The Operator
Anderson, Distilled

💸 One pop-up night, all the way down

The ingredient margin is fine. What eats the business is fixed cost and uncounted labour — the shopping trip, the drive, the teardown, the Instagram post. Two thirds of the work is invisible in any recipe. ▶ _7LLcb

His own conclusion is the useful part: the pop-up was never a profit centre, it was paid market research. Free events for his apartment neighbours first, then charging, then watching whether the same people came back week after week. That repeat-purchase signal is what he treated as permission to raise money.

🏗️ What opening the actual restaurant cost

• Over $100,000 invested, from four sources: personal savings, small family loans, a small-business loan from a local Cleveland organisation, and a Kickstarter (rewards were gift cards and discounted pizza classes).
• The property search took over a year and a half. Lease signed mid-October 2024, then three months to open.
• He took a second-generation space — a pizzeria that had closed, keeping its oven, spiral mixer, walk-in coolers and hood. He names this as the single biggest lever on both cost and time versus building out raw space.
• He hired a commercial lease negotiation consultant, who got two months of free rent after opening and struck restrictive clauses. On a $100k project that consultant is rounding error.
• He also hired a pizza consultant for ~10 weeks (remote plus on-site) for permits, health-code compliance, POS setup, service systems and recipe documentation. Not for recipes — for the parts that are not cooking.
• ~200 applications → ~10 staff. Virtual interviews plus trial days.
• Since opening: 12+ hours a day, close to 7 days a week.

▶ HV6Atw ▶ L_qHT9

📊 What a good pizzeria actually earns

Two revenue benchmarks he got directly from owners:

• Brooklyn DOP — over $1.8M/year, ranked #5 slice in America. ▶ Bl8Dfy
• Pizzeria Deloro — 40,000 pizzas a year, over $1.5M revenue, which implies roughly $37 a pizza. ▶ lMJorm
• St. Francis of Pizza (Ohio's best, per his visit) — 950 sq ft, no walk-in cooler at all, and the owner hand-mixed dough for the first 8-9 months before he could afford a mixer. ▶ 1O66MZ

And the warning, from the owner of the $1.8M shop: a restaurant can look busy and still be "six months away from closing." The classic failure mode he names is food bills quietly piling up until you owe a supplier $10-20k. When sales go slow, full-time staff hours get protected first, and "everyone has to kind of roll with it."

⚙️ How Good Pizza holds quality without him

• Water is pre-chilled to a fixed temperature before mixing, so dough temperature is a controlled input rather than a daily surprise.
• Timed mixes, not feel: 8 minutes low speed → 20 minute autolyse → 16 minutes high speed.
• Bulk dough is split into four bins immediately after mixing. A big mass ferments unevenly because the centre cools far slower than the edges — he calls this the mass effect, and dividing is the fix. Brooklyn DOP does the same thing with speed racks and a light water mist.
• 72 hours total: 24h poolish + 24h bulk + 24h balled.
• A written dough-stretching guide with photographs, because stretching technique alone changes the pizza even when the dough is identical.
• A written pizza evaluation rubric, borrowed from Dan Richer at Razza, with crust width checked by caliper (~1 inch ± ¼).
• Oven procedure keys off both air temp and stone temp, with a target stone window of 600-675°F and explicit instructions for staying ahead during a rush.
• Cheese is a four-cheese blend shredded in-house every couple of days on a "cheese hog" — chosen over pre-shredded to avoid anti-caking agents and to control shred size, which controls melt speed.

▶ _un_Wq ▶ hlAR1P

The transferable rule: standardise everything that can be standardised and log it, so a bad plate can be diagnosed instead of guessed at. Reserve hand craft for the one or two steps where it measurably changes the result — for him, hand-stretching. He uses a machine to ball dough and a human to stretch it, and he can tell you why.

