What I found was not nostalgia. It was this week’s argument, written eighteen years ago by someone who did not know he was making it.
In 2008 I wrote a fifty-page white paper for the CATA Alliance called SAP Procurement for Public Sector. The question it asked was narrow and of its moment: was SAP’s new PPS module a genuine advance, or the same product re-branded?
I answered the question, published it, and moved on. I did not treat it as part of anything. There was no framework then, no named method, no thesis to defend. I was writing what I understood at the time.
This week I read it again, because I needed one figure out of it.
Six things in that document are still in circulation today. Two of them I published as new arguments within the last seven days.
1. The workarounds
Here is what I wrote in 2008 about what happens when a procurement department inherits a module that does not fit how it works:
The need to “shape” or “mold” a misaligned module to better align with an organization’s procurement practice increases the frequency of having to create “work arounds” or band-aid fixes.
And the consequence: a substantial and incremental cost in time and resources, starting at implementation and continuing indefinitely — followed by varying degrees of cynicism at the operational levels of an organization.
Two days ago I published a post arguing that the gap agentic AI cannot close is the accumulated set of workarounds that exist because the formal design does not fit operating reality.
The 2008 version says the same thing, and is more specific about the cost.
2. Where the decision actually gets made
Most purchasing departments inherit their software as an adjunct downstream byproduct of either an original Finance (ERP) or IT-centric initiative. This has usually meant that their input has been relegated to the realm of the afterthought versus providing decisive and proactive input when it matters the most — prior to an actual decision being made.
Earlier this month a director at a large consumer goods company described their situation to me. Finance had chosen the platform. Procurement inherited it. They now sit on the IT side of the same relationship and can see both halves of the frustration.
They were describing 2026. The paragraph describing it is from 2008, and this week’s discussion reminded me of it.
3. The gate methodology, and why it does not reach far enough
The paper examined a structured review approach then in circulation for improving SAP implementations — five phases, five gates, each positioned at the exit of a major project stage. Scope and plan, design, build, test, go-live.
My assessment at the time:
While the Gate concept is certainly creative, it is still based upon the premise of working within the flawed confines of a traditional enterprise-based project.
Every gate sits inside the project. The first one opens after the scope has been set.
⭐ This week I read a major analyst firm’s model for accelerating AI partnerships. It has four sprints. Sprint 0 is shortlisting and onboarding.
Same structure, eighteen years later, different technology. The zero-point is still the moment a shortlist exists — which means everything that determines whether the shortlist is the right one has already happened, unexamined.
4. The line that became a method
Buried in the middle of the paper, referring back to an October 2007 post of mine on why some e-procurement initiatives succeed and others do not:
Success was directly linked to the proper alignment of technology with the way in which the organization operated in the real-world.
That sentence is the whole of what I now call Phase 0™, written before I had a name for it, in a document about a software module.
5. What made Virginia work — and it was not the platform
The paper documents the Commonwealth of Virginia’s eVA program at a point when it was still unusual to call it a success:
- Less than 1% of identified spend running through the system in 2001 — roughly $3.5 billion of the Commonwealth’s $5 billion total — rising to 80–90% by 2007
- Registered suppliers from 20,000 to 34,000
- Contract awards distributed to 23% of the supply base in 2001, more than 40% by 2007
But the numbers are not the finding. This is, from my interview with Bob Sievert, Director of the eProcurement Bureau for the Commonwealth of Virginia, published 12 September 2007 — an excerpt he and the Virginia management team subsequently circulated widely inside the Commonwealth:
Bob’s statement that government is not just a “single business,” but is actually comprised of many different “lines of business” tweaked my interest. This was due to the fact that the majority of e-procurement initiatives are championed by senior level managers who recognize the potential of a technology-centric program but lack a firm understanding of operational challenges and therefore underestimate the impact of a proposed strategy at the department level.
The recognition on the part of Virginia that government goes beyond a mere org chart but is actually comprised of Higher Education, K-12, Corrections, Public Safety, Transportation, Health, Social Services and Construction etc. meant that they really understood the “special needs, special rules and special challenges” associated with the procurement practice of each entity both individually and collectively.
As a result, they avoided the trap of eVA becoming a software project as Bob put it, and were thereby able shift the emphasis from an exercise in cost justification, to one of process understanding and refinement. And while the Ariba application has done the job it was required to do, eVA’s effectiveness has little to do with the technology and more to do with the methodology the Virginia brain trust employed. It is when technology (nee software) is seen as the primary vehicle to drive results that it becomes ineffectual and mostly irrelevant. The 75 to 85% e-procurement initiative failure rate gives testimony to this fact.
