Restaurant Gary Danko
A prix-fixe that quietly runs the whole playbook
Some menus need a behavioral overhaul. This one is closer to a finished proof: a room that opened in 1999 and engineers the choice without anyone noticing.
Menu-craft grade
Tiered courses that engineer the middle, round whole-dollar prices with no cents, a caviar anchor, curated build-your-own freedom and a tableside cheese cart that doubles as reassurance. It already runs the playbook.
Graded on how well the menu uses behavioral economics, not the food.
- Type
- Independent
- Where
- San Francisco, Fisherman's Wharf
- Cuisine
- Contemporary American / French
- Footprint
- 1 location
- Since
- 1999
- Ownership
- Chef-owned, James Beard winner
The setup
Gary Danko's menu is deceptively plain: pick 3, 4 or 5 courses ($97 / $120 / $143) from a single list of appetizers, fish and seafood, meat and game birds, cheese and dessert. The chef's-choice tasting is simply the 5-course price; there is no separate tasting tier. No photos, no exclamation points. It reads like restraint. It is actually a stack of behavioral levers, executed so smoothly they look like good taste.
If most of our breakdowns are 'here is what to change,' this one is closer to 'here is why it works,' and the two or three places even a master could push a little further.
On the menu
Prices are round whole dollars with no cents, '3 Courses $97,' caviar 'Imperial Golden $145.' The menu prints the dollar sign, but never charm-prices: no .95 endings, no cents anywhere. Individual dishes carry no price at all; you pay by course count. Dish names are spare, ingredient-first lists, with no sales adjectives. (As sampled, 2026; menus change.)
Choose any dishes from any section
↳ the tiering is the strategy
with Salsify, Leeks and Lettuce Cream
↳ signature appetizer (as sampled, 2026; dishes rotate seasonally)
with Yukon Potatoes, Swiss Chard, Bordelaise Butter and Cassis-Shallot Marmalade
↳ signature meat course (as sampled, 2026; dishes rotate seasonally)
Presented Tableside
↳ 15 to 20 cheeses, rolled to your table
1 ounce with Signature Buckwheat Blini
↳ the luxury anchor (California Select / Black Sea Reserve / Imperial Golden)
What they get right
The behavioral economics already at work.
The 4-course is the engineered middle
Three tiers create a Goldilocks structure. Extremeness aversion pushes diners off the cheapest (3) and the priciest (5) toward the middle, and adding the 5-course option mathematically raises the 4-course's share. With the chef's tasting priced at the same 5-course $143, the top is capped, not extended.
a top tier shifted the middle's share ~20 points, Simonson & Tversky, 1992
Round whole-dollar prices, no cents
'$97,' not '$96.95.' The menu prints the dollar sign, but the prices are round whole numbers with no cents, and round numbers read as 'premium' where 9-endings read as 'discount.' Pricing dishes by course count, rather than per item, also mutes how many individual charges a diner is tallying. Exactly right for this room.
round-vs-charm pricing signals premium vs discount, Schindler & Naipaul on round-vs-charm pricing
Choose-your-own-courses is the format Sutherland loves
Rory Sutherland's favorite menu gives you a curated structure you're still free to deviate from. Danko's build-it-yourself prix-fixe does exactly that: it maximizes what he calls 'the feeling of having made a good decision.'
Sutherland on the Chapito set-menu; 'people maximise the feeling of having made a good decision'
The caviar service anchors the page
A Tsar Nicoulai Osetra Caviar Service at $105 / $125 / $145, sitting at the top of the menu, is a high anchor. Once a diner has seen a single ounce of caviar priced near the cost of a whole 5-course dinner, $97 to $143 for the full prix-fixe reads as reasonable by contrast. The anchor does its work whether or not anyone orders it.
an early high number reframes everything below it, anchoring; Tversky & Kahneman 1974
The cheese cart is a costly signal
Rolling 15 to 20 cheeses to the table nightly is theater you can see and choose from: it removes uncertainty and turns service into ceremony. Sutherland: the ritual is part of the flavor. The dessert course ends the meal on a high (the peak-end rule).
What we’d test
The rewrite, with the expected lift and the honest caveat.
Point at the middle
The tiering already favors the 4-course; a light wording touch, noting it as the most-chosen, or simply listing it first in the price line, would amplify on the page the compromise effect Danko already gets for free.
Expect More diners self-select the margin-friendly tier
Caveat Menu wording and ordering only; this room sells understatement, so keep the cue subtle.
A sentence of story on the signatures
The menu is a pure ingredient list. For one or two signatures, a short line of provenance or method in the dish copy can lift perceived value, the cleanest, best-evidenced naming effect there is.
Expect Higher perceived value on the hero dishes
Caveat Menu copy only; fine-dining diners prize restraint, so test one or two dishes, don't narrate the whole card.
Move the caviar anchor to the very top
If the caviar service is not already the first thing the eye lands on, leading the page with it sets the highest reference point before any course price is read, making the $97 to $143 tiers feel modest by comparison.
Expect Course prices read as more reasonable against the caviar anchor
Caveat Menu layout and ordering only; no change to dishes, prices or service.
What diners actually say
Synthesized from public reviews, the reality check that grounds every recommendation.
They praise
- Impeccable, personable service
- The cheese-cart ritual
- Consistency over decades
- Build-your-own flexibility
They criticize
- Feels 'dated' to some
- Cheese cart less lavish than it was
- Value questioned at the price
The verdict
Gary Danko is what the output of a behavioral-economics menu audit looks like when it is done right: tiered courses that engineer the middle, round whole-dollar prices with no cents, a caviar service that anchors the page, curated freedom and ritual that doubles as reassurance. The remaining gains are amplification and story: point harder at the middle tier, give a signature dish a line of provenance, and make sure the caviar anchor leads the page.
Sources
- Restaurant Gary Danko, official menu (PDF)
- Restaurant Gary Danko, menus
- SFist, Gary Danko loses its Michelin star (2024)
- Gary Danko, MICHELIN Guide (recommended, no star)
- Compromise / extremeness aversion, Simonson & Tversky, 1992
- Cornell, removing the “$” lifted spend ≈8%/person (Yang, Kimes & Sessarego, 2009)
- Rory Sutherland, “The presentation of choice” (the Chapito set menu)
Your menu next
Get this for your own menu, free.
Send your menu and we’ll send back the same breakdown, what you get right, what we’d test, and why.
Zuni Cafe
A 75.00 chicken that takes about 75 minutes, and the menu says so. A practical warning that also happens to reduce uncertainty, printed in the margin.
Read it →