The real problem with the pizza industry

The headlines are not kind to pizza right now. Pizzerias have fallen from Americans' favorite restaurant category to sixth place, trailing coffee shops and Mexican restaurants. Restaurant counts have been declining since 2019.

Two forces are squeezing the category from opposite sides: Inflation has pushed consumers to cut back on how often they order, and third-party delivery platforms have systematically captured the customer relationship, taking the margin, the data, and the loyalty that used to belong to the operator. The result is visible in the numbers. Consumers are ordering smaller pizzas with fewer toppings, franchise owners are watching margins compress from both ends, and structural shifts like the rise of GLP-1 medications are quietly pulling portion sizes and order frequency down further.

Most people look at that and see a category in cultural decline. I look at it and see an industry with an operational problem it has not yet been willing to name, and a new class of technology that could solve it, if operators are willing to let it act.

Industry challenges

The trend coverage tends to blame stagnant menus and shifting tastes, and those factors matter. GLP-1 medications are driving smaller portion sizes, tariffs are pushing ingredient costs higher, and consumers are ordering less often and spending less when they do. Those are real headwinds.

The brands that are actually growing through this period are doing it by running better operations. Domino's posted 5.2% same-store sales growth in Q3 2025, driven by a loyalty program overhaul and a DoorDash partnership that improved delivery economics. The throughline is not standalone menu innovation. It is operator intelligence, powered by systems that do not wait for a manager to notice a problem.

Pizza is one of the most operationally complex food categories to run profitably. Managing dough yield, tracking topping costs, and forecasting labor across variable demand all require discipline and connected data. Yet recent data showed that only 4.9% of pizzeria operators have adopted AI voice or chat ordering, and just 4.1% have implemented automated inventory management.

Over 70% of operators invested in new technology in the past 12 months, but most of that spending went toward refrigeration, ovens, and POS upgrades. The operational intelligence layer is still largely empty, and almost no one has moved to the next step: an Agentic OS that does not just surface information but takes action on it.

Technology solutions

The distinction matters more than it sounds. A dashboard that flags a food cost spike is useful. An agentic system that detects the variance, traces it to a specific ingredient, adjusts the next purchase order, and alerts the manager before the week closes is a different category of tool entirely. That capability exists today. Adoption rates have not kept pace with the urgency.

AI agents can produce impactful results when working on a connected system. When a loyalty program does not talk to the POS, and the POS does not connect to inventory or labor scheduling, AI is making decisions without the full 360-degree view of ordering, operations, and commerce data. The recommendations are based on siloed data, or at best, delayed data from a traditional integration, rather than a complete view of what is happening across the entire store.

When AI is connected to live ordering data, inventory levels, and labor schedules, it can do more than report. It can shift a prep task earlier when it sees a demand spike building. It can flag a delivery window as unprofitable and recommend rerouting before the order is accepted. It can identify which loyalty customers are at risk of churning and trigger a targeted offer without anyone pulling a report. The manager still makes the call on anything consequential, but the system is already working the problem before the question gets asked.

Operational pressures

Labor costs average 23–28% of sales across the industry, and the average hourly wage rose 3.86% in 2025 with no indication of stabilizing. Ingredient costs are up broadly, with tariffs adding unpredictability to supply chains that operators were already managing on thin margins. With 64% of customers now preferring to place orders digitally, operators who lack a first-party ordering channel are handing that relationship to a third party along with the margin and the data equity that comes with it.

Each of those pressures is addressable, but only when operators have real-time visibility and systems empowered to act on what they see. Labor scheduling is the clearest example. A well-integrated agentic platform does not wait for a manager to review last week's sales before building next week's schedule. It reads current reservation data, delivery order trends, and historical patterns, then drafts a schedule optimized for coverage and cost. A manager reviews and approves it. The work is already done.

Data shows that 66% of pizzeria operators are optimistic about the next 12-18 months, and 78.8% expect an uptick in gross annual sales. That optimism is reasonable when paired with operational discipline. On its own, it does not close the gap between your best store and your worst store.

Growth strategies

The operators best positioned to grow share a few attributes.

• They are investing in connected technology rather than layering disconnected tools;

• They are deploying AI that takes action on data rather than just displaying it;

• They are building loyalty programs designed to change behavior, not just send blanket discounts to a list that goes nowhere.

The restaurant technology industry sold a lot of software to pizza operators over the past decade, most of it positioned as best-in-class standalone tools. Much of that software is now part of the problem. When operators manage a dozen disconnected platforms, data does not flow across them, and AI built on fragmented data produces noise, not outcomes.

The operators who have already moved toward an Agentic OS with real context equity are pulling ahead in ways that compound over time, and the distance between them and everyone else is growing.

The question is not whether your ordering data talks to your labor model. It is whether your system is doing something about it when it sees a problem, before you do. Whether your AI is generating reports or generating outcomes.

Pizza is under pressure, and the tools to respond have never been more capable. The decision to treat the operational problem as the actual problem, and to deploy AI that takes action on it, is the one that matters most.