If the owner thinks AI is hype, don't pitch AI.

Pitch a problem he already cares about. Owners don't buy technology. They buy fewer missed calls, faster quotes and less time on paperwork. The tool is just how you get there.

This is for managers and employees at small businesses who see an opening and can't get a yes. Here's how I'd approach it.

Why owners say no

Most skepticism is reasonable. The owner has watched a decade of software promises. He has been burned by a website, a CRM or an agency that overpromised. He worries about three things:

  • Cost and distraction. Another subscription, another thing to manage.
  • Risk. A tool sends something wrong to a customer and the brand pays.
  • Work. If the idea lands on his plate, he's the one who has to make it work.

Your pitch has to answer all three, which means it should be small, cheap, reversible and owned by you.

Pick a problem he can see

Start with something the owner already complains about, in his own words. Some examples that tend to land:

  • "We lose jobs because nobody calls back fast enough."
  • "Quotes take two days and sometimes never go out."
  • "Nobody asks old customers to come back."

Look for the place where money is entering the business and leaking out before it becomes a job. That's usually follow-up, and I explain the reasoning in my rule for what to automate first. Avoid leading with content creation or anything that sounds like a toy.

The one-page pitch

Keep it to one page. Use these headings.

  1. The problem. One sentence, in his words, with a rough cost. "About ten calls a week go to voicemail. If even two are real jobs, that's real money."
  2. The pilot. One narrow thing. "Text back every missed call within two minutes with a message and a booking link."
  3. What it costs. Tool cost per month and your time. Be honest, including what you don't know.
  4. What we measure. Two or three numbers, their current values and a check date. Response time, booked jobs, missed calls.
  5. The guardrails. What humans review, what the system can't do on its own and how to turn it off.
  6. The ask. "Thirty days. If the numbers don't move, we stop."

That last line carries most of the weight. You aren't asking him to believe in AI. You're asking for a thirty-day test with a stop condition.

Answers to the usual objections

I'm writing these as the responses I'd give, not as a script.

"AI makes mistakes." True. That's why a person reviews anything that touches a customer or money until it has a track record. Start with internal drafts or simple, controlled messages.

"My customers want a real person." They want a fast answer from someone who knows the job. A text that says "we got your call, here's when we'll reach you" is better than silence. You're improving the part customers already don't get.

"I don't have time to learn this." You won't need to. You own the pilot. His part is to look at one number each week.

"What about our data and privacy?" Fair question. Check the vendor's own documentation on how data is stored and used before you connect anything, and keep customer details out of tools you haven't vetted. If you're unsure, ask before you proceed.

"We tried something like this before." Ask what happened. Usually the answer is that nobody owned it or nobody measured it, which is the pattern in what to check before buying AI. Show how this pilot is set up differently.

Make it safe to say yes

Owners say yes to small, reversible things. So:

  • Use the cheapest tool that can do the job.
  • Keep a human in the loop for anything customer-facing.
  • Write down the baseline before you start.
  • Report in plain numbers, not tool screenshots.
  • Name yourself as the owner of the pilot and the weekly check.

If you want to build the skills that make you the person who can run this, AI skills for operations managers is a good next read. And for the margin argument behind all of it, see AI Isn't Magic. It's Margin.

Timing and framing matter

Don't pitch in the middle of a bad week. Pick a calm moment, ask for ten minutes and bring the page. Lead with the problem, not the technology, and let him react before you get to the tool. If he asks a question you can't answer, say so and come back with the answer. Owners trust people who admit what they don't know far more than people who sound certain about everything.

It also helps to bring a second person with a stake in the problem, such as the person who answers the phones. When the owner hears the pain from the one living it, it stops being your idea and becomes a shared fix.

If he still says no

Some owners won't move, and that's their call. You can still use AI on your own work, such as drafting, summarizing and cleaning data, and show results that way. A small, visible win from your own desk is often a better argument than any pitch. If you'd like help framing a pilot, or want an outside voice in the room, start with Di-Hy's free audit.

FAQ

How do I pitch AI to my boss?

Pitch a problem, not a tool. Bring a one-page plan with a narrow pilot, a cost, two or three measurable numbers and a stop date.

What if the owner thinks AI is a fad?

Don't argue. Pick one specific, costly problem, like missed calls, and propose a thirty-day test with a clear way to stop.

What should the first AI pilot be?

Something near revenue and low risk, like missed-call follow-up or quote reminders. Avoid anything where a mistake could reach a customer unreviewed.

How do I show the pilot worked?

Record the baseline before launch and check the same numbers on a set date. Hours saved only count if the time went somewhere useful.