Case Study 2 — The No-Tipping Experiments: When the Arithmetic Was Right and the Answer Was Still No

Background

This is a case about a decision that was analyzed carefully, argued honestly, implemented by sophisticated operators — and reversed. It is here because it teaches the limits of everything Chapter 32 just taught you, and because the mechanism it turns on is exactly the fixed/variable split.

The American tipping system has a structural oddity that every operator eventually notices. A large share of front-of-house compensation is paid by the guest, directly, and never passes through the restaurant's income statement at all. In jurisdictions that permit a tip credit, the restaurant may count a portion of those tips toward its minimum-wage obligation, so a server's cash wage can be a small fraction of their actual earnings. Meanwhile the cooks — who cannot receive tips from guests in the ordinary course, and whose eligibility for tip pools is a contested area of law that has changed more than once — are paid entirely by the restaurant at market rates.

The result is a compensation structure with three well-documented consequences: a large and often resented pay gap between front and back of house; a front-of-house pay level that varies with volume, weather, and section assignment rather than with skill; and a body of research and litigation establishing that tipping outcomes correlate with things that have nothing to do with service quality.

Starting in the mid-2010s, several prominent American operators tried to fix it by eliminating tipping outright — raising menu prices, paying service staff a higher wage funded by those prices, and using part of the increase to raise back-of-house pay.

What is documented and reliable (Tier 1 — public record):

  • Union Square Hospitality Group publicly announced a program it called Hospitality Included in October 2015, eliminating tipping across its restaurants, raising menu prices, and paying service staff a higher wage. It was implemented over a period of years and drew extensive public attention.
  • USHG publicly ended the program in 2020, as it reopened following the pandemic dining-room closures, stating that it needed to get money into its employees' hands as quickly as possible.
  • Joe's Crab Shack, then part of Ignite Restaurant Group, publicly tested a no-tipping model across a set of its locations in 2015 and publicly abandoned the test in 2016.
  • A number of independent restaurants adopted no-tipping or mandatory-service-charge models over the same period. Some retained them. Many reversed. Both outcomes are well attested.
  • Several states have no tip credit at all, which changes the arithmetic of the decision substantially, and the federal and state rules governing tip pooling, tip credit notice, and service charges have been revised more than once during this period.
  • A service charge is legally distinct from a gratuity. It is generally the restaurant's revenue, subject to different tax and distribution treatment, and disclosure requirements vary by jurisdiction. Chapter 20 §20.4 covers the distinction; it is the reason the service-charge model and the no-tipping-with-higher-prices model are not the same thing.

What I will not do: attribute financial outcomes to any named company that were not publicly reported, speculate about internal deliberations, or invent a statistic about guest or staff response. The arithmetic below is a clearly labeled constructed composite.

The operating issue

Here is the decision, stated as a cost-structure question — which is what it actually is.

Eliminating tipping moves a large block of compensation from the guest's pocket onto the restaurant's income statement, funded by higher menu prices. In break-even terms that is three separate moves at once, and they do not point the same way:

  1. Revenue rises (higher menu prices), which is good.
  2. The labor line rises by more than the tips it replaces, because the whole point of the exercise is to raise back-of-house pay too, and because employer payroll taxes and workers' compensation now apply to money that previously arrived as a gratuity.
  3. The variable cost ratio rises and the CM ratio falls, because the added labor is a cost of doing business and the added revenue does not carry a proportionally better margin.

That third point is the counterintuitive one and it is the whole case. You can raise prices 20%, hold covers flat, and end up with a higher break-even and a thinner cushion.

A restaurant that makes the change

(Constructed composite. A 90-seat full-service restaurant, dinner six nights a week, 40,000 annual covers at a \$50 average check. Cost structure per the benchmarks in Chapters 1 and 31. This restaurant does not exist, and the figures are illustrative.)

FIGURE CS32.2-A — Before and after, at flat covers            [constructed composite]

                              BEFORE (tipped)         AFTER (no tipping, +20% prices)
  ───────────────────────────────────────────────────────────────────────────────────
  Covers                             40,000                     40,000
  Average check                      $50.00                     $60.00
  Revenue                        $2,000,000  100.0%         $2,400,000  100.0%
  COGS                              600,000   30.0%            600,000   25.0%
  Labor                             600,000   30.0%          1,020,000   42.5%
  Occupancy                         160,000    8.0%            160,000    6.7%
  Other operating                   300,000   15.0%            312,000   13.0%
  General & administrative           60,000    3.0%             60,000    2.5%
  ───────────────────────────────────────────────────────────────────────────────────
  OPERATING PROFIT                  280,000   14.0%            248,000   10.3%

  Note on the labor line: the $420,000 increase is (a) roughly $360,000 of tip income
  that guests previously paid directly and that now runs through payroll, and (b) about
  $60,000 of back-of-house raises funded by the price increase. Other operating rises
  $12,000 on card processing at ~2.8% of the additional $400,000 of revenue.

