The purpose isn’t to explain yesterday’s failure. It’s to recognize tomorrow’s before it happens.
Procurement Insights · Jon W. Hansen, FCIPS
The Hansen Case Labs run six cases: Santa Clara County, Thomas Cook, Virginia eVA, North Carolina, Bay Pines and the VHA, and I-35W. A seventh session — a sixteen-year public argument between competent practitioners that has never settled — runs as the encore.
Take one of them. Most people know how Thomas Cook ended.
They do not know about 1,100 IT suppliers. Or more than thirty data centres. Or more than a hundred desktop images and around five hundred IT projects, after twelve acquisitions in three years. Or a Lessons Learnt slide carrying five generic statements after twelve months of transformation.
That material sits in a deck presented publicly in 2010 by the group procurement director, and handed to me at a conference I was hosting. Nine years later the company filed for compulsory liquidation.
How many people still have that deck?
That question is the whole subject of this post.
Two very different artifacts
When an aircraft goes down, two things eventually exist.
The first is the accident report. Investigators work backward from the wreckage, reconstruct the sequence that produced the outcome, and publish their findings.
The second is the flight recorder. Not a conclusion — a record. What was actually said and done, minute by minute, by people who did not yet know any of it mattered.
Those are not the same thing, and only one of them lets you sit in the seat.
The published account of any failed initiative is the accident report. Somebody assembled it after the outcome was known, from what could be recovered, and named a cause. That is what forms most people’s view of Thomas Cook, or Santa Clara, or any of the others.
Think of the Procurement Insights archive as the flight recorder.
Internal documents sent to me while an implementation was still running. An award submission written by the people doing the work, while they were still doing it. A recorded interview a County Executive agreed to give when the questions were uncomfortable. A transformation deck handed over in 2010 by the firm that produced it.
None of that was written to explain a failure. Most of it was written by people who expected to succeed.
That is the property that matters, and it is not the same as the material simply being old. It was created before anyone knew what it would turn out to be evidence of — which means none of it was selected, framed or emphasised to support a conclusion that did not yet exist.
Why the recorder is credible
A flight recorder is not trusted because the investigator is trustworthy. It is trusted because it was running at the time and nobody edited it afterward.
That is the entire basis on which the archive works, and it is why the rule has never changed here: published once, never edited. Where a post has been updated, the update is appended above the original with its own date and the original left intact. The timestamps are in the platform metadata. Anyone can check them.
Which means the numbers in the archive do not always agree with each other. A figure reported in 2008 may differ from the same figure reported in 2016. That is not a defect — it is the proof. Across nineteen years of contemporaneous reporting, a record that agreed with itself to the decimal on every retelling would warrant more suspicion, not less, than one that preserves its seams.
One honest limit, because the analogy would otherwise flatter me. A real flight recorder captures every channel for the whole flight. Mine does not. It was running on some channels and not others, kept by someone who was present for parts and absent for others. There are gaps.
What it holds is contemporaneous, dated, and unedited. That is the property that matters, and it is rarer than completeness.
The debris field
Here is what most discussion of case teaching misses.
An investigation does not begin with a narrative. It begins with wreckage scattered across a search area, and most of it is missing. The pieces recovered are recovered because somebody went looking. The report at the end is assembled from what was found — never from everything.
And a piece of wreckage does not arrive labelled.
Nobody hands you the fragment and says this one matters. You have to recognize it among the pieces that do not, while you are still standing in the field.
That is the actual skill. Not analyzing evidence you have been told is significant — recognizing significance in material that arrives undifferentiated.
Santa Clara turns on a decision made in 2003, described in eight words in a background section of a committee document, with no reasons given. Nothing marks it. Everything downstream follows from it.
Why knowing the ending does not spoil it
Most of these outcomes are public. Anyone can look up what happened to Thomas Cook, or read the Civil Grand Jury report adopted in June 2024 on Santa Clara County’s procurement.
So people arrive knowing the ending, and usually with a view already formed about why.
That view survives contact with the published account — because the published account is what formed it.
It does not survive contact with the recorder.
Knowing a company collapsed tells you nothing about which document mattered, or which eight-word line carried the whole thing, or which of six competing signals was the one to act on in the third month. Knowing that a plane went down tells you nothing about which piece in the debris field explains why.
Investigators work backward from the outcome to establish cause. Then simulators put pilots back at an earlier point and run the conditions forward, so they can recognize the pattern developing while there is still time to change it.
Trace backward to establish causality. Move forward to develop recognition.
The pilot in the simulator often knows which flight it is. That is not the exercise. The exercise is whether they can see the moment it was still preventable, and what they would have done differently while standing inside the uncertainty rather than looking back at it.
What you are in the cockpit to do
The labs place you in the seat and play the record forward, decision by decision. At each stop, three questions:
What would you do in this situation? Why would you do it that way? How would it have affected the outcome?
