The 8-word formula for restaurant success.

A few months ago, I sat down with Jason Cochran, Sweetgreen's COO. There was one part of our conversation that ended up on every whiteboard at our office:

More sales. More profit. Less cost. Better people.

Eight words. Simple enough to sound obvious, but when you start applying them to things you've already convinced yourself are good ideas you really see their value.

Jason was describing how he thinks about creating value in a business. Tying sales, profit, and cost to developing people. What interested me was how much easier they made it to separate a useful idea from a simply exciting one.

A week later in my follow-up, I told him we'd written the formula down and ran every decision through this model. If it didn't produce a positive number, we classified it as ‘noise’.

I've since found that the harder part is being honest about what counts as value.

A churro shop doesn't need every good idea

At ChurroMania, we like to say we reinvented the churro. We didn't invent it. We took something familiar and built a distinctive experience around it while keeping the product recognizable.

That gives us plenty of room to innovate, but it also gives us a reason (and defines pretty good criteria) to leave some opportunities alone.

The DubaiMania, our Dubai-chocolate churro innovation, made sense within that experience. The chocolate and pistachio trend gave us something new to do with the product people were already coming to us for. We could make it feel special without asking the guest to understand a different business.

In our experience, it supported a premium price without adding a comparable operating burden. More importantly, it gave people another example of what they could expect from us. Someone who enjoyed it could come back for that experience or be interested in our next churro innovation. That's the relationship we want to build, whether the trend is still popular six months later or not.

Now imagine we added matcha to follow the similarly exploding matcha trend.

We do already serve drinks, so Churro based shakes like our Frappmania make sense for us. We also serve coffee and hot chocolate, which traditionally have a place alongside our core churro product.

However, making high quality matcha (‘high-quality’ being another thing we pride ourselves on) would require us to learn something new and teach it across 150 stores and different geographies. We'd need ingredient sourcing, preparation standards, training, and enough consistency that the drink didn't depend on who happened to be working. For anyone that’s made their own matcha, you know that sounds easier said than done. Nobody wants a watered-down matcha just because we decided to participate in the trend.

We don't see an equally clear reason our guest would return to ChurroMania for it. We might attract someone once out of sheer curiosity. We might sell a few drinks. But before taking on the extra work, we'd want to know what the addition contributes to the experience people choose us for. Cue Jason’s model.

Dubai Mania builds on that experience, matcha would need to make its case. There are different goals the model forces us to optimize for even if both products can theoretically generate sales. This model asks us to think beyond that.

The work arrives after the idea

Delivery is a familiar version of this problem. Opening a new sales channel sounds straightforward and like a no brainer. In reality, the store has to handle another order flow, learn another platform, manage promotions, reconcile the data, and work out what it truly earned after commissions and several other random platform costs that exportable reports hide behind minutiae and terminology. Some don’t even tell you until several days later.

DoorDash's published U.S. marketplace delivery plans charge commissions of 15%, 25%, or 30%, depending on the plan. That doesn't mean delivery is a bad business. An incremental order can still be valuable. It means the revenue alone can't answer the question. Cue Jason’s model.

First, the team has to perform the extra work. If a channel adds profitable orders but makes the store harder to run, we need to understand that cost rather than leaving the manager to absorb it invisibly, especially if we expect this to scale across an entire system of stores.

The same applies to a new product, a software subscription, or the next location. Something can be useful and still demand more operating capacity than we have available. The decision gets better when we count the training, judgment, and ongoing attention alongside the expected sales.

That's why I don't treat Jason's formula as arithmetic. Profit and cost overlap. “Better people” doesn't necessarily fit neatly into a financial equation. The point is to examine the business from more than one direction before we decide an initiative worked. This also reminds me of Charlie Munger's idea of having multiple mental models informed by multiple disciplines to make decisions. The world is much more complicated than just numbers, although they are very useful and should be mastered. Questions and information are one tool of the model that evaluates different things, and decisions become much clearer and more useful.

Better people includes the people you already employ

When operators talk about better people, most people hear better recruitment. Find a stronger manager. Hire someone with more experience. Replace the employee who can't keep up.

Sometimes we do need to hire. But I've watched people at ChurroMania develop into operators whose judgment is difficult to replace (Steve Jobs would call them ‘A Players’): they understand the product and the guest and they can distinguish a real problem from noise. We don't want to keep looking outside for ability while underusing the people who already have it. Giving those people better tools is a form of development. So is giving them the time to coach others, the room to make decisions, and feedback on how those decisions worked.

We began automating reporting, labor planning, and inventory work because those tasks were holding people back. The important question wasn't only whether the labor percentage moved, it was what a good manager could do with time they previously spent assembling the plan. Still, a tool can also make that person less effective. If it adds another dashboard to check, another process to learn, and another set of exceptions to resolve, its advertised monetary return may hide the attention it consumes instead.

This became an important test for us at Clave: can the technology do the work without becoming yet another substantial part of the operator's work? The time it releases should have somewhere useful to go, not back into the tool in different ways.

Research outside the restaurant industry offers a useful example of what better-equipped people can look like. A study of customer-support agents found that AI assistance increased resolved issues per hour by 15% on average, with a 30% improvement among less-skilled and less-experienced workers.

What the guest gets from the decision

We don't want a restaurant where every improvement simply ends at a lower cost line. We want the person running it to have more ability to make the place worth visiting.

That may mean releasing a new item that feels unmistakably ours. It could also mean a manager who has time to teach a new employee how to look after a guest properly. Or it may mean declining a trendy addition because the team has something more valuable to work on. In the end, the financial result does matter (we operate businesses after all) but when we evaluate only the most immediate number, we can miss the work that makes the next visit happen.

I still like the eight words on the whiteboard. They make me account for more of what we're doing. A decision should improve the business for the people buying from it and the people making it work. If I can't explain how, I haven't finished thinking about the idea.