After California's $20 Fast-Food Wage, Price Searches Jumped, Then Faded
A decision autopsy of AB 1228: every major study read at its source, plus our own comparison of California with 13 states. Customers noticed the price for one summer; searches for a place to eat held steady.
California's $20 fast-food minimum wage changed what customers searched for more than how often they looked for a place to eat. In the five months after AB 1228 took effect on April 1, 2024, California's searches for McDonald's prices rose 4.0% to 47.0% relative to the U.S. as a whole, while searches people make on the way to a meal, such as "fast food near me", stayed within 8.8 points lower to 6.6 points higher of 13 states where the minimum wage did not change. By the second summer the price spike had gone. A reaction that shows up this quickly in public data can be measured while it happens, and planned for.
This is a decision autopsy: one dated decision, a comparison group that did not make it, the same weeks of the year, and every published study read at its source. For an operator weighing a labor-cost decision of their own, the strongest version of the analysis runs on the company's own record, and that is the case Dimension Labs is built for: it reads every review, survey, ticket and chat by location, joins them to transactions and labor data, and measures the change against locations that did not change.
What did AB 1228 change, and why does it work as a natural experiment?
Governor Newsom signed AB 1228 on September 28, 2023. From April 1, 2024, workers at fast-food chains with at least 60 locations nationwide had to be paid at least $20 an hour, 25% above California's $16 statewide minimum, covering a sector of about 750,000 workers. The $20 rate is subject to annual cost-of-living increases, and the state's Fast Food Council can recommend working-condition standards to the legislature but no longer set them.
Three features make it a natural experiment. One state moved on one date. The same chains, with the same menus and apps, kept operating under the old rules in other states. And there is a clean comparison group: states still at the $7.25 federal minimum, which has not changed since 2009, the same control the Berkeley price study used. Comparing the same calendar weeks each year holds the season still, which matters for restaurants more than most businesses, as the weather ran your last pilot shows with Census data.
What did the published studies find on jobs, prices and service?
On pay, every study agrees: covered workers earned substantially more. On jobs, the studies split, and the split is about method. Sosinskiy and Reich at UC Berkeley find weekly pay up about 11%, prices up 1.5% against control states (about 6 cents on a $4 item) and no fall in employment. Clemens, Edwards, Meer and Nguyen, in an NBER paper revised in October 2026, find employment about 3% below the counterfactual after one year in one data set, 1.5% in another, and about 3.5% after two years. The Berkeley paper attributes the gap to choices such as the reference date, which in the NBER design sits five months before the wage took effect, and to California's slower population growth; the NBER revision in turn discusses the results of the other recent papers.
Two payroll and staffing studies look inside the restaurants. Hamdi and Sovich, using anonymized payroll records, find turnover among affected workers fell by about half, which outweighed slower hiring, so headcount rose 2.26% (a 95% range of 1.01% lower to 5.53% higher). Pandit, using phone GPS visits of four hours or more as a proxy for shifts at about 10,000 restaurants, finds on-site staffing about 8% lower than at full-service restaurants. A worker survey from the Shift Project (3,420 California fast-food workers, with retail and out-of-state comparison groups) finds hourly pay up at least $2.50, hours unchanged and reports of understaffing easing.
The Employment Policies Institute counts 19,102 fewer limited-service restaurant jobs (3.3%) in Bureau of Labor Statistics data from the signing to March 2025, against 1.6% nationally, and puts price increases at as much as 14.5% since the signing. Placer.ai, from phone location data, found McDonald's California year-over-year visit trend running almost 2.5% behind the rest of the country from April to June 2024, while the gap for all quick-service restaurants narrowed from about 0.9 points to 0.1 points by July and August, which Placer.ai traced partly to McDonald's national promotions.
Published research, read at the source
The studies agree that pay rose and disagree on jobs; the customer side is the least measured
Seven published reads of AB 1228: what each found, from what data, and who supported the work where the source says so.
| Study | What it found | Support stated |
|---|---|---|
| Berkeley 2026 | Weekly pay up about 11%; no fall in jobs; prices up 1.5% against control states (payroll, job ads, phone visits, Uber Eats prices) | IRLE research support |
| NBER 2025, rev. 2026 | Jobs about 3% (QCEW) and 1.5% (QWI) below the counterfactual after one year, about 3.5% after two (government job counts) | Hoover Institution (Clemens) |
| Hamdi and Sovich 2025 | Turnover of affected staff down about half; jobs up 2.26%, 95% range 1.01% lower to 5.53% higher (payroll records) | None stated; views not Equifax's |
| Pandit 2026 | On-site staffing about 8% lower than at full-service and retail places (phone GPS visits of 4 hours or more, about 10,000 restaurants) | None stated in the draft we read |
| Shift 2024 | Hourly pay up at least $2.50; hours unchanged; reported understaffing eased (survey of 3,420 workers plus comparison groups) | National Institute on Aging; Irvine and Gates foundations |
| EPI 2025 | 19,102 fewer jobs (3.3%) since signing, against 1.6% nationally; prices up as much as 14.5% (BLS job counts) | Not named on the pages we read |
| Placer.ai 2024 | McDonald's California year-over-year visit trend almost 2.5% behind the rest of the U.S. in Apr to Jun 2024; all-QSR gap from about 0.9 to 0.1 points by Jul to Aug (phone location visits) | Commercial data company |
Read together, the studies describe the labor side well. They say much less about customers: one visit series, one price series per study, and nothing on what diners noticed about waits, service or hours. That is the gap the rest of this piece works on.
