302: How Restaurants Can Survive During A Pandemic With Hanson Li

Podcast: Download (Duration: 54:13 — 62.3MB)

A well-run restaurant operates on a 10% to 12% margin, because food costs 27% to 30% of sales, labor costs 35% to 40%, and occupancy adds another 10% to 12%. That leaves almost no room for error, which is why the restaurants that survived 2020 were the ones that rebuilt their business around takeout, delivery, and grocery.

Hanson Li is an old Stanford classmate of mine who runs Salt Partners Group, a development and investment company in food and beverage. His portfolio spans 14 restaurants from Humphry Slocombe ice cream to Atelier Crenn, a Michelin three-star.

This episode covers the full economics of running a restaurant, what it costs to open one, and the specific pivots his group made when dining rooms closed.

Key takeaways

• Prime cost is 70% of sales: 27% to 30% food plus 35% to 40% labor.
• A 10% to 12% operating margin is considered good in the restaurant industry.
• Opening a 2,000 square foot Bay Area restaurant costs $500,000 to $2 million, with plumbing behind the walls driving the high end.
• Alcohol runs 80% to 90% gross margin and roughly 30% of a full-service restaurant’s sales.
• A San Francisco full liquor license cost about $250,000 pre-pandemic and fell to around $160,000.
• Takeout replaced only 10% to 30% of pre-pandemic revenue.
• A $345-per-person Michelin three-star became a $38 three-course kit that consistently sold out.
• Third party delivery takes 15% to 30%, which is why they redeployed staff as their own drivers.

How Hanson Li got into the restaurant business

Hanson was born in Hong Kong, where food was central to family life, and credits Stanford with his love of cooking.

He had kitchen access all four years and made midnight nachos for his dorm, which is where feeding people became the thing he enjoyed.

His degree was biology and sociology, on a pre-law track aimed at genetics law, until he fell asleep in his first biochemistry class.

He opened his first restaurant at 28 in 2002, a coffee house in downtown Austin that he and two friends bought and rebranded for a couple hundred thousand dollars. It is still open 18 years later.

Salt Partners started in 2014, after ten years of investment banking and private equity mostly in China, when he decided at nearly 40 that finance was not a long-run fit.

What a restaurant’s cost structure actually looks like

Starting from sales at 100%, cost of goods runs 27% to 30%. That covers produce, proteins, and wine.

Team costs run 35% to 40% in San Francisco, including salary, benefits, and payroll taxes.

Those two together make 70%, which the industry calls prime cost.

Occupancy is the next bucket at 10% to 12%, covering rent, utilities, and insurance. That puts you at 80% to 83% before general and administrative costs.

An operating margin of 10% to 12% is a good outcome within the industry, and it should produce a solid return on the original investment.

Why restaurant owners overpay on merchant processing

Hanson’s founding hypothesis for Salt Partners was that owner-operators working seven days a week let expenses drift because they have no time to examine them.

His favorite example is credit card processing. Owners open a business bank account, get signed up for merchant processing at the same time, and never learn that the rate is negotiable.

Rates are shoppable and processors are switchable, which is one of the first things his group revisits.

On a 10% margin, that fee difference is material.

The same problem applies to food costs. A busy restaurant might take deliveries from 20 vendors a day, and answering what you paid for eggs last week often means going through physical invoices.

His group captures that information electronically across restaurants, so prices that have crept up become visible.

Why restaurant food margins are lower than you think

Food at 27% to 30% of sales means a 70% gross margin on food, which surprises people who assume a $1.50 bowl of rice is nearly all profit.

The rice is cheap and the beef balances it out, and the blended result is that 70%.

That compares unfavorably to software and well against most physical goods businesses.

Restaurants buy from wholesale distributors on a completely separate supply chain, in pack sizes built for volume. Eggs arrive 180 at a time in 15 dozen flats, and flour comes in 10 or 25 pound bags.

Bulk pricing helps on some items and not others. Far fewer distributors will deliver eggs to a restaurant than sell them to consumers, so those distributors hold pricing power.

