No one read it in 2004. Would you read this paper in 2026?
In 2004 I wrote a paper called Technology’s Diminishing Role in an Emerging Process-Driven World. I had no blog, no platform and no distribution channel, so I sent it out by email, manually, to people who had every reason to care about the argument.
Nothing came back. Not disagreement. Not a rebuttal. Silence.
It is like sending out invitations to a birthday party and having nobody RSVP.
Two years later Summit Magazine printed it, in their September 2006 issue, and it entered the record properly. So the sequence is: written and circulated by email in 2004, published by Summit in September 2006, and quoted on this blog in 2008 — which is the version that survived.
But the 2004 circulation is the part worth talking about, because of what happened afterwards — which is to say, nothing happened, for a long time, and then the pattern the paper described kept reappearing.
I don’t want you to mistake that last sentence for some kind of psychic ability. I didn’t have any. The power you are seeing comes from the Procurement Insights contemporaneous archive. I could have just as easily been wrong, and the archive would have recorded that too — because every post carries a published stamp and a modified stamp, and anything that moves is logged. We don’t erase tapes.
What the paper actually said
I quoted it in a 2008 post on this blog, which is how the text survived. Here is the passage that matters, written in 2006 and reflecting the market as it stood between 2001 and 2005:
Studies indicate that between 75 and 85 percent of all e-procurement initiatives in the period failed to deliver the expected results.
The editor added a note putting large-scale ERP programs at 90 to 95 percent.
The argument around those figures was that as enterprise technology capability advanced, the variable determining whether an investment produced operational value was moving — away from the technology itself and toward the process integrity and operating conditions surrounding the deployment.
That claim went into a market buying ERP optimism and accelerating SaaS enthusiasm. Among the people I sent it to, nobody argued about whether the constraint had moved. Nobody replied at all.
The same number, twenty years later
Last week I published a set of figures from four advisory firms, gathered across five technology generations.
Gartner reported a 50 percent CRM failure rate in the early 2000s. Sixty percent for big data in 2015, revised upward afterward. BCG found a 35 percent success rate across more than 850 companies on digital transformation. Gartner projects 80 percent of data governance initiatives will fail by 2027. And for AI in 2026: 22 percent scaled, 39 percent reporting measurable impact, 5 percent generating significant value.
Every one of those sits inside the band the 2006 text reported for e-procurement.
Those figures are not commensurable — each firm counts something different, and the band is the finding rather than the series. But the range has not moved in twenty years, across technologies that have almost nothing in common with each other.
What being ignored actually demonstrates
I want to be careful here, because there is a version of this post that is about having been right, and that version is not worth writing. Being right about a failure rate is not an achievement. The rate was visible to anyone who looked, and several people with better instruments than mine had already documented the underlying pattern.
The useful part is the silence.
An argument that the constraint had moved out of the technology layer was, in 2004, unanswerable in a specific way. It did not tell anyone what to buy. It did not create a category, a budget line or a project. It offered a diagnosis in a market that was purchasing solutions, and a diagnosis with nothing to sell attached to it has no natural recipient.
Technology problems are easy for organizations to recognize because they have products, owners and budgets. Operating-condition problems often have none of the three.
Nicholas Carr published IT Doesn’t Matter in the Harvard Business Review the year before, and got a very different reception — Ballmer called it hogwash, vendors lined up to respond, the magazine ran seventeen pages of letters. Carr had a platform and a title that provoked. The argument still changed almost nothing.
So there are two ways for an accurate correction to fail. It can be argued with loudly and left where it stands. Or it can produce no response at all. Neither outcome depends on whether the correction was right.
What the twenty years added
The paper made a claim it could not test. There was no way in 2004 to know whether the structural argument would hold, because the conditions that would test it had not arrived.
They have now. Across two decades of successive technology waves, capability has advanced dramatically while the reported non-success band has stayed stubbornly familiar. That is no longer a prediction. It is a documented pattern, and the documentation exists because the observations were made as the waves happened rather than assembled afterward.
Which is the only reason this post is possible. The paper survived because I quoted it in 2008 in a piece about something else entirely — a vendor profile, written for a different purpose, that happened to carry the passage forward. I found it by accident last week, looking for something unrelated.
A record kept only for the parts you expect to need is not a record. It is a summary of what you already believed.
The part that has changed
One thing is different now, and it is the reason the argument may finally land where it did not in 2004.
Back then the consequences of an unvalidated commitment arrived years later, usually after the person who made it had moved on. The gap absorbed the failure. Nobody had to reckon with it inside their own tenure.
That gap has closed. AI did not eliminate the old implementation problem. It compressed the time between commitment and consequence — the first consequences can now arrive in months, often inside the tenure and budget cycle of the executive who authorized the spend.
The failure rate has not changed. What changed is who is still in the chair when it lands.
That is a considerably better reason to engage the argument than being told it was made twenty years ago.
Coda
I am not going to claim the silence was unjust. A paper emailed manually by someone with no platform, into a market with no reason to listen, is not a scandal. It is what happens.
But it is worth knowing that the argument was available, in 2004, to anyone who wanted it — and that the twenty years since have produced a great deal of evidence consistent with it, and little reason to abandon the underlying proposition.