🧯 Failure modes he watched happen

• Unmixed sauce. Frank Pinello (Best Pizza, Williamsburg): "I've looked at the bottom of my container and there's just all salt on the bottom... you can ruin dozens and dozens of pizzas." A prep error that scales.
• Overproofed pan dough collapses during service and loses its crisp bottom. A timing risk, not a recipe risk.
• Fermentation drift in large batches, which is what the four-bin split exists to prevent.
• Cooler space is a real cost centre. Dough needs 3-6 days of cold ferment and most shops skip it because "space is money in a restaurant." He designed his fridge capacity around the ferment, and calls giving the dough time "the biggest upgrade in our pizza" — bigger than any oven or ingredient.
• Key-person risk, stated out loud. At Pizzeria Deloro, the $1.5M shop: "If Taylor is not here on Friday and Saturday, we just close." One dough-thrower is a single point of failure for a seven-figure business. That is what the written stretching guide and the caliper are really insuring against.

📐 A real production sheet, from a 40,000-pizza-a-year shop

His visit to Pizzeria Deloro produced the most concrete scaling numbers in the corpus:

• Dough balls scaled by format: 10oz for a 12-inch round, 22oz for an 18-inch, 28oz for Sicilian and grandma pans.
• A 100lb-rated mixer deliberately run at 50lb batches, because under-filling produces silkier dough. Anderson's own shop does the same thing in the other direction: a 50lb bag of flour yields ~80 dough balls, and he notes he could push it to 120-140.
• Two-stage cold ferment: individually proofed ~12 hours, 24 hours in one walk-in, then 24-72 more in a warmer one.
• Portioned by weight at the pass: an 8oz sauce ladle, 9-10oz of cheese, and cheese reduced deliberately for wet toppings (mushrooms, olives, onions, marinated peppers) to stop the pie going soupy.
• Par-baking needs humidity. They par-bake in a steam-injected combi oven at 350°F to hold ~70% moisture. This is the fix for the dry, crusty par-bake Anderson kept getting at home, and it is the difference between par-baking as a compromise and par-baking as an upgrade.
• They bake deliberately dark, and moved pan pizza from the deck oven to a conveyor purely for consistency. ▶ YFGOv2

✨ The cheapest differentiator he found

Every single pizza at his shop gets finished after the bake, and he calls it "a crucial part of what makes our pizza special": fresh basil, parmesan, olive oil, hot honey, chives, chimichurri, goat cheese, pecorino, adobo, teriyaki glaze, burrata, stracciatella. It costs almost nothing, survives the hold-and-reheat model intact, and is the one thing a par-baked slice cannot lose on the rack. He also puts sauce on top of the cheese at the shop, which reads more sauce-forward and keeps the crust from going soggy. ▶ 7aoyi3

Standard NYC slice-shop practice is to bake pies ahead and reheat to order. Anderson's finding is that this changes what the correct recipe is, which makes it the single most restaurant-relevant thing in his corpus:

• Baking once, serving immediately: 55-57% hydration is the sweet spot, because it has to crisp in a single bake.
• Baking ahead and reheating: 60%+ is better, and 65%+ gives the airiest L'Industrie/Scarr's-style result. The dough can afford to be wetter and chewier because the reheat restores the crisp.

The winning reheat, from a seven-method test: preheated baking steel at the oven's max (550°F) with the broiler on, about 60 seconds. Optionally microwave 10-15 seconds first to protect interior moisture, then finish on the steel. Losers: stovetop (soggy top), microwave alone (no crisp at all), and any longer, gentler oven bake — which just dries the slice out. Speed is the dominant variable, not gentleness. ▶ Y-ZN4L ▶ l9X18j

His own shop runs the same shape: par-bake ahead, hold on speed racks rather than in a traditional slice window, reheat to order, then finish with post-bake basil, parmesan and olive oil so the plate still reads fresh. This is the fourth independent arrival at a cook-ahead service model in this project.