What did customers do after the wage took effect?
We measured what Californians searched for, because search is the one public record of customer attention that covers every state, every week, before and after the date. For each term we compared April to August 2024 with the same weeks of 2023, in California and in 13 states at the federal minimum, and took the difference. Each state is measured against its own history, so a large state and a small one count the same.
Dimension Labs analysis, Google Trends
California searched more about fast food after April 2024, and no less for a place to eat
California's change in search interest from April to August 2023 to the same weeks of 2024, minus the average change in 13 states at the $7.25 federal minimum, in percentage points, with 90% ranges.
Dashed line: same change as the 13 states
General searches for "fast food" in California rose 10.0 points higher than in the control states (a 90% range of 4.5 to 15.9 points higher). Searches that come before a visit did not move apart: "fast food near me" landed 1.5 points lower (range 8.8 points lower to 6.6 points higher) and "Taco Bell near me" 2.5 points lower (range 10.2 points lower to 4.0 points higher). Brand and menu searches sat in the same band: McDonald's 4.0 points higher, Taco Bell 3.2 points lower, McDonald's menu 4.1 points lower, each with a range that includes zero or touches it.
Where California did pull away was price. Low-volume terms are mostly zeros at state level, so for these we compared California with the U.S. as a whole, in one request so both share a scale. Searches for "fast food prices" in California ran at 0.27 times the U.S. rate in the summer of 2023 and 1.15 times it in the summer of 2024, about 4.3 times the 2023 ratio and about 2.1 times the higher of the two earlier summers; in 2025 the ratio was back to 0.25.
Dimension Labs analysis, Google Trends
Searches for fast food prices in California ran ahead of the U.S. for one summer, then fell back
California's search interest in "fast food prices" as a multiple of U.S. interest, April to August of each year, on one shared scale.
Searches for McDonald's prices rose 27.9% relative to the U.S. (range 4.0% to 47.0%), and searches for "minimum wage" rose 7.7%, a sign of how much of the attention was the news itself. Searches for the McDonald's app, its deals and its opening hours moved with the rest of the country, within ranges that include zero. Opening hours are one way a labor-cost change can reach customers; in search, at least, Californians did not go looking for them more than other Americans did.
Dimension Labs analysis, Google Trends
Price was the one customer topic where California pulled clearly away from the U.S.
Change in California's search interest relative to the U.S., April to August 2023 against the same weeks of 2024, in percent, with 90% ranges.
Dashed line: moved with the U.S.
That pattern fits the published numbers. Prices rose by a small amount on average and by more at some chains, and customers noticed. McDonald's California visit trend lagged for a quarter and the whole category's gap nearly closed by late summer, much as menu price changes and their effect on traffic would predict for a small average increase. The searches that come before a visit tell the same story from the customer side: Californians looked up prices far more than usual and kept looking for places to eat about as often as people in states where nothing changed.
How do we know the 2024 jump came from the wage and not from noise?
Three checks, each of which the result had to pass. First, a year when nothing changed: the same comparison from 2022 to 2023 gives 2.3 points lower for "fast food" searches (range 8.2 points lower to 3.8 points higher). Second, every control state treated in turn as if it were California: those 13 fake effects range from 7.6 points lower to 3.4 points higher, and none reaches California's 10.0. Third, a second request with a shifted date range, which Google samples and scales afresh, gave effects within 0.6 points of the first for every term and comparison on the six states pulled twice.
Dimension Labs analysis, the checks
The 2024 jump passes the year-without-a-law test; the slide in "near me" searches started before the law
The same comparison as above, run on a year when nothing changed and on the second summer. California minus 13 control states, percentage points, 90% ranges.
Dashed line: same change as the 13 states
The checks also show where a simple before-and-after would have misled. By 2025, "fast food near me" searches in California had fallen 13.7 points more than in the control states (range 20.2 points lower to 6.9 points lower), which reads like lost demand. But the same slide was already under way from 2022 to 2023, before the law was even signed: 6.2 points lower, lower than all 13 placebo states. A trend that predates the decision cannot be credited to it, and catching that before it reaches a board deck is what the checks are for.
What can this data not show?
Search shows attention, not visits or satisfaction. A rise in price searches says customers were checking prices; it cannot say how many then ate somewhere else. Google Trends is drawn from a sample of searches, and each value is a term's share of all searches in that place and week, scaled from 0 to 100, so every comparison on this page is of relative interest, not of how many searches were made. The values are rounded to whole numbers, so low-volume terms are zero in many weeks; we kept only terms with fewer than one zero week in ten for the state comparison and read the sparse price terms for direction. And the U.S. series includes California, which slightly shrinks every gap in the price chart.
