office-workers-drinking-coffee

The Three Questions Every Business Owner Should Answer Before Investing in AI

Posted July 23, 2026 in AI, FSM Podcast Resources by

There is no shortage of AI use cases to pursue. Ask any team to brainstorm where AI could help and you’ll end up with a list that runs across every function in the business. Marketing wants it for content. Sales wants it for prospecting. Operations wants it for reporting. Customer service wants it for response time. Finance wants it for analysis. The ideas are real, the enthusiasm is genuine, and the list keeps growing.

The problem isn’t generating ideas. The problem is figuring out which ones are actually worth pursuing, in what order, and how to make that call without spending months in evaluation mode or committing budget to something that doesn’t deliver.

Henry Hays has a three-question framework he applies to every proposed AI use case before recommending it to a client. It’s not a complicated framework. But it’s a disciplined one, and discipline is what turns a long list of possibilities into a short list of priorities.

Question One: Does It Make the Business More Efficient?

The first question is whether the use case makes the business more efficient in a way that can actually be measured.

That qualifier matters. Efficiency as a concept is easy to claim and hard to verify. Almost any AI application can be described as making something more efficient. The question is whether the efficiency can be quantified in a way that means something to the business: time saved per task, volume of work handled with the same headcount, reduction in error rates, faster turnaround on a process that has a known cost when it’s slow.

If the efficiency case for a use case can’t be made in specific, measurable terms, that’s a signal to either sharpen the use case until it can or move on to one that’s more clearly defined. Vague efficiency claims are how AI investments get approved and then quietly fail to justify themselves when someone asks what the return has been.

The specificity requirement also forces a useful conversation about what the AI is actually doing. A use case that produces measurable efficiency has a clear workflow behind it: defined inputs, a specific process the AI is handling, and a defined output. That clarity is what makes implementation straightforward and evaluation possible.

Question Two: Does It Make Your People More Productive?

The second question shifts from the business level to the individual level. Does this use case make the people doing the work more productive in a way that can also be measured?

Productivity at the individual level and efficiency at the business level aren’t the same thing, and both matter. A use case that automates a process the business runs without meaningfully changing what any individual on the team can accomplish may produce efficiency without producing productivity. A use case that gives a skilled person back two hours a week by removing a task that required their time but not their expertise produces both.

The productivity question also surfaces the human dimension of the use case in a useful way. For AI to make someone more productive, that person has to actually use it. Which means the use case has to fit into how they work, produce outputs they trust, and save them effort they’d rather not spend. Use cases that look efficient on paper but create friction for the people they’re supposed to help tend to get used inconsistently and eventually abandoned.

Hays is specific about the measurement requirement here too. It’s not enough to say the AI will make the team more productive. The plan should include a way to know whether it’s working: a baseline of how long a task currently takes, a target for what it should take with AI support, and a process for checking whether the target is being reached.

Question Three: Does It Create a Path to More Revenue?

The third question connects the first two to the business outcome that ultimately justifies any investment: does this use case create a path to more revenue?

That path doesn’t have to be direct. An AI application that gives your customer service team back three hours a day doesn’t automatically produce more revenue. But if those three hours get redirected toward higher-value customer interactions, that’s a path. If the efficiency gain allows the business to handle more volume without adding headcount, that’s a path. If the productivity improvement allows a sales team to have more conversations or follow up more consistently, that’s a path.

The revenue question is also useful as a filter for use cases that are technically interesting but not strategically important. AI can do many things that are impressive without contributing meaningfully to how a business grows. The revenue question cuts through the impressive and asks: does this actually matter to the business?

Not every use case has to have a direct revenue connection to be worth pursuing. Some efficiency and productivity gains are worth capturing even if the revenue link is indirect. But if a proposed use case can’t answer the revenue question at all, that’s worth taking seriously before committing to it.

What the Framework Does to a Long List

One of the most practical applications of this three-question framework is what it does to the long list of AI use cases that most companies generate when they start thinking about this seriously.

Hays describes a common pattern: a company identifies 19 ideas across six different parts of the business. Everyone is excited. The ideas are real. But 19 is too many to pursue, and there’s no obvious way to prioritize without some kind of filter.

Run those 19 ideas through the three questions, honestly, and most of them fall away. Not because they’re bad ideas, but because they can’t clearly answer all three questions in specific, measurable terms. What’s left is a much shorter list of use cases that have a real business case behind them, a clear implementation path, and a way to know whether they’re working.

That shorter list is where the investment should go first. The other ideas don’t disappear. They wait until there’s proof of concept from the first round, and then they get reassessed with the benefit of what the organization learned.

Applying the Framework in Practice

Using this framework well requires two things that are harder than they sound: specificity and honesty.

Specificity means that answers like “it will save time” or “it will help the team” don’t count. The framework only works if the efficiency, productivity, and revenue answers are concrete enough to be verified after the fact. If the expected outcome can’t be described in terms that would let someone check in six months and know whether the investment delivered, the use case isn’t well-defined enough to pursue yet.

Honesty means being willing to let good-sounding ideas fail the test. Some use cases will be advocated for by people who are enthusiastic about them, and enthusiasm is not a business case. The framework is useful precisely because it provides a shared standard that isn’t about enthusiasm. It’s about whether the use case will actually move the business.

Companies that apply this framework consistently find that their AI investments produce clearer results, get adopted more reliably, and build internal confidence faster than companies that pursue AI initiatives based on excitement or competitive pressure alone. The framework doesn’t slow things down. It makes the things that move worth moving.

Listen to the full podcast episode here.

Related Resources

Tags:
, , , , ,