, which ought to make a reader suspicious. This one is ours and it went wrong: in the direction this paper is least comfortable with.

A field service operation, fourteen engineers, first-time fix rate 68 percent against a required 82. We ran the analysis exactly as this paper recommends. Nobody was blamed. We followed one job end to end and found a plausible systemic cause: the van stock list had not been revised after a product change, so engineers were arriving without the part in roughly three visits out of ten. The list was rebuilt. It was cheap, it was real, and we wrote it up as the finding.

Six weeks later first-time fix stood at 71 percent. Moved, and nowhere near 82. The service manager rang to ask what we had missed. We did not know, which is the answer no consultant wants to give, so we went back and looked again.

The two were certified. First-time fix reached 78.3 percent and stopped there, still short of the 82, and the last four points we never accounted for. That is where the case ends, because that is where it actually ended.

Why we missed it

The interesting part is not that training turned out to be part of the answer. It is why we, who knew this method and had written about it, did not see two engineers at twenty-three per cent.

We had cut the gap by region and by product type. Both cuts came back uniform (every region a little short, every product type a little short) which is the pattern that says look at what they all share. So we looked at what everybody shared, found something real, and stopped. We never made the cut by engineer, and at that cut the same data was violently concentrated. The records to make it had been in the first extract all along.

The claim, in its exact form

So the claim in this paper needs stating one more time in its exact form. It is not that causes are environmental. It is that the environment is harder to see, and therefore has to be examined first, because the visible explanation will still be there afterwards. Examining it first is not the same as accepting it. An analysis that finds a systemic cause and stops has done half the job, and the half it skipped is the half everybody else does.

What this does not claim

An argument this convenient for one side deserves its boundaries stated first. Three of them, plainly.

• People do cause problems. Sometimes the cause really is capacity, or knowledge, or a person in the wrong role. The claim here is about the order of examination, not about the outcome of it. Look at the work first because it is harder to see, not because the answer is known in advance.
• The attribution evidence is imported. That literature studied how people explain the behavior of others, largely outside organizations. Applying it to how a management team explains a delivery number is an inference, and it is stated as one throughout.
• Making the work visible is not the same as fixing it. A timestamp on a handoff reveals a wait. Removing the wait requires two departments to change something, and that is a different kind of work with a different failure rate.

What can you actually see?

Ten minutes, on the problem in front of you now. Answer these five and you will know whether you are looking at a person or at a system. Leave one blank and you have found the first thing to establish.

Two problems, one method, sixteen papers

Two models carry the method. Gilbert’s six conditions ask what kind of cause; Rummler and Brache’s three levels ask where it sits. Both are sound and this series uses both, and it leads with Gilbert, because his unit is the accomplishment, which is what an organization already counts, and because his six conditions are the categories an organization can actually change.

Where the method comes from

Two further pieces close the set and carry no number, because each points at all the numbered papers. Before the guide sits Arets L&D Business Models, the destination piece for the people who receive the requests: four models a learning function can run, and the route from HR-driven to business-driven. Last sits Solutions that cannot work, the field guide, ordered by the sentence that lands on your desk: seventeen forms a performance problem arrives in, each with an address in one of the numbered papers. Three orders in all: the destination you choose, the question you are on, and the form that landed on your desk.

Sixteen papers, six parts

Number 01 is the map and number 02 is the foundation. From number 05 on, every number is depth on exactly one step of the method. The filled block is the paper you are holding; the two unnumbered pieces close the set.

Where every claim in this paper comes from

The three books named earlier in this paper are listed in full in number 02, so that they do not recur in every reference list.