Five Questions to Ask Before Starting Any AI Project

Why These Questions Matter Before Any AI Project

Before starting any AI project, the questions that matter most are not technical; they are foundational. We’ve seen millions spent on AI initiatives that were doomed from the start, not because the technology was wrong, but because nobody paused to think critically about the fundamentals.

Here are five questions we ask at the start of every engagement. They’re simple, but the answers are revealing.

1. What Specific Problem Are You Solving?

Not “we want to use AI” or “we need to be more innovative.” What specific, measurable problem is causing pain today?

Good answer: “Our review team misses 12% of issues because they’re handling 200 cases per hour.”

Bad answer: “We want to leverage AI to transform our operations.”

If you can’t articulate the problem in one sentence, you’re not ready to start.

2. How Are You Solving It Today?

Understanding the current process, including the humans involved and the expertise they bring, is essential. This tells you:

• What’s actually working (and shouldn’t change)
• Where the genuine bottlenecks are
• What expertise needs to be preserved or augmented

Pay particular attention to the knowledge that is not written down anywhere. Every organisation has people whose experience makes them disproportionately effective: who can read a situation quickly, who know which edge cases matter, whose judgement has been calibrated by years of exposure to the problem. This tacit knowledge is often the most important input to how the process actually works, and it is the part most likely to be ignored when someone maps current-state workflows on a whiteboard.

Skip this step and you risk automating the wrong things.

3. What Does Success Look Like?

Define your metrics before you build anything. Be specific:

• Accuracy: What error rate is acceptable?
• Speed: How much faster does it need to be?
• Cost: What ROI are you targeting, and over what timeframe?
• Adoption: What percentage of your team needs to use it?

Without clear success criteria, you’ll never know whether your project actually delivered value.

4. Do You Have the Right Data?

AI needs data. But not just any data: the right data, in sufficient quantity, with adequate quality. Before committing to an AI approach, assess:

• Availability: Do you actually have the data, or do you need to start collecting it?
• Quality: Is it clean, consistent, and well-labelled?
• Volume: Do you have enough for the approach you’re considering?
• Bias: Does your data represent the full range of scenarios you’ll encounter?

Data problems are the single most common reason AI projects stall after kickoff.

5. Who Will Use It, and Do They Want It?

The best AI system in the world is worthless if nobody uses it. Understanding your end users is critical:

• What’s their current skill level with technology?
• Are they threatened by or enthusiastic about AI?
• How will it integrate with their existing workflow?
• What training and support will they need?

The human element isn’t a nice-to-have; it’s the difference between a successful deployment and expensive shelfware.

Use This Framework Before Every AI Project

Print these questions out. Put them on a whiteboard. Make them the first slide in every AI project kickoff. They won’t guarantee success, but they’ll dramatically reduce the probability of failure.

If after working through them you want a more structured view of your organisation’s readiness, a fixed-fee advisory session can go deeper on your specific situation.

Want help working through these questions for your organisation? We’re here to help.