🍅 The sauce, which is what his customers actually compliment

Blind-tasting eight canned tomato brands, Red Pack won, with 7-Eleven and Bianco di Napoli tied second. Raw sauce beat cooked — simmering "loses too much of that brightness" and drifts toward pasta sauce — but raw needs added sugar. His doses per 28oz: ~2g salt, 0.6-0.7g sugar per ounce of sauce, ~3 tsp oregano, ~½ tsp garlic, ~14 basil leaves. Thicker beat thinner, because thin sauce makes the crust soggy. Olive oil in the sauce was unnecessary once cheese was on it.

At the shop he uses Stanislaus Alta Cucina, and says guests compliment the sauce more than anything else — which he attributes almost entirely to the tomato brand rather than technique. Note the procurement catch: the best-scoring tomatoes only come in #10 food-service cans, which is exactly the sort of thing that separates a restaurant from a home kitchen. ▶ t9xRdK

🟫 Detroit, after a five-month quest

The headline technique is par-topping. Bake at 550°F for 7 minutes with only meat and a thin cheese layer, then add the rest of the cheese and the sauce stripes and bake 4 minutes more, with an optional 2-3 minute broil. This sets the crust structure before it has to carry the full weight of the toppings, which fixes the caved centre and the burnt cheese edge at once.

Supporting findings: 75% hydration (60% was dense, 90% floppy and caved); aluminium and blue steel pans performed identically; Wisconsin brick cheese beat mozzarella, with 40% muenster / 40% Monterey Jack / 20% mild white cheddar as the accessible substitute; grated or small-cubed cheese beat large cubes and slices, melting fully and using less cheese for the same coverage; sauce goes on after the first bake; 550°F beat 400°F outright. Cold ferment 1-4 days, "3 days is the sweet spot." ▶ Eq6TQY ▶ V6O8MO

Brick cheese does not transfer to NY-style. Under a short hot bake it separates into a grainy mess, and unlike mozzarella a broken brick-cheese emulsion cannot be fixed by reheating. An ingredient's suitability is a property of the bake, not the ingredient.

Seven videos, and it lands directly on top of a Chlebowski row. He ran the cut test twice, and the two runs disagree in a way that turns out to be the actual finding.

Test one: four cuts on a griddle, and beef knuckle (sirloin tip) won. He attributed that to slice uniformity rather than the cut, because the knuckle was the one he bought pre-sliced while his butcher's ribeye and strip came back unevenly hand-cut. Brisket lost badly. ▶ VhHzdr

Test two: five cuts, all mandolin-sliced to identical thickness. With slicing controlled, the ranking inverted: chuck eye was worst (stringy), knuckle was second-worst (dry, too lean), ribeye and NY strip were indistinguishable and both good, and dry-aged NY strip tasted best. His practical pick became plain ribeye. ▶ EcUGSn

Read together they resolve cleanly: slice uniformity dominates cut choice so completely that it can make a lean cut beat a fatty one. Control the slice and the cut differences reappear, with fat winning. For a kitchen that is an equipment decision before it is a purchasing decision, and it explains Chlebowski's near-tie (chuck 7.9 vs ribeye 8.1) as the outcome of a test where slicing was not the dominant variable.

And the cooking method: start the meat frozen. A hard sear alone came out dry and rubbery, a gentle cook came out tender but bland; starting from frozen lets the exterior sear hard while the inside stays cold enough not to overcook. Salt at the end, not before, or the slices weep and dry out.