  SORTED BY BEHAVIOR
  Total FIXED cost                 $580,000                   $640,000
  Total VARIABLE cost            $1,140,000   57.0%          $1,512,000   63.0%
  ───────────────────────────────────────────────────────────────────────────────────
  CM ratio                            43.0%                      37.0%
  BREAK-EVEN SALES               $1,348,837                 $1,729,730
    as % of revenue                    67.4%                      72.1%
  BREAK-EVEN COVERS                  26,977                     28,829
  BREAK-EVEN COVERS PER NIGHT            86                         92
  Margin of safety                    32.6%                      27.9%
  Degree of operating leverage          3.07                       3.58

What it shows

Operating profit fell even though revenue rose 20% and covers did not move. \$280,000 to \$248,000 — down \$32,000, which is the back-of-house raises (\$60,000) net of the price increase's contribution above the tips it replaced, less the additional processing cost. That is not a failure of execution. That is the arithmetic of funding a raise: if you raise pay by \$60,000 and fund it with price, you have to sell enough additional contribution to cover \$60,000, and 20% more revenue at a worse CM ratio did not quite get there.

Break-even rose \$380,893 — six covers a night. From 86 to 92 on a room doing 128. The cushion fell from 32.6% to 27.9%, and the degree of operating leverage rose from 3.07 to 3.58, meaning every future revenue miss now costs about seventeen percent more in profit terms than it used to.

The tolerance for guest loss is negative. To match the old \$280,000 of operating profit, the restaurant needs \$920,000 of contribution, which at a 37.0% CM ratio requires \$2,486,486 of revenue — 41,441 covers, or 3.6% MORE traffic than before the change. The change does not merely require that guests accept a 20% price increase without leaving. It requires that they accept it and bring friends.

And an 8% cover loss is close to catastrophic. 36,800 covers at \$60 is \$2,208,000 of revenue, \$816,960 of contribution, and \$176,960 of operating profit — a 36.8% decline from the pre-change figure, produced by a traffic loss most operators would describe as modest. Margin of safety falls to 21.7%.

None of that means the decision was wrong. It means the decision was expensive, and that its justification had to be something other than near-term operating profit. The publicly stated justifications were, in fact, other things: pay equity between front and back of house, income stability for service staff, removal of a compensation mechanism with documented bias problems, and a hospitality experience without a transaction at the end of it. Those are real goods. They are simply not goods a break-even analysis can price.

There were also genuine cost-side benefits the composite above does not capture, and they belong in an honest accounting:

  • A large category of wage-and-hour exposure disappears. No tip credit means no tip-credit notice requirement, no dual-jobs / tip-credit-eligibility analysis, no tip-pool composition question, and no litigation risk in any of them. Chapter 20 explains why that is not a small thing: restaurants generate more wage-and-hour liability per dollar of revenue than almost any other industry, and most of it is unintentional. Removing an entire class of it has real expected value, even though you cannot put a clean number on it.
  • Scheduling becomes more honest. When servers earn a wage rather than tips, the manager who over-schedules is spending the restaurant's money rather than diluting the staff's, which surfaces a cost that was previously invisible. That is uncomfortable and it is better.
  • Retention effects, if they materialize, are worth real money. Chapter 17 computes the cost of turnover. In an industry running roughly 75% annual turnover, a compensation model that reduces departures pays for a meaningful share of itself.

Outcome

The record is mixed and it is worth reading precisely rather than as a verdict.

Union Square Hospitality Group ran Hospitality Included for roughly five years and ended it in 2020 during the reopening after the dining-room closures, stating publicly that it needed to get money to its employees as quickly as possible. Joe's Crab Shack's chain-level test ran for about a year and was abandoned in 2016. A number of independent restaurants adopted no-tipping models and some still operate them; others reversed within a year or two.