You commit before the next piece arrives. Then the boundary advances and the record shows what the organization actually did and what followed.
Your reasoning is then mapped against the strands the record shows were determinative, along with one, two or three plausible alternative paths that could have followed from a different decision.
Every one of those alternatives is treated as a projection. What happened is documented. What might have happened is not, and never will be. Where the analysis produces competing scenarios, the disagreement is shown rather than resolved — because a single confident answer about a path nobody took is exactly the shape of evidence this practice exists to distrust.
And where the record is silent, it stays silent. Nobody knows why Santa Clara declined the procurement module in 2003. That silence is part of the case, not a gap to be filled.
What does not transfer
Here is the thing that has to be said plainly, because it is the objection the format invites.
Doing the same steps in your organization will not produce the same outcome.
I wrote about this in December, quoting Bob Sievert — who ran one of the cases in the set:
You cannot implement to someone else’s outcome. You can only implement from your actual starting point.
You do not leave with Virginia’s answer, or Santa Clara’s. Those belong to Virginia and Santa Clara. They were produced by those organizations’ people, constraints, incentives and absorptive capacity, none of which you have.
What transfers is the capacity to recognize which strand is load-bearing in a situation you have not seen before.
And that can only be built on somebody else’s situation — because in your own you are inside it, invested in it, and already holding a view. Someone else’s debris field is where you learn to read one.
The purpose is not to copy a championship. It is to develop a new and expanded way of thinking critically about the one you are actually in.
What this is for
The industry produces a great deal of analysis about why initiatives fail. Most of it is sound, and I have no argument with the conclusions. Analyst research tells you the shape of the distribution, which matters — you cannot learn how common a pattern is from one case.
But you cannot practise on a percentage.
The purpose is not to explain yesterday’s failure. It is to recognize tomorrow’s before it happens — and recognition is a skill you can only build by sitting in the seat before the outcome is known, with the evidence arriving in the order it actually arrived, and committing to a call before the next piece lands.
That is what the flight recorder is for. And it is why the question I opened with is not rhetorical.
How many people still have that deck?
The strand mapping and scenario work described above are performed using the Hansen Strand Commonality™ Model and the ARA™ RAM 2025™ multimodel platform.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
Hansen Models™ · Procurement Insights
-30-
Related
The Accident Report Names a Cause. The Flight Recorder Lets You Sit in the Seat Before the Outcome Even Happened.
Posted on August 23, 2026
0
The purpose isn’t to explain yesterday’s failure. It’s to recognize tomorrow’s before it happens.
Procurement Insights · Jon W. Hansen, FCIPS
The Hansen Case Labs run six cases: Santa Clara County, Thomas Cook, Virginia eVA, North Carolina, Bay Pines and the VHA, and I-35W. A seventh session — a sixteen-year public argument between competent practitioners that has never settled — runs as the encore.
Take one of them. Most people know how Thomas Cook ended.
They do not know about 1,100 IT suppliers. Or more than thirty data centres. Or more than a hundred desktop images and around five hundred IT projects, after twelve acquisitions in three years. Or a Lessons Learnt slide carrying five generic statements after twelve months of transformation.
That material sits in a deck presented publicly in 2010 by the group procurement director, and handed to me at a conference I was hosting. Nine years later the company filed for compulsory liquidation.
How many people still have that deck?
That question is the whole subject of this post.
Two very different artifacts
When an aircraft goes down, two things eventually exist.
The first is the accident report. Investigators work backward from the wreckage, reconstruct the sequence that produced the outcome, and publish their findings.
The second is the flight recorder. Not a conclusion — a record. What was actually said and done, minute by minute, by people who did not yet know any of it mattered.
Those are not the same thing, and only one of them lets you sit in the seat.
The published account of any failed initiative is the accident report. Somebody assembled it after the outcome was known, from what could be recovered, and named a cause. That is what forms most people’s view of Thomas Cook, or Santa Clara, or any of the others.
Think of the Procurement Insights archive as the flight recorder.
Internal documents sent to me while an implementation was still running. An award submission written by the people doing the work, while they were still doing it. A recorded interview a County Executive agreed to give when the questions were uncomfortable. A transformation deck handed over in 2010 by the firm that produced it.
None of that was written to explain a failure. Most of it was written by people who expected to succeed.
That is the property that matters, and it is not the same as the material simply being old. It was created before anyone knew what it would turn out to be evidence of — which means none of it was selected, framed or emphasised to support a conclusion that did not yet exist.
Why the recorder is credible
A flight recorder is not trusted because the investigator is trustworthy. It is trusted because it was running at the time and nobody edited it afterward.