The roll got its own quest. Three YouTube recipes all missed. He then reverse-engineered Amoroso rolls from the nutrition facts panel rather than the ingredient list, matching weight, protein, carbs and fat — because the ingredient list tells you what is in it but not in what proportion. Final: 59% hydration, 1% yeast, 2% diastatic malt, 1% salt, 10% Crisco. Diastatic malt was the single variable that mattered most; dropping salt to ~1% gave a lighter, more open crumb because salt tightens the gluten network. And a $3.50 disposable roasting pan over a baking steel got nearly the same steam result as a dedicated bread pan. ▶ gMz0Z9 ▶ bf7Xyf

💵 The $0.97 pizza, versus his own $2.31 one

He rebuilt a whole pizza from Walmart and Aldi discount ingredients and got an 11.5-inch pie to 97 cents, then ate it beside the pizza he actually sells at the pop-up, which costs $2.31. The expensive one was better, but not dramatically. His conclusion: "the process itself is way more important than the actual recipe you use."

The cost levers, ranked by how much they actually moved the number: cheese weight was by far the biggest (121g → 97g), diameter second, and hydration was nearly irrelevant — going 65% to 70% saved one cent. He runs this off a costing spreadsheet that models size, cheese and hydration against per-pizza cost, which is the kind of tool this project will eventually need.

In the same video he blind-ranked ten canned tomato brands: Alta Cucina and Cento All-in-One tied for first, Bianco DiNapoli third, and Aldi's $1.39 Happy Harvest beat Jersey Fresh, while Walmart's Great Value came last. So the tomato that his customers compliment is a genuine spend, but the tier below it is an Aldi can. ▶ QY87yz

One honest wrinkle, since it cuts against the cheese finding above: here the cheap whole-milk mozzarella performed worse than part-skim at a 5-minute bake, going rubbery and falling apart. That does not overturn the whole-milk verdict so much as confirm his own stated principle that brand and quality matter more than the fat split — a bad whole-milk cheese loses to a decent part-skim one.

Sodium citrate, for the third time

He builds a cheesesteak whiz on it, exactly as Chlebowski does for mac and cheese. Same conclusion independently: it is the compound, not the cheese, that makes a sauce stable. Three of four corpora now converge on sodium citrate, which promotes it from a trick to a house standard. Honest footnote: he ultimately judged his homemade whiz only subtly better than the jar, and "not really worth it" unless you eat cheesesteaks constantly. ▶ tU8j3g

Salt by weight — the same trap, found twice

Chlebowski proved every salt is the same sodium chloride. Anderson found the operational consequence: 1.5 tsp of salt is 4.22g of Diamond Crystal but 9.3g of another coarse brand — more than double. In a dough that is the difference between working and ruined. Weigh it. ▶ dh4rRc

Don't pay for the premium cut — now confirmed from two directions

Chlebowski scored chuck at $9.99/lb a 7.9 against ribeye at $23/lb an 8.1 for cheesesteak. Anderson's blind test agrees and explains it: what actually drives tenderness in thin griddled beef is slice thinness and uniformity, which is a knife problem, not a butcher-counter problem. Two corpora, one purchasing decision.

And the thesis-level agreement

Chlebowski's whole corpus argues that most premiums are not perceptible. Anderson, from inside a working restaurant, says most pizzerias use worse ingredients than a home cook can buy and win anyway on process. Both point the same way: spend on process, technique and time, not on the label. For a menu built out of other people's validated work, that is the most reassuring finding in the project — the edge we would be buying is cheap to source and hard to copy.

He is not really a fourth menu lane. Pizza is a lane, and a good one, but the reason this run matters is that Anderson is the only creator in the project who has tested the business, not just the food — and the business is where the thesis was thinnest.

The Menu Lab's premise has been that aggregating validated technique is the edge. Anderson's corpus supports the premise and prices it: technique really does beat ingredient spend, and he demonstrates it from inside a real P&L. But he also shows the edge is not the hard part. The hard part is $100k, an 18-month property search, 200 applications, 12-hour days seven days a week, and a consistency system elaborate enough to include a caliper.

The most useful single thing he gives us is the sequence: free events for people who will tell you the truth → paid pop-up → watch for repeat customers → only then raise money. That is a validation ladder with a real number attached at every rung, and it is the same shape I flagged earlier as the sane path. It is now sourced rather than asserted.