The AI Opportunity Audit: Choose One Idea Worth Testing Next

In this guide

1. The useful starting point
2. Start with friction, experience and recurring interest
3. Use an evidence table instead of a popularity contest
4. An illustrative example: revive the idea or launch another brand?
5. Run the audit in one sitting
6. Ask what AI actually changes
7. Keep the test smaller than the story
8. Make the decision visible
9. Questions you might be asking
10. Sources and further reading

The useful starting point

List your recurring problems and ideas, then compare evidence, audience access, relevant experience and delivery effort. Use AI to question assumptions and reduce the size of the first test. An idea’s longevity can signal sustained interest; it does not establish current demand.

When new tools make more projects possible, choosing becomes more important. You may have an old domain, a notebook full of ideas and several half-built experiments. The useful question is not whether AI can produce another website. It is which problem deserves your next focused block of effort.

An opportunity audit gives you a way to make that choice. It combines what you keep returning to with what people need now and what you can reasonably test. AI can help organize the comparison, but the evidence and the decision remain visible.

Start with friction, experience and recurring interest

Make three short lists. First, tasks that repeatedly frustrate you or people you know. Second, work you understand well enough to judge and deliver. Third, ideas that have survived several rounds of enthusiasm. Look for overlap without assuming that overlap guarantees a business.

Returning to an idea over years can suggest genuine interest, accumulated knowledge or a problem you still notice. It can also reflect an attractive possibility you have not tested. Respect the consistency while asking for fresh evidence. The people, alternatives and practical constraints may have changed.

Use an evidence table instead of a popularity contest

For each candidate, record the intended user, the recent problem observed, existing alternatives and the next person you can speak with. Add your relevant experience, the smallest useful deliverable and the main risk. Mark missing information clearly. A blank cell is more useful than an invented estimate.

Avoid giving every idea a precise numerical score if the inputs are guesses. A simple decision such as “ready for conversations,” “needs access to an audience” or “overlaps an existing project” may be more honest. Use the audit to choose a next action, not to manufacture certainty.

Review these ideas as possible small experiments: [ideas and evidence]. For each, separate observed facts, old evidence and assumptions. Identify the intended user, the current alternative, my relevant experience and the smallest test. Flag overlap with projects I already have. Recommend one next action based on evidence and available time. Do not invent demand, search traffic, domain value or likely revenue.

An illustrative example: revive the idea or launch another brand?

Imagine an old idea for helping professionals create a workshop. The founder still enjoys teaching, has relevant experience and now has tools that make drafting materials easier. There is also an existing education site where a small pilot could be offered.

The tempting response is to reactivate a separate domain and build a large course platform. The audit reveals a smaller question: will a reachable professional commit to a session that produces one useful workshop outline? A simple invitation and a clear deliverable can investigate that before a new brand is needed.

This example is illustrative. The history of the idea supports attention, while a current conversation and commitment would support demand. Improved production tools address delivery effort; they do not answer whether the offer matters to the buyer.

Run the audit in one sitting

1. Choose no more than five candidates. A complete inventory can wait; the immediate goal is a decision.

2. Write the real evidence beside each idea, including its date and source. Distinguish a recent request from an old note.

3. Identify one reachable audience for each. If access is missing, make that the next test rather than building first.

4. Ask AI to challenge the assumptions and suggest a smaller deliverable. Review its reasoning against the evidence you supplied.

5. Choose one experiment, a time limit and a question it must answer. Put the other ideas in a visible waiting list.

6. Write what would make you continue, revise or stop. Set the review point before doing the exciting production work.

Ask what AI actually changes

For the chosen idea, describe the specific capability that might help. It could organize approved research, draft a workbook, create a prototype or prepare variations for review. Compare that with the existing process and include checking, correction and setup in the effort estimate.

Research on AI productivity has produced different results across settings. That is a reason to measure your own workflow rather than importing a headline percentage. Faster generation may be valuable, but only if the resulting work is accurate and useful enough to deliver.

Keep the test smaller than the story

An exciting vision can contain several businesses at once: a directory, a course, a service, a community and software. Choose the piece that answers the most important uncertainty. If the uncertainty is buyer interest, a polished backend will not resolve it. If the uncertainty is delivery quality, a waitlist alone will not resolve it.

Record what the experiment cannot tell you. One useful pilot does not establish a large market. A conversation does not prove repeatability. A working prototype does not prove that support will be manageable. These limits help you choose the next test without dismissing the evidence you have gained.

Make the decision visible

Finish with a short note: “I am testing this idea because…, the main uncertainty is…, the next action is…, and I will review it when….” If you publish the experiment, separate plans from completed work and illustrative examples from actual results.

The audit succeeds when it gives one idea a fair test and lets the others wait. You do not have to abandon every possibility. You need a way to earn the next commitment through evidence instead of adding another project each time a tool expands what seems possible.

Questions you might be asking

Does returning to an idea for years mean I should build it?

It may indicate sustained interest or relevant experience, but it does not prove current demand. Look for recent evidence, a reachable audience and a small test before making a larger commitment.

Should I reactivate a separate domain for the experiment?

Only if a distinct brand serves a clear purpose. If an existing site reaches the right audience, a focused page or pilot there may answer the first question with less setup.

How do I choose between equally appealing ideas?

Compare the quality of evidence, audience access, your ability to deliver and the cost of the next test. Prefer the experiment that can resolve an important uncertainty within your available capacity.

Sources and further reading

• SBA: Market research and competitive analysis
• Anthropic: Starting with simple AI systems
• METR: Measuring actual productivity effects

Sources checked October 16, 2026. Tool features and plan availability can change; review the linked publisher documentation before connecting an account.

Put one idea to work

Use the idea revival worksheet to turn a recurring interest into a small test with a clear decision.

Choose an idea worth testing →Explore more from Wealth Experiment