That is the entire basis on which the archive works, and it is why the rule has never changed here: published once, never edited. Where a post has been updated, the update is appended above the original with its own date and the original left intact. The timestamps are in the platform metadata. Anyone can check them.
Which means the numbers in the archive do not always agree with each other. A figure reported in 2008 may differ from the same figure reported in 2016. That is not a defect — it is the proof. Across nineteen years of contemporaneous reporting, a record that agreed with itself to the decimal on every retelling would warrant more suspicion, not less, than one that preserves its seams.
One honest limit, because the analogy would otherwise flatter me. A real flight recorder captures every channel for the whole flight. Mine does not. It was running on some channels and not others, kept by someone who was present for parts and absent for others. There are gaps.
What it holds is contemporaneous, dated, and unedited. That is the property that matters, and it is rarer than completeness.
The debris field
Here is what most discussion of case teaching misses.
An investigation does not begin with a narrative. It begins with wreckage scattered across a search area, and most of it is missing. The pieces recovered are recovered because somebody went looking. The report at the end is assembled from what was found — never from everything.
And a piece of wreckage does not arrive labelled.
Nobody hands you the fragment and says this one matters. You have to recognize it among the pieces that do not, while you are still standing in the field.
That is the actual skill. Not analyzing evidence you have been told is significant — recognizing significance in material that arrives undifferentiated.
Santa Clara turns on a decision made in 2003, described in eight words in a background section of a committee document, with no reasons given. Nothing marks it. Everything downstream follows from it.
Why knowing the ending does not spoil it
Most of these outcomes are public. Anyone can look up what happened to Thomas Cook, or read the Civil Grand Jury report adopted in June 2024 on Santa Clara County’s procurement.
So people arrive knowing the ending, and usually with a view already formed about why.
That view survives contact with the published account — because the published account is what formed it.
It does not survive contact with the recorder.
Knowing a company collapsed tells you nothing about which document mattered, or which eight-word line carried the whole thing, or which of six competing signals was the one to act on in the third month. Knowing that a plane went down tells you nothing about which piece in the debris field explains why.
Investigators work backward from the outcome to establish cause. Then simulators put pilots back at an earlier point and run the conditions forward, so they can recognize the pattern developing while there is still time to change it.
Trace backward to establish causality. Move forward to develop recognition.
The pilot in the simulator often knows which flight it is. That is not the exercise. The exercise is whether they can see the moment it was still preventable, and what they would have done differently while standing inside the uncertainty rather than looking back at it.
What you are in the cockpit to do
The labs place you in the seat and play the record forward, decision by decision. At each stop, three questions:
What would you do in this situation? Why would you do it that way? How would it have affected the outcome?
You commit before the next piece arrives. Then the boundary advances and the record shows what the organization actually did and what followed.
Your reasoning is then mapped against the strands the record shows were determinative, along with one, two or three plausible alternative paths that could have followed from a different decision.
Every one of those alternatives is treated as a projection. What happened is documented. What might have happened is not, and never will be. Where the analysis produces competing scenarios, the disagreement is shown rather than resolved — because a single confident answer about a path nobody took is exactly the shape of evidence this practice exists to distrust.
And where the record is silent, it stays silent. Nobody knows why Santa Clara declined the procurement module in 2003. That silence is part of the case, not a gap to be filled.
What does not transfer
Here is the thing that has to be said plainly, because it is the objection the format invites.
Doing the same steps in your organization will not produce the same outcome.
I wrote about this in December, quoting Bob Sievert — who ran one of the cases in the set:
You do not leave with Virginia’s answer, or Santa Clara’s. Those belong to Virginia and Santa Clara. They were produced by those organizations’ people, constraints, incentives and absorptive capacity, none of which you have.
What transfers is the capacity to recognize which strand is load-bearing in a situation you have not seen before.
And that can only be built on somebody else’s situation — because in your own you are inside it, invested in it, and already holding a view. Someone else’s debris field is where you learn to read one.
The purpose is not to copy a championship. It is to develop a new and expanded way of thinking critically about the one you are actually in.
What this is for
The industry produces a great deal of analysis about why initiatives fail. Most of it is sound, and I have no argument with the conclusions. Analyst research tells you the shape of the distribution, which matters — you cannot learn how common a pattern is from one case.
But you cannot practise on a percentage.
The purpose is not to explain yesterday’s failure. It is to recognize tomorrow’s before it happens — and recognition is a skill you can only build by sitting in the seat before the outcome is known, with the evidence arriving in the order it actually arrived, and committing to a call before the next piece lands.
That is what the flight recorder is for. And it is why the question I opened with is not rhetorical.
How many people still have that deck?
The strand mapping and scenario work described above are performed using the Hansen Strand Commonality™ Model and the ARA™ RAM 2025™ multimodel platform.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
Hansen Models™ · Procurement Insights
-30-
Share this:
Like this:
Related