Decision records and tradeoffs: choosing between options with reasons

By the end, you can

• Explain what an options appraisal is, and why weighing a decision's tradeoffs means comparing named options against each other rather than judging one option alone (PM-4).
• Use AI to draft a first-pass options appraisal for a project decision, and explain why the named decision-maker, not the draft, makes the actual call (PM-2, PM-4).
• Write a decision record for a sample project decision that states the options considered, the choice made, who made it and why (PM-4).

Before you start

This is Module 3, Lesson 2 of the AI-Powered Project Management course. It builds on Lesson 1's risk register — some of the decisions this lesson covers exist precisely because a risk on that register turned into something a project has to actually choose how to handle. It assumes you have completed Lesson 1 and have used at least one AI chat assistant for drafting text.

Weighing options is not the same as picking one that sounds right

A project decision rarely has one obviously correct answer sitting in isolation — it has options, each with real tradeoffs, and picking between them is a comparison, not a verdict on a single choice. APM's project management glossary defines options appraisal as "a systematic process of evaluating different choices to determine the most effective solution to a problem," adding that it involves "identifying options, assessing their feasibility, and comparing their potential impacts." The word worth noticing is "comparing" — an options appraisal is not a case for or against one option, it is a side-by-side look at more than one, so the tradeoffs between them are visible before anyone commits.

Some decisions in project work have their own name because they come up so often. APM's glossary defines a make or buy decision as "the decision to make a deliverable internally or to buy a finished deliverable from a supplier; for example, develop a software application in-house or purchase an existing application" — a decision with two named options and a real tradeoff between them: cost and control on one side, speed and lower up-front effort on the other. The glossary also defines an investment decision as "the decision made by the sponsor and governance board that justifies the investment in a project, programme or portfolio," noting that such decisions "rely on robust investment appraisal" — naming, again, that a real decision rests on a comparison, not a single option examined alone.

AI can draft the appraisal; the decision-maker makes the call

Given a decision a project faces — build or buy, which vendor, which of two dates — an AI assistant can draft a genuinely useful first-pass options appraisal: a list of the realistic options, a plausible set of pros and cons for each, and a rough comparison. That draft can save real time — and because the person asking often already privately favours one option, part of the decision-maker's check is reading the draft for quiet favouritism, rather than assuming the comparison is even-handed just because a machine produced it.

It is not the same as making the decision. The Open Worldwide Application Security Project's (OWASP) generative-AI security guidance on misinformation names the risk directly: "overreliance occurs when users place excessive trust in LLM-generated content, failing to verify its accuracy." An AI-drafted appraisal can only weigh what it was told; it has no access to a vendor's actual reputation with your organisation, a genuine budget ceiling nobody wrote down, or a political reality about which option a key stakeholder will actually accept. The US National Institute of Standards and Technology's (NIST) AI Risk Management Framework's description of an AI system as something that "can, for a given set of objectives, generate outputs such as predictions, recommendations, or decisions" is exactly why a drafted appraisal is a recommendation, not a decision — nothing in that description includes the system being accountable for what it recommends.

Making the actual call — committing the project to one option over another — is itself the kind of high-impact action OWASP's guidance on excessive agency has in mind when it recommends that a human "approve high-impact actions before they are taken." A named decision-maker, not the appraisal that compared the options, is who takes that step, and who answers for it afterward.

The decision record: writing down what was chosen, and why

An options appraisal compares choices. A decision record is what a project writes down once one of those choices has actually been made — the options that were considered, the option chosen, who made the choice, and the reason. Without it, a decision survives only in the memory of whoever was in the room, and six weeks later nobody can say with confidence why the cheaper vendor was passed over.

A decision record does not need to be long. For a make-or-buy decision, for example: "Decision: buy the off-the-shelf scheduling software rather than build one in-house. Options considered: build in-house (12 weeks, full control, ties up the only two developers on staff), buy [vendor] (live within a week, £400/month, does not support one minor workflow we currently use). Chosen by: the operations lead, with the sponsor's sign-off. Reason: the two-developer team cannot be spared for 12 weeks without delaying two other committed projects; the missing workflow can be handled manually for now." Anyone reading that record afterward, whether or not they were in the meeting, can see what was chosen and why — which is the entire point of writing it down.

A worked example: a community health clinic's scheduling system

A small community health clinic's operations lead is choosing between two patient-scheduling software options: build a simple in-house tool using an existing volunteer developer, or buy a low-cost off-the-shelf package. She asks an AI assistant to draft an options appraisal from a description of both.

The assistant lays out a reasonable comparison: building in-house offers full control over the clinic's specific intake questions but depends on one volunteer developer's limited spare time; buying the off-the-shelf package is faster to launch and comes with support, but its intake form can't be customised to match the clinic's required consent questions.

The operations lead checks the draft against what she actually knows. She confirms directly with the volunteer developer that he can only commit four hours a week — the AI's appraisal treated "depends on volunteer time" as a soft risk, but she now rates it as a likely delay, not a possibility. She also knows something the assistant had no way to: the off-the-shelf vendor has, in a call she already had, agreed informally to add a custom consent-question field within a month at no extra cost, closing the gap the appraisal flagged as a real limitation. She writes a decision record: "Decision: buy the off-the-shelf package. Options considered: in-house build (full control, but volunteer capacity makes the timeline unreliable), off-the-shelf (faster, missing one consent field, but vendor has agreed to add it within a month). Chosen by: operations lead, with clinic director's sign-off. Reason: reliable launch timeline matters more than full customisation for this clinic's current patient volume, and the one real gap has a committed fix." She files it alongside Module 1's project brief.

Accessibility notes

This lesson is text-first, with no images, audio, video or downloadable artifacts. The practice exercise's model answer sits behind a native disclosure control that is reachable and operable by keyboard and correctly announced by screen readers. The knowledge check uses native radio-button inputs with a visible question and options, and posts its result to a live status region so assistive technology announces the outcome without a page reload.

Sources and limits

This lesson synthesises the sources below into a practical learning model. It is not a security standard, legal advice or a guarantee that any particular agent design is safe.

1. Project management glossary — APM (Association for Project Management). Defines options appraisal as a systematic process of evaluating different choices to determine the most effective solution to a problem, involving identifying options, assessing their feasibility and comparing their potential impacts.
2. Project management glossary — APM (Association for Project Management). Defines a make or buy decision as the decision to make a deliverable internally or to buy a finished deliverable from a supplier, and an investment decision as the decision made by the sponsor and governance board that justifies the investment in a project, programme or portfolio.
3. AI Risk Management Framework 1.0 — NIST AI Resource Center. Frames an AI system as an engineered system that generates outputs such as predictions, recommendations or decisions — not a self-directing decision-maker.
4. LLM09:2025 Misinformation — OWASP Gen AI Security Project. Defines overreliance as placing excessive trust in AI-generated content without verifying its accuracy.
5. LLM06:2025 Excessive Agency — OWASP Gen AI Security Project. Recommends human-in-the-loop control requiring a person to approve high-impact actions before an LLM-connected system takes them.