FAQ | Impact Zone

GETTING STARTED

Getting Started

Where should we start?

Pick one workflow that is not customer facing to start with. Research shows the best returns are in software development, consulting quality, and writing speed. Ensure you have a clear beginning and a result you can actually measure. Skip that discipline and you're not piloting AI, you're buying expensive doubt.

You already know which bottleneck is costing you the most right now. Start there.

BUSINESS CASE & ROI

Business Case & ROI

How do we measure ROI?

Name the metric before you build anything, hours saved, error rate, cost per transaction, and measure where you stand today. Without that baseline, you can't prove the after to your board, and neither can anyone selling you the tool.

88% → 6%

88% of organizations use AI somewhere. Only 6% capture over 5% of EBIT from it, that gap is the value strategy closes. (McKinsey State of AI, 2025)

What are you actually trying to change? Answer that before you sign anything.

DATA, SECURITY & PRIVACY

Data, Security & Privacy

Is our data safe if we use AI tools?

Ask the vendor plainly how your data gets used: is it training their model, how long is it stored, is it actually anonymized or just labeled that way. If you can't get a straight answer before you sign, you won't get one after, and that's the answer your board will ask for.

What about compliance and regulation?

The rules around your data don't change because AI is now handling it, they get stricter, and the legislation is still catching up. Build your own standard for what data goes in and what stays out, don't wait for the law to draw that line for you.

Write your data policy before the tool goes live, not after something leaks.

IMPLEMENTATION & INTEGRATION

Implementation & Integration

Will this work with our existing systems?

This is where the real cost hides, not the AI tool itself, but what it takes to connect it to everything you already run. Ask specifically how a provider plans to access, clean, and integrate with your systems; that's where timelines and budgets actually slip.

What's a realistic rollout timeline?

Plan around your slow season, not your development calendar. Pilot one workflow, measure it, then expand, rather than launching when your busiest quarter is already eating every hour your team has.

You already know which system is the tangled one. Start the integration conversation there.

ETHICS, RISK & GOVERNANCE

Ethics, Risk & Governance

How do we prevent AI mistakes from causing real damage?

Keep a human in the loop anywhere a mistake would actually cost you something: money, safety, someone's trust in you. Build the review step in before launch, not after the first expensive mistake teaches you why you needed it.

Who's responsible when AI gets something wrong?

You are. The tool doesn't take the blame, and neither does the vendor who sold it to you. Assign who reviews, who signs off, and who owns the outcome before launch, or you're shipping a decision nobody actually owns.

How do we avoid bias in AI-driven decisions?

AI reflects whatever bias was already in your historical data, especially in hiring, lending, and how you treat the people you serve. Audit the outputs against real fairness before it costs you something you can't take back.

Name who's accountable before you need to. That's the whole practice.

CHOOSING A CONSULTANT OR VENDOR

Choosing a Consultant or Vendor

Should we hire a consultant or figure this out ourselves?

If you can name your problem, your data situation, and the number you're trying to move, you may not need help for a first small pilot. Bring someone in once you're choosing between unfamiliar tools or wiring into a complicated system, that's where a consultant's pattern-matching across other organizations actually pays for itself.