Executive Clarity · The Platform

The Trust Question

By Andreas BORN

Published

The volume of plausible SAP content is now unlimited. Detection and labels do not tell an executive what to act on — curation and accountability do. A selection framework for transformation leaders, and how sapperment applies it to itself.

An SAP transformation decision is only as good as the information it stands on.

That has always been true. What changed is the supply side.

The volume of plausible content about SAP Cloud ERP, RISE, GROW, Clean Core and Business AI is now effectively unlimited. Some of it is excellent. Much of it is vendor messaging in an analyst costume. And a growing share is generated at near-zero cost by people who have never sat in a cutover weekend.

The instinctive response is detection: tell me what was machine-made, and I will know what to discard.

Medium — a platform whose entire business is trusted reading — publicly rejected that instinct this month. In Write for humans, not algorithms, the company explains why it has chosen not to offer AI detection: the tools are unreliable in both directions, they punish honest writers while sophisticated actors route around them, and above all they answer the wrong question. The reader’s real question, Medium argues, was never what produced this — it was can I trust this. Their answer is what they call the curation era: human editors, accountable publications, communities, and reader-controlled feeds. By their own account, the large majority of stories in a reader’s feed are there because a person recommended them.

The reader’s question is not forensic. It is fiduciary.

I think Medium is right — and I think the argument matters far more in enterprise software than in general-interest publishing. A weak essay costs a reader ten minutes. A weak input into an ERP platform decision costs years and a nine-figure budget.

What detection cannot tell an executive

The provenance of a text is a poor proxy for its reliability, in both directions.

Content produced entirely by humans has misled SAP customers for decades: sponsored analyst rankings, reference stories with the difficult phases removed, partner thought leadership whose conclusion was fixed before the research began. No detector flags any of it.

Conversely, the mode of production says nothing about whether the underlying experience is real. The question that separates usable advice from noise is older than any of today’s tools: does a named person with real exposure stand behind this claim, and can I check it?

The selection framework

What follows is judgment, built on the pattern above — my criteria, not a documented standard.

Before an external analysis, benchmark or recommendation enters your decision material, four tests:

1. Named accountability. Is there a person — not a brand — whose reputation is attached to the claim? Would that person lose something if it turned out to be wrong? Anonymous authority is not authority.

2. Disclosed interest. Who benefits if you follow the advice? An interest is not disqualifying — undisclosed interest is. The most dangerous input in a transformation programme is the one whose commercial motive you discover in month nine.

3. Separated fact and judgment. Does the piece distinguish what is documented from what is inferred and from what is opinion? Writing that blends the three is not necessarily dishonest — but it is unusable, because you cannot tell which parts survive contact with your own facts.

4. Checkable claims. Are material statements sourced, dated, and specific enough to verify? A claim that cannot be checked is a mood, not an input.

None of these tests asks how the text was produced. All of them ask whether someone stands behind it.

How sapperment applies this to itself

It would be cheap to propose a standard and exempt the platform proposing it. So, for the record: sapperment’s answer to the trust question is structural, and it is documented on Trust is the product. Authorship is by invitation, never by open call, with one accountable owner per topic. Every substantive piece separates fact, inference and point of view — you can see the seam in this article. Migrated publications carry their provenance. The library shows a verification status on every record, including the unflattering ones. And nobody pays to be published here, or is paid to publish.

Hold us to it. The framework above is only credible if it can be turned on its author.

Monday morning

Before the next steering committee, ask of the three most influential external inputs in your programme:

  • Whose name is attached — and what do they lose if it is wrong?
  • What commercial interest sits behind it, and is it disclosed?
  • Can you separate its facts from its opinions without guessing?
  • Which of its material claims has anyone on your team actually verified?

If an input fails all four, it is not information. It is atmosphere.

Sources