The invitations went out. Nobody came. The party happened anyway.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
Related
Nobody RSVP’d
Posted on September 9, 2026
0
No one read it in 2004. Would you read this paper in 2026?
In 2004 I wrote a paper called Technology’s Diminishing Role in an Emerging Process-Driven World. I had no blog, no platform and no distribution channel, so I sent it out by email, manually, to people who had every reason to care about the argument.
Nothing came back. Not disagreement. Not a rebuttal. Silence.
It is like sending out invitations to a birthday party and having nobody RSVP.
Two years later Summit Magazine printed it, in their September 2006 issue, and it entered the record properly. So the sequence is: written and circulated by email in 2004, published by Summit in September 2006, and quoted on this blog in 2008 — which is the version that survived.
But the 2004 circulation is the part worth talking about, because of what happened afterwards — which is to say, nothing happened, for a long time, and then the pattern the paper described kept reappearing.
I don’t want you to mistake that last sentence for some kind of psychic ability. I didn’t have any. The power you are seeing comes from the Procurement Insights contemporaneous archive. I could have just as easily been wrong, and the archive would have recorded that too — because every post carries a published stamp and a modified stamp, and anything that moves is logged. We don’t erase tapes.
What the paper actually said
I quoted it in a 2008 post on this blog, which is how the text survived. Here is the passage that matters, written in 2006 and reflecting the market as it stood between 2001 and 2005:
The editor added a note putting large-scale ERP programs at 90 to 95 percent.
The argument around those figures was that as enterprise technology capability advanced, the variable determining whether an investment produced operational value was moving — away from the technology itself and toward the process integrity and operating conditions surrounding the deployment.
That claim went into a market buying ERP optimism and accelerating SaaS enthusiasm. Among the people I sent it to, nobody argued about whether the constraint had moved. Nobody replied at all.
The same number, twenty years later
Last week I published a set of figures from four advisory firms, gathered across five technology generations.
Gartner reported a 50 percent CRM failure rate in the early 2000s. Sixty percent for big data in 2015, revised upward afterward. BCG found a 35 percent success rate across more than 850 companies on digital transformation. Gartner projects 80 percent of data governance initiatives will fail by 2027. And for AI in 2026: 22 percent scaled, 39 percent reporting measurable impact, 5 percent generating significant value.
Every one of those sits inside the band the 2006 text reported for e-procurement.
Those figures are not commensurable — each firm counts something different, and the band is the finding rather than the series. But the range has not moved in twenty years, across technologies that have almost nothing in common with each other.
What being ignored actually demonstrates
I want to be careful here, because there is a version of this post that is about having been right, and that version is not worth writing. Being right about a failure rate is not an achievement. The rate was visible to anyone who looked, and several people with better instruments than mine had already documented the underlying pattern.
The useful part is the silence.
An argument that the constraint had moved out of the technology layer was, in 2004, unanswerable in a specific way. It did not tell anyone what to buy. It did not create a category, a budget line or a project. It offered a diagnosis in a market that was purchasing solutions, and a diagnosis with nothing to sell attached to it has no natural recipient.
Technology problems are easy for organizations to recognize because they have products, owners and budgets. Operating-condition problems often have none of the three.
Nicholas Carr published IT Doesn’t Matter in the Harvard Business Review the year before, and got a very different reception — Ballmer called it hogwash, vendors lined up to respond, the magazine ran seventeen pages of letters. Carr had a platform and a title that provoked. The argument still changed almost nothing.
So there are two ways for an accurate correction to fail. It can be argued with loudly and left where it stands. Or it can produce no response at all. Neither outcome depends on whether the correction was right.
What the twenty years added
The paper made a claim it could not test. There was no way in 2004 to know whether the structural argument would hold, because the conditions that would test it had not arrived.
They have now. Across two decades of successive technology waves, capability has advanced dramatically while the reported non-success band has stayed stubbornly familiar. That is no longer a prediction. It is a documented pattern, and the documentation exists because the observations were made as the waves happened rather than assembled afterward.
Which is the only reason this post is possible. The paper survived because I quoted it in 2008 in a piece about something else entirely — a vendor profile, written for a different purpose, that happened to carry the passage forward. I found it by accident last week, looking for something unrelated.
A record kept only for the parts you expect to need is not a record. It is a summary of what you already believed.
The part that has changed
One thing is different now, and it is the reason the argument may finally land where it did not in 2004.
Back then the consequences of an unvalidated commitment arrived years later, usually after the person who made it had moved on. The gap absorbed the failure. Nobody had to reckon with it inside their own tenure.
That gap has closed. AI did not eliminate the old implementation problem. It compressed the time between commitment and consequence — the first consequences can now arrive in months, often inside the tenure and budget cycle of the executive who authorized the spend.
The failure rate has not changed. What changed is who is still in the chair when it lands.
That is a considerably better reason to engage the argument than being told it was made twenty years ago.
Coda
I am not going to claim the silence was unjust. A paper emailed manually by someone with no platform, into a market with no reason to listen, is not a scandal. It is what happens.
But it is worth knowing that the argument was available, in 2004, to anyone who wanted it — and that the twenty years since have produced a great deal of evidence consistent with it, and little reason to abandon the underlying proposition.
The invitations went out. Nobody came. The party happened anyway.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
Share this:
Like this:
Related