Hidden Costs: Don't Let AI Drain Your Operating Budget

Many companies are discovering that AI can create an unexpected problem: operating costs.

Consider a typical franchise workflow. A franchisee schedules an appointment. The CRM updates automatically, and a confirmation email goes out followed by a reminder text. A technician is assigned, the calendar updates, an invoice is generated, and the customer receives a thank-you message.

The question isn’t whether AI can handle every step. It’s whether it should.

For decades, software followed a simple economic model. Companies either purchased software outright or paid a subscription based on the number of users. Once it was running, the cost of performing another calculation or sending another automated email was essentially zero.

AI changes that equation.

Every prompt consumes tokens. Every document analyzed consumes tokens. Every conversation, every response, and every request has a cost attached to it. AI isn’t traditional software. It’s more like hiring a consultant who charges for every question you ask. One request is inexpensive, but one million unnecessary requests can become a significant operating expense. Managing those ongoing AI costs has become a discipline of its own, often referred to as “tokenomics.”

That’s where many organizations make an expensive mistake. They’re asking AI to perform tasks that traditional software has already mastered. Calculating royalties, determining whether a store is open, updating a CRM, sending reminder texts, scheduling follow-up emails, or automating marketing campaigns are deterministic processes. The same input produces the same output every time. Existing software already performs these functions efficiently and at a fraction of the cost. Using AI for those tasks is like hiring an attorney to tell you today’s date.

One franchise system with more than 200 locations learned this lesson firsthand. The company set out to replace its existing customer communication platform with an AI-first texting and email system. On paper, the economics looked compelling. A one-time development investment would replace a recurring monthly software expense while reducing cloud hosting costs.

Fortunately, the operations and IT teams tested the platform at a single location before rolling it out across the system. Customer engagement improved dramatically, but so did AI usage. The pilot generated far more text conversations than anticipated, creating AI costs that would have erased most of the projected savings if deployed across the entire system.

Rather than abandoning AI, the company redesigned its approach. Routine sales and marketing communications were handled through traditional automation while AI was reserved for complex interactions such as warranty questions, customer complaints, and other conversations that required judgment. The result was lower operating costs, stronger customer engagement, and a more scalable technology platform.

The lesson wasn’t that AI was too expensive. It was that AI is too valuable to waste on work conventional software already knows how to do.

AI delivers its greatest value where rules end and judgment begins. Here are tasks where AI can provide meaningful business value:

• Writing a thoughtful response to an unhappy customer
• Summarizing a franchise business coach’s notes
• Identifying unusual patterns across hundreds of locations
• Helping a franchisee improve performance by analyzing months of operational data

The strongest franchise systems won’t be the ones that use AI everywhere. They’ll be the ones that use it intentionally. Let software handle rules and automation handle repetitive tasks. Let AI step in where creativity, reasoning, and human judgment make the difference.

Too many executives ask, “Can AI do this?”

The better question is, “Should AI do this?”

The franchise brands that understand “tokenomics” will gain more than lower technology costs. They’ll build AI systems that deliver measurable value today and remain economically sustainable as they scale for years to come.

Dennis Leskowski, CFE, is a frequent author, speaker, and advocate for technology in franchising. A former brand CTO, he is currently the chief product officer for ClientTether, a franchise-specific CRM with automated sales and marketing channels with judicious AI engagement.