Not every decision needs an independent review. Some of them badly do.

The distinction is whether the decision is reversible, and whether the people advising you are compensated by the outcome.

Why the source of the advice matters

An assessment written by the firm that would implement it is a sales document. That is not a claim about anyone's integrity — it is what happens when the assessment and the implementation revenue sit in the same place.

The market splits accordingly. Large audit practices are built for enterprise-scale engagements and priced for them. Migration vendors offer an assessment that is the opening move of an implementation sale. Neither reliably gives a mid-market IT leader an independent answer.

Decisions worth testing first

The common thread is cost of reversal. Where unwinding the decision would take budget cycles rather than weeks, testing it beforehand is cheap by comparison.

Multi-year platform commitments, particularly where the framing of the problem came from the vendor proposing the platform.
Cloud migration scope — which workloads should move, as distinct from which ones can.
Migration timing that is currently being set by a licence renewal date rather than by architectural readiness.
Automation investment, before a platform is bought, because the sequencing question determines whether the licence gets used.
Compliance and disaster-recovery posture, where requirements would otherwise be retrofitted onto a design that never accounted for them.

Decisions that do not need one

Plenty do not, and saying so is part of scoping honestly.

Like-for-like refreshes where the architecture is not in question and the decision is procurement.
Decisions already made, where the review is being commissioned to ratify them. That is worth establishing before anyone writes a report nobody intends to act on.
Problems that are genuinely operational rather than architectural — an unmonitored estate needs coverage, not a roadmap.
Requirements where the binding constraint is a capability gap in the team, in which case recruitment answers it more directly than an assessment would.

Scope it to a question, not an estate

A single decision costs less to assess than an entire environment, and it produces a sharper answer. An estate-wide review yields an inventory; a scoped question yields a position you can act on.

The questions that establish whether an engagement is worth running at all are usually these four, asked in the first hour:

What decision is this assessment supposed to unblock, and what happens if you defer it another quarter?
What has already been ruled out, and by whom?
Which constraint is real — budget, compliance, a renewal date, or an internal position someone has already taken?
If we tell you to do nothing, is that an acceptable outcome here?

The last one matters more than it sounds. If a “do nothing” finding would be unwelcome, the engagement is ratification rather than assessment, and it is better to establish that before the work starts than in the debrief.

What a review should leave you holding

Findings against your actual environment, not a maturity-model template
A prioritized roadmap with sequencing and dependencies
A written position on the specific decision you were trying to make
Full ownership of the deliverable, including the right to tender it competitively

That last point is the practical test of independence. If you cannot take the deliverable to a competitor to price the execution, the assessment was not independent of the execution.

Where to go next.

Consultancy

Test the decision before you commit the budget.

A scoping call defines the decision and what the review needs to cover. If it does not need reviewing, you will hear that.

Book an architecture scoping call →Email us