⭐ Read that bolded sentence again, and then read the quote I published in 2016 — technology becomes ineffective the more dependent we become on it. Same claim, eight years earlier, with the mechanism attached: it is not dependence in the abstract, it is dependence on technology as the primary vehicle to drive results.
And the paper records something else, which I would ask any vendor to sit with. Virginia’s senior management believed they would have been equally successful with any other vendor’s product.
6. What the case references were actually worth
The paper reproduces the reference accounts available at the time — Arapahoe County, Seattle Public Schools, Erie County, San Luis Obispo County, the City of Ottawa — alongside the failures: Hershey at $112 million, FoxMeyer’s $500 million lawsuit, Hewlett-Packard’s $400 million revenue loss, Cadbury’s £12 million hit, Whirlpool, Dow Chemical, Boeing, Dell, Waste Management.
And it asked the question that made the exercise worth doing:
The key question surrounding the above referenced results is whether or not they are a fair and reliable indicator of success.
⭐ Within weeks of that paper, one of the reference accounts I had been examining collapsed — a multi-year public sector program, roughly $38 million, ending in a political crisis. It had been a reference account in December. It was a catastrophe in January. Nothing changed over Christmas except that somebody outside the reporting chain picked up a phone.
Today’s case reference could be tomorrow’s embarrassment. I wrote that two years later in a follow-up post.
What the document actually is
Being early and being adopted are different events, and the record only proves the first one.
I was not building an argument in 2008. No framework to protect, no method to sell, no thesis that needed the evidence to come out a particular way. I answered a question about a software module and moved on to the next thing.
⭐ That is what makes it usable now. A record assembled to prove something is worth very little, because everyone understands the writer chose what went into it. A document written in 2008 could not have been selected to substantiate a theory that did not yet exist.
It is also why I could not have retrofitted it. The 2008 paper contains an argument I did not know I was making, in vocabulary I no longer use, about a product that no longer matters, with cases I have not cited in years.
And it says the same thing I published this week.
The uncomfortable part
Eighteen years. Four technology eras. Same finding.
That is either evidence of an invariant, or evidence that I have been repeating myself since 2008 without noticing.
I would rather state both possibilities than pretend only one is available. Here is the test that separates them, and it is the one I would put to anybody making a claim like this:
If the underlying condition were not constant, the technology would have moved it by now. Four generations of platform have been installed against this problem — ERP, SaaS, analytics, and now agentic AI. If the failure rate collapses in the agentic era without organizations changing anything about readiness, I am wrong, and I will publish that.
It has not collapsed yet. And multiple industry reports suggest it isn’t about to — I have covered several of them here recently.
-30-
This analysis draws on the Procurement Insights archive — an independent record, carrying zero vendor sponsorships, that I have published openly since 2007 and that consolidates documented client work, lectures, and articles reaching back to 1998 — nearly three decades of contemporaneous observation, gathered in one place rather than created there. Every claim in it is held to the Provenance Ledger™: a verify-before-publish discipline that traces each assertion to a primary source and never quietly edits the record once it is posted. That record is the evidence base for two working lenses — Invariant Physics™, the constant that however far the technology advances, the operating logic must be in place first; and Implementation Physics™, its per-engagement application: the discipline of doing the readiness work before the platform, not after. Getting it right, rather than being right.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
Related
I Wrote a White Paper About SAP in 2008. I Have Just Read It for the First Time in Years.
Posted on August 18, 2026
0
What I found was not nostalgia. It was this week’s argument, written eighteen years ago by someone who did not know he was making it.
In 2008 I wrote a fifty-page white paper for the CATA Alliance called SAP Procurement for Public Sector. The question it asked was narrow and of its moment: was SAP’s new PPS module a genuine advance, or the same product re-branded?
I answered the question, published it, and moved on. I did not treat it as part of anything. There was no framework then, no named method, no thesis to defend. I was writing what I understood at the time.
This week I read it again, because I needed one figure out of it.
Six things in that document are still in circulation today. Two of them I published as new arguments within the last seven days.
1. The workarounds
Here is what I wrote in 2008 about what happens when a procurement department inherits a module that does not fit how it works:
And the consequence: a substantial and incremental cost in time and resources, starting at implementation and continuing indefinitely — followed by varying degrees of cynicism at the operational levels of an organization.
Two days ago I published a post arguing that the gap agentic AI cannot close is the accumulated set of workarounds that exist because the formal design does not fit operating reality.
The 2008 version says the same thing, and is more specific about the cost.
2. Where the decision actually gets made
Earlier this month a director at a large consumer goods company described their situation to me. Finance had chosen the platform. Procurement inherited it. They now sit on the IT side of the same relationship and can see both halves of the frustration.