What it costs to open a restaurant

A 2,000 square foot Bay Area restaurant runs $500,000 to $2 million.

Two million buys the higher end and half a million buys something simpler.

Kitchen complexity is what moves the number. Finishes can vary tenfold between cheap and expensive materials, and the genuinely expensive work is what you cannot see.

Plumbing is the usual culprit. Rerouting it can escalate costs astronomically depending on the layout.

Taking over an existing space can be cheaper if you go in accepting that things stay where they are. Moving an existing wall usually costs more than building a new one.

Funding is limited. Most banks will not lend on a new restaurant without attached real estate, so the money generally comes from friends, family, and your own circle.

How alcohol economics work in a restaurant

Alcohol carries an 80% to 90% gross margin and represents roughly 30% of sales at a full-service fine dining restaurant.

California issues a couple dozen license types, covering brewing, corner liquor stores, brew pubs, and full-service restaurants separately.

San Francisco caps certain license types, which creates a taxi medallion effect on price.

A full San Francisco liquor license cost roughly $250,000 before the pandemic. It fell to around $160,000 within months, and Hanson expected further declines as more businesses failed.

Losing that 30% of sales is what made dining room closures so severe, since takeout initially could not include alcohol at all until the rules changed a week in.

How much revenue takeout actually replaces

Takeout and delivery brought in 10% to 30% of what these restaurants were doing before.

Total consumer spending on outside food fell even as takeout rose, and Hanson notes he eats maybe half his meals out in a normal week.

The alcohol gap explains much of the rest. People drink the wine they already have at home rather than ordering it with a delivery.

Staffing followed the revenue. Bars closed entirely on March 16 with no takeout or delivery legally possible, and even restaurants still operating had to let 70% to 80% of staff go.

How a Michelin three-star restaurant pivoted to takeout

The Crenn group combined Atelier Crenn, Bar Crenn, and Petit Crenn into a single takeout operation after the first week of shelter in place.

The price moved from $345 per person to a $38 three-course meal to go.

The decision was driven by the team’s desire to keep feeding people, and by the belief that hospitality still mattered during a difficult period.

The wine program adapted too. A three-bottle blind tasting kit ships with the bottles wrapped in foil and a printed tasting guide, which proved popular as entertainment for people stuck at home.

The kits sold out consistently, and a meaningful share of buyers were experiencing Crenn for the first time because the normal price point made it a special occasion place.

Packaging became the ongoing design problem. People come to a restaurant for the experience as much as the food, and the team kept iterating on how to deliver that hospitality in a box.

How to run a virtual dinner party from a restaurant

The next step Hanson was planning was a virtual dinner party built around the meal kits.

The format is a Friday evening hour where roughly 20 guests log on with their kits, alongside Chef Dominique Crenn, Hanson, or wine director Courtney.

The structure is one-to-many. The wine director walks the group through the wines for half an hour, or a chef live streams how to prep, reheat, and plate the dinner.

The framing behind it is that when you cannot bring guests to the restaurant, you bring the restaurant to their home.

Why the ice cream shops stayed open

Humphry Slocombe kept every retail store open through the shutdown.

The reason is that ice cream already had a robust delivery business across Postmates, UberEats, DoorDash, and Caviar.

That channel grew sharply with everyone at home, which offset the lost walk-in traffic enough to keep the shops running.

The stores changed function. They became pickup points for delivery providers, selling pints rather than scooped cones.

How corporate sponsorship funded restaurant meals

Brown Sugar Kitchen in Oakland, run by Chef Tanya Holland for over ten years, got funded by a neighbor.

Marqeta, a payments company headquartered a few blocks away, reached out at the start of the pandemic and paid the restaurant to make meals for local nonprofits.

The arrangement produced a couple hundred to 350 meals a day, three days a week, to three different Oakland nonprofits.

Lexus funded a parallel program at the Crenn group, sending 200 meals every Friday to UCSF for a commitment of 3,000 meals over 14 weeks.

Both came from existing relationships. Dominique Crenn is a Lexus brand ambassador, and Marqeta had hosted their holiday party at the restaurant in December.

