The Pre-Mortem: 5 Questions to Ask Before Hiring an AI Consultant

What is a Pre-Mortem?

A pre-mortem is simple: instead of waiting for your project to fail and asking "what went wrong?", you imagine it's already failed and ask "why did it fail?" before you start. It's the most valuable exercise in project management, and it takes 30 minutes.

We apply this thinking to every engagement. Here are the five questions you should ask any AI consultant before you hire them.

1. "Can you show me requirements you've written for a similar project?"

Why this matters: The number one cause of AI project failure is poor requirements. If your consultant can't show you a clear, structured requirements document from a previous project, they probably don't write them. And if they don't write proper requirements, your project will be scoped on assumptions and delivered on guesswork.

Red flag: "We'll figure out the requirements as we go." This is consultant-speak for "we don't do proper analysis."

2. "What does a realistic timeline look like?"

Why this matters: AI projects take longer than vendors claim. A focused chatbot implementation takes 8-12 weeks, not 4. A computer vision system for manufacturing takes 16-24 weeks, not 8. Anyone promising faster delivery is either cutting corners on requirements and testing, or hasn't done it before.

Red flag: Any timeline that doesn't include a dedicated discovery and requirements phase at the start.

3. "How do you define success, and who decides when it's been achieved?"

Why this matters: If nobody defines acceptance criteria before the project starts, nobody will agree on whether it's been delivered. This is the single most common cause of vendor disputes, and it's entirely preventable.

Red flag: "We'll define success metrics once we see what the model can do." Success should be defined in business terms before any technical work begins.

4. "What happens if the project stalls or the approach doesn't work?"

Why this matters: Not every AI approach works. Good consultants have contingency plans, pivot strategies, and honest conversations about risk. Bad consultants blame the data, the team, or the timeline.

Red flag: Any consultant who guarantees results without qualification. AI outcomes depend on data quality, organisational readiness, and dozens of other factors.

5. "Do you have a commercial relationship with the platforms you're recommending?"

Why this matters: Many AI consultants have referral arrangements with technology vendors. That means their "impartial" recommendation might be influenced by commission rather than suitability. There's nothing wrong with vendor partnerships in principle, but you deserve to know about them.

Red flag: Resistance to answering this question directly.

Bonus: The Question You Should Ask Yourself

"Have I defined the business problem clearly enough that a stranger could understand it?"

If you can't explain the problem you're trying to solve in plain English, without jargon, you're not ready to hire a consultant. Start with the problem definition. The solution comes second.

Our Approach

At EfficiencyAI, we encourage clients to ask us all five of these questions. We can show you our requirements documents, we'll give you honest timelines, we define acceptance criteria before work begins, we have contingency plans for every engagement, and we have zero vendor partnerships or referral arrangements.

That's what impartial, analysis-led consulting looks like.