They were describing 2026. The paragraph describing it is from 2008, and this week’s discussion reminded me of it.
3. The gate methodology, and why it does not reach far enough
The paper examined a structured review approach then in circulation for improving SAP implementations — five phases, five gates, each positioned at the exit of a major project stage. Scope and plan, design, build, test, go-live.
My assessment at the time:
Every gate sits inside the project. The first one opens after the scope has been set.
⭐ This week I read a major analyst firm’s model for accelerating AI partnerships. It has four sprints. Sprint 0 is shortlisting and onboarding.
Same structure, eighteen years later, different technology. The zero-point is still the moment a shortlist exists — which means everything that determines whether the shortlist is the right one has already happened, unexamined.
4. The line that became a method
Buried in the middle of the paper, referring back to an October 2007 post of mine on why some e-procurement initiatives succeed and others do not:
That sentence is the whole of what I now call Phase 0™, written before I had a name for it, in a document about a software module.
5. What made Virginia work — and it was not the platform
The paper documents the Commonwealth of Virginia’s eVA program at a point when it was still unusual to call it a success:
But the numbers are not the finding. This is, from my interview with Bob Sievert, Director of the eProcurement Bureau for the Commonwealth of Virginia, published 12 September 2007 — an excerpt he and the Virginia management team subsequently circulated widely inside the Commonwealth:
⭐ Read that bolded sentence again, and then read the quote I published in 2016 — technology becomes ineffective the more dependent we become on it. Same claim, eight years earlier, with the mechanism attached: it is not dependence in the abstract, it is dependence on technology as the primary vehicle to drive results.
And the paper records something else, which I would ask any vendor to sit with. Virginia’s senior management believed they would have been equally successful with any other vendor’s product.
6. What the case references were actually worth
The paper reproduces the reference accounts available at the time — Arapahoe County, Seattle Public Schools, Erie County, San Luis Obispo County, the City of Ottawa — alongside the failures: Hershey at $112 million, FoxMeyer’s $500 million lawsuit, Hewlett-Packard’s $400 million revenue loss, Cadbury’s £12 million hit, Whirlpool, Dow Chemical, Boeing, Dell, Waste Management.
And it asked the question that made the exercise worth doing:
⭐ Within weeks of that paper, one of the reference accounts I had been examining collapsed — a multi-year public sector program, roughly $38 million, ending in a political crisis. It had been a reference account in December. It was a catastrophe in January. Nothing changed over Christmas except that somebody outside the reporting chain picked up a phone.
Today’s case reference could be tomorrow’s embarrassment. I wrote that two years later in a follow-up post.
What the document actually is
Being early and being adopted are different events, and the record only proves the first one.
I was not building an argument in 2008. No framework to protect, no method to sell, no thesis that needed the evidence to come out a particular way. I answered a question about a software module and moved on to the next thing.
⭐ That is what makes it usable now. A record assembled to prove something is worth very little, because everyone understands the writer chose what went into it. A document written in 2008 could not have been selected to substantiate a theory that did not yet exist.
It is also why I could not have retrofitted it. The 2008 paper contains an argument I did not know I was making, in vocabulary I no longer use, about a product that no longer matters, with cases I have not cited in years.
And it says the same thing I published this week.
The uncomfortable part
Eighteen years. Four technology eras. Same finding.
That is either evidence of an invariant, or evidence that I have been repeating myself since 2008 without noticing.
I would rather state both possibilities than pretend only one is available. Here is the test that separates them, and it is the one I would put to anybody making a claim like this:
If the underlying condition were not constant, the technology would have moved it by now. Four generations of platform have been installed against this problem — ERP, SaaS, analytics, and now agentic AI. If the failure rate collapses in the agentic era without organizations changing anything about readiness, I am wrong, and I will publish that.
It has not collapsed yet. And multiple industry reports suggest it isn’t about to — I have covered several of them here recently.
-30-
This analysis draws on the Procurement Insights archive — an independent record, carrying zero vendor sponsorships, that I have published openly since 2007 and that consolidates documented client work, lectures, and articles reaching back to 1998 — nearly three decades of contemporaneous observation, gathered in one place rather than created there. Every claim in it is held to the Provenance Ledger™: a verify-before-publish discipline that traces each assertion to a primary source and never quietly edits the record once it is posted. That record is the evidence base for two working lenses — Invariant Physics™, the constant that however far the technology advances, the operating logic must be in place first; and Implementation Physics™, its per-engagement application: the discipline of doing the readiness work before the platform, not after. Getting it right, rather than being right.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
Share this:
Like this:
Related