The mechanisms most commonly cited in public discussion of the reversals were, in rough order:

  1. Server compensation and retention. Strong servers in busy rooms could earn more under tipping, particularly on peak nights, and some left for tipped houses. The model compressed the upside of the best shifts, which is precisely the shifts the best servers want.
  2. Price legibility. A menu priced 20% above the restaurant next door reads as expensive on a listing, a review site, or a delivery app, none of which explain the compensation model in the headline. Guests compare prices before they compare policies.
  3. Guest confusion and residual tipping. Guests tipped anyway, or asked whether to, or felt the ambiguity as friction at exactly the moment Chapter 23 says the last ninety seconds matter most.
  4. The competitive-context problem. A single restaurant changing its compensation model in a market where nobody else has is carrying the entire adjustment cost alone. The system-level version of the change may well be better than the tipping system; the unilateral version is a different proposition.

What has grown instead, and substantially, is the mandatory service charge — a disclosed percentage added to the check. It achieves a large part of the same redistribution while keeping menu prices comparable to competitors', which addresses mechanism 2 directly. It is also legally distinct from a gratuity in ways that matter for tax, distribution, and disclosure, and the rules vary by jurisdiction. Verify locally, with counsel, before implementing one.

The lesson

Break-even analysis told the truth here and it was not sufficient. Everything in Figure CS32.2-A is correct: the change raises the fixed base, lowers the CM ratio, raises break-even by six covers a night, and requires more traffic than before just to stand still. An operator who ran that arithmetic knew exactly what they were buying.

What the arithmetic could not do is the four things that decided the outcome.

  1. It cannot price staff response. The model's effect on your ability to hire and keep the best servers is the largest single variable in the decision, and it does not appear anywhere in a fixed/variable split. Chapter 21's argument — that your people are the product — is the missing term.
  2. It cannot price guest comprehension. A 20% price increase and a 20% price increase that replaces tipping are the same number and completely different guest experiences, and break-even treats them identically.
  3. It cannot see the competitive context. The formula has one restaurant in it. The decision has a whole market in it.
  4. It cannot value the things that are not money. Pay equity between the line and the floor, income stability for people who cannot budget against weather, and the removal of a compensation mechanism with documented bias problems are legitimate objectives. They may be worth more than \$32,000 a year of operating profit. Break-even's job is to tell you that the price is \$32,000 a year and six covers a night, stated plainly, so you can decide whether it is worth it. That is the correct use of this chapter: not a verdict, a price tag.

And one operational takeaway, which is the most useful thing in this case for a small operator: any structural change that moves compensation from the guest onto your income statement raises your fixed base and lowers your CM ratio, and therefore raises your break-even twice over. That is true of no-tipping models, service-charge models, tip-pool restructurings that add employer cost, and any move from tipped to salaried service positions. Compute the new break-even before you announce anything, and state it in covers per night, because that is the form in which your managers will have to live with it.

Discussion questions

  1. Using the composite figures, compute the price increase that would have been required to hold operating profit at \$280,000 with covers flat, rather than the 20% actually modeled. Then argue whether that larger increase makes the decision more or less likely to survive guest response.

  2. The composite shows break-even rising from 86 to 92 covers a night. Rework it assuming the operator funds only the front-of-house wage change and skips the \$60,000 of back-of-house raises. What happens to break-even, and what happens to the argument for making the change at all?

  3. Several states have no tip credit, meaning employers already pay full minimum wage to tipped staff before tips. Explain how that changes the arithmetic of eliminating tipping, and predict whether no-tipping models should be more or less common in those states. What would you look for to test your prediction?

  4. A mandatory service charge and a 20% menu price increase move roughly the same money. Compare them on four dimensions: the CM ratio, price legibility to a guest comparing restaurants, the legal and disclosure treatment, and the effect on your break-even. Which would you choose for Bellwether, and what would you need to verify first?

  5. Case Study 1 argued that low fixed cost and high margin of safety are structural advantages. This case describes operators deliberately raising their fixed base for non-financial reasons. Reconcile the two — is there a principled way to decide when a fixed-cost increase is worth it? Write the test you would apply.

  6. Bellwether's plan carries labor at \$500,000 (32.3%), Chapter 19's schedule says \$570,461 (36.8%), and Chapter 20's classification says \$597,461 (38.5%). Suppose the chef-owner also wants to move the two most senior servers to salary to retain them. Using §32.7's method, price that decision in covers per night and write the four-sentence argument for or against.