If you cannot describe it, do not automate it

Mark Ogne of Forrester argues that organisations chasing ambitious AI use cases have already taken a risk decision without noticing — the pressure to show value is running ahead of the readiness to support it.

The principle he offers is compact: the higher the consequence of the use case, the stronger the foundation needed beneath it. The foundations he names are shared organisational knowledge, documented and consistent workflows, defined accountability, continuous governance and rigorous measurement.

The diagnostic

The warning sign Ogne calls clearest is being unable to state how an important decision is currently made. If the answer is I'm not sure or it depends, automating or augmenting that decision is a risky bet — because there is no description of the current behaviour to compare the new behaviour against.

That is the part worth keeping. A process nobody can articulate cannot be evaluated after automation either: without a stated rule, there is no way to tell whether the system is reproducing the judgement or quietly replacing it with something else.

Read it as a framework rather than a finding. The piece carries no survey data or statistics; it rests on the author's conversations with marketing leaders, and Forrester presents it as such.