The structure works twice over. Corporate social impact budget flows to a restaurant that needs revenue, and the food reaches medical workers and nonprofits whose volunteer bases had collapsed.

How restaurants turned into grocery stores

The mechanism is the separate supply chain. Restaurants buy produce in bulk from distributors consumers cannot access.

Hanson’s group repacked those bulk goods into $30 boxes with eggs, bacon, and initially a box of disposable gloves.

Other restaurants went further. A chef in the Mission District converted his restaurant into a general store selling everything from canned goods to fresh fish by the pound.

The advantage is access. Restaurant supply chains reached seafood and staples that grocery stores were short on, during a period when eggs, flour, and yeast were all difficult to find retail.

The other advantage is density. A neighborhood has one or two large grocery stores and a hundred restaurants, so restaurants can act as distributed pickup points.

Why restaurants should run their own delivery

Third party delivery platforms take 15% to 30% off the top.

Hanson’s reasoning is straightforward. Paying 30% to a delivery company is money that could instead pay his own staff to do the same work.

The conditions made it easier than usual. Customers were almost always home, so delivery was far simpler than what DoorDash and UberEats normally manage.

That also kept hours for staff who would otherwise have been laid off.

The approach varies by restaurant. Crenn kits are pickup only, some restaurants run their own delivery, and Humphry Slocombe works with every third party platform.

What happens to restaurant economics at 50% capacity

Hanson’s anxiety about reopening centered on capacity restrictions and six-foot table spacing.

Run the cost structure at half revenue. Variable costs drop, and much of the model is fixed.

Labor is only partly variable. A base level of staff is required whether you serve a full dining room or half of one.

Rent, utilities, and insurance do not move at all.

For a restaurant already operating at 1% to 5% margin, serving half the guests makes reopening economically pointless, which is why he expected many storefronts to stay empty.

What the restaurant industry looks like long term

Hanson’s prediction splits into two phases: the period before a vaccine, and the world after it.

Consumer habits will have changed either way. After 18 months of ordering in, people will have discovered that a good local Italian restaurant delivers food as good as a chain dining room for less.

Chefs who resisted third party delivery and packaged food have been forced through that barrier, and he expects a genuine expansion in food sold online over the following years.

The role of the dining room shifts as a result. Good food used to be the main reason to go to a restaurant, and it becomes one reason among several.

What remains is what cannot be boxed. Gathering with people, hospitality, and the experience itself become more important in a world where everyone is used to takeout.

Frequently asked questions

What is a good profit margin for a restaurant?

Ten to 12% operating margin is considered good in the industry. Food runs 27% to 30% of sales, labor 35% to 40%, and occupancy another 10% to 12%.

What is prime cost in a restaurant?

Prime cost is food cost plus labor cost combined, typically around 70% of sales. It is the primary number restaurant operators manage against.

How much does it cost to open a restaurant?

A 2,000 square foot Bay Area restaurant runs $500,000 to $2 million. Kitchen complexity and plumbing behind the walls drive the difference far more than visible finishes.

What is the margin on alcohol at a restaurant?

Eighty to 90% gross margin, and alcohol represents roughly 30% of sales at a full-service fine dining restaurant, which is why losing it hurts so much.

How much does a liquor license cost in San Francisco?

A full liquor license ran about $250,000 before the pandemic, dropping to roughly $160,000 as businesses closed. San Francisco caps certain license types, creating a taxi medallion effect on price.

How much revenue does takeout replace for a restaurant?

Ten to 30% of pre-pandemic sales in Hanson Li’s portfolio. The alcohol that normally accounts for 30% of a full-service restaurant’s revenue largely disappears with takeout.

Should restaurants use DoorDash or their own delivery?

Third party platforms take 15% to 30%, so restaurants with available staff can redirect that money to their own drivers, especially when customers are reliably home.

Why do restaurants overpay on credit card processing?

Because owners typically get signed up for merchant processing when opening a business bank account and never learn the rate is negotiable and the processor is switchable.