What makes a business decision defensible?

A contemporaneous record of the process: a document written at the moment of deciding that shows what was decided and by whom, among which options, on what evidence and of what quality, under what assumptions, and with how much declared certainty. It is the same standard used in corporate law: the business judgment rule asks whether a decision was informed and made in good faith with the information available at the time. Whether it worked out is outside that evaluation.

Should you document before or after deciding?

At the moment. A record reconstructed after the outcome is worth little: memory does not archive the original reasons, it rewrites them with what it already knows about how things turned out (hindsight bias), and any experienced reader notices. The contemporaneous record, even if imperfect, is the one with evidentiary and learning value.

Method

How to document a business decision so it holds up

By the Verdika team · Jul 19, 2026 · 6 min read

In short

A record written in real time, while you decide, can answer months later: what was decided and by whom, among which options, on what evidence and of what quality, under what assumptions, and with how much declared certainty. That is what defends a decision to a third party, regardless of how it turned out.

When someone questions a decision of yours (a client, a board, an investor, an auditor) they almost never argue the outcome in the abstract. They argue whether you had reason to decide what you decided with what you knew at the time. Memory is not enough to answer that. You need a document. Corporate law settled this question decades ago with the business judgment rule: it requires deciding on an informed basis, in good faith and without a conflict of interest¹. Whether it worked out does not enter the equation. Documenting well builds your defense before you need it.

The record gets written in the moment

This is the rule that invalidates the most decisions in practice. Memory does not archive the original reasons for a decision: it rewrites them with what it already knows about the outcome. Once you know how things turned out, the reasons line themselves up on their own (this is called hindsight bias), and any experienced reader can tell a contemporaneous record apart from a justification written after the fact. That is why keeping a decision journal is recommended: write down, at the moment of deciding, what you expect to happen and why, so you can compare it honestly later².

What it needs to contain

#	Element	What it answers to a third party
1	Decision, date and owner	What exactly was decided, when, and who is accountable for it.
2	Options considered	Real alternatives, compared against each other. Without them, the analysis is a conclusion with makeup on.
3	Evidence with source and weight	What data backs the choice: source, year, sample, and how much each one counted. Data with no source is an opinion in good typography.
4	Falsifiable assumptions	What the decision rests on, and what observation would change it.
5	Declared certainty	How much confidence there was, tied to the quality of the evidence. Enthusiasm does not count.
6	Outcome, logged afterward	How it turned out. Turns loose decisions into a calibration history.

Evidence needs a named source

"The market is growing" sounds like data, but nobody can verify it, and verifying is exactly what makes a record defensible. "The U.S. Census reports X% annual growth in the sector, 2019 to 2024" (a sample figure) works: anyone can check it. Every data point in the record answers three questions: where it came from, when it is from, and what its scope is (what population it measured, and how far it can be extrapolated). When the evidence is not enough, say so: "there is no sufficient source on this point; the assumption is ours." Declaring the limit strengthens the document; stretching it contaminates the whole thing.

Certainty gets declared, then checked against reality

A record that does not say how certain it was cannot be evaluated later: any outcome is compatible with it. Declaring "70% confidence" is a commitment, and that commitment is exactly what defends you: if the evidence was partial and you said so, a bad outcome confirms your honesty instead of contradicting it. Across several logged decisions, comparing declared certainty against real outcomes builds a calibration history (what your 70% is actually worth when you say 70%), which is the hardest asset to fake that exists in consulting. Regulation is pushing the same direction: the EU's AI regulation requires transparency and documentation when automated systems take part in decisions that affect people³. The trend favors decisions that show their reasoning. Asking for trust stops being enough.

A quick defensibility test: hand the record to someone who was not part of the decision and ask them to attack it. If they need to ask you something the document does not answer, that is the gap.

The three mistakes that ruin the defense

The first: documenting after you already know the outcome. The reconstruction shows: an experienced reader can tell when the reasons were lined up to fit how things turned out. The second: certainty with nothing behind it. Declaring 90% confidence on thin evidence does not convince anyone who checks the data; the gap between what was declared and what the evidence supports is exposed immediately. The third: the assumptions left unwritten. What you do not write down is exactly what gets asked afterward, and an improvised answer, with the outcome already on the table, convinces few. All three share the same root: they mistake persuasion for reasoning.

How Verdika handles it

Verdika delivers this record as the natural output of the analysis. The memo comes out with the options ranked, the evidence cited by source, year, sample and weight, and the assumptions and risks named one by one. Confidence has a mathematical ceiling tied to the quality of the real evidence: without enough evidence, it cannot exceed 40%, no matter how much you'd like to push it higher. Every memo logs its full traceability (what evidence produced what confidence, with which model, and on what date) and later asks how the decision turned out. That is how, decision by decision, a public, verifiable calibration history gets built.

Bring the next hard decision with a record from the start.

Tell it in your own words. The defensible document comes out on its own.

Analyze my decision

Frequently asked

What makes a decision defensible?

A contemporaneous record of the process: decision, options, evidence with source and weight, assumptions, declared certainty and owner. The standard is the same one used in corporate law: informed and made in good faith with what was known at the time¹. Whether it worked out is outside the equation.

What if the decision turned out badly?

A well-documented decision that turned out badly defends itself better than a lucky call with no record. The bad outcome is explained by the world's uncertainty: negligence is ruled out by the document itself. And that outcome, once logged, feeds your calibration.

Isn't this a lot of work for every decision?

Skip it for reversible, cheap decisions: the record is for the ones that put cash, people or reputation on the line. Most of the work is formatting, and that can be automated. The honesty of the assumptions cannot be delegated. That part stays yours.

References

Business judgment rule — Cornell Law School, Legal Information Institute: law.cornell.edu/wex/business_judgment_rule.
The decision journal as an antidote to hindsight bias — Farnam Street: fs.blog/decision-journal.
Regulation (EU) 2024/1689 (the EU AI Act), official text: eur-lex.europa.eu/eli/reg/2024/1689/oj.

You might also like: How to write a decision memo · The 409 startup failures study