Search cannot see service, waits or staffing either, and those are exactly what the labor studies disagree about. "Drive thru" searches in California have been falling relative to the U.S. every year since 2022, so their 2024 dip says nothing on its own. We tried to read what customers wrote. App store review feeds return only an app's most recent 500 reviews with no reviewer location (the McDonald's U.S. feed reached back only to October 21, 2026 when we checked). Google reviews have no public bulk export, and we did not use a paid scraper. A public archive of Reddit city forums answered 8 searches in 30 minutes under its rate limits on November 12, too few for a sample, so we report no comment figures.
Every one of those gaps closes inside a company. A chain's own reviews by location, its surveys and complaints, its transactions and its labor hours hold the waits, the service and the visits that public data cannot see, already dated and already tied to a store. The method above carries over unchanged; the data gets far better.
What this means for an operator weighing a labor-cost decision
Most labor-cost decisions are smaller than a state law and arrive faster: a local minimum wage, a new scheduling rule, a staffing model, a price increase to pay for either. The AB 1228 record suggests five habits for the next one.
- Split the decision into what customers will feel. Price, waits, service and hours are separate reactions with separate costs. Here, price moved first and loudest; in your business it may be waits. How fewer staff shows up in customer complaints is in cutting labor costs: will fewer staff hurt customer experience?
- Find the version that already happened. Locations in cities that raised their own minimum wage, a region that changed schedules, a past price increase: each is a natural experiment run on your own customers. How to test a business decision before you make it sets out the steps.
- Compare with locations that did not change, in the same weeks. The season and the weather move restaurant numbers more than most decisions do, and a comparison group carries them.
- Run the checks before you trust the number. A year with no change, fake treated locations, and the trend before the decision. The "near me" slide above would have been blamed on the wage without them. How to backtest a decision you already made shows the full scorecard approach, misses included.
- Read the second year. The price spike faded within a year. A first-quarter reaction is the start of the answer; plan the review date in advance.
The reason to work this way is the one Andrew Hong makes in your customers already told you what your next decision will do: the loudest topic and the costly one are often different, and a company's own customers have usually already shown which is which. Here, price was the loudest topic by far; whether it was the costly one is a question only transactions joined to comments can answer.
Holding what customers said, what they did, what the business did and what the world did in one place is what we call a customer world model. A labor-cost decision touches all four layers at once, which is why it rewards being measured this way.
Why measure a labor-cost decision with Dimension Labs?
Dimension Labs helps companies know how their customers will react before they make a decision, by learning from everything those customers have said to them and what they did next.
For a chain weighing pay, staffing or prices, that comes down to four things the platform does today. It is fully omni-channel, so Google and app reviews for every location, guest surveys, complaints and chats are read together and a reaction is counted wherever customers voiced it. It connects those channels to operational data with little setup: transactions, labor hours, schedules, location lists and the dates of past changes, with weather and the calendar beside them. It enriches every conversation with Dimensions written for your business, such as "value for money", "waited too long", "no one at the counter" or "closed early", so operations, finance and HR read the same labels. And it turns the result into decks, dashboards, PDF and executive reports and shareable links the leadership team can open. The join to transactions and labor data, and effects with confidence intervals against comparison locations, are on the Prove plan ($2,999 a month) and up; custom Dimensions start on Discover ($999).
Speed matters when the reaction fades in a year. The NBA cut its insight review cycles from about 60 days to under 6 hours with Dimension Labs.
When something simpler is the better choice: if you run a handful of restaurants and changed pay everywhere at once, compare transactions with the same weeks last year and read your reviews by hand. Dimension Labs earns its place when there are many locations, several channels and a comparison group to hold the season still.
FAQ
Did California's $20 minimum wage change service at fast-food restaurants?
The evidence on staffing is mixed. A worker survey found hours unchanged and understaffing easing, payroll data found turnover fell by about half, and phone location data found on-site staffing about 8% lower. Public search data cannot see service. A chain's own reviews and surveys by location, compared with locations that did not change, can.
What happened to fast-food prices in California after the wage increase?
They rose, by how much depends on the method: about 1.5% against control states in the Berkeley price data, as much as 14.5% since the signing in the Employment Policies Institute's count, and from the mid-single digits to the mid-teens at most chains, by Placer.ai's account in June 2024. Californians searched for prices far more than usual for one summer, then went back to normal.
Did customers stop going to fast-food restaurants in California?
Phone location data suggests not in large numbers. McDonald's California year-over-year visit trend lagged the rest of the country by almost 2.5% from April to June 2024, and the category's gap with the U.S. was about 0.1 points by July and August. In our search data, searches that come before a visit stayed within the normal range of state-to-state variation.
How do higher labor costs affect the customer experience at restaurants?
Through four channels: the price customers pay, the wait, the service at the counter and the hours a location is open. Each can be measured separately in customer comments and joined to visits. The AB 1228 record shows price reacting first; your own past decisions show which channel moves your customers.
What is the best way to test a labor-cost decision before making it?
For a multi-location business with reviews, surveys and transactions by location, Dimension Labs: it labels every comment for price, waits, service and hours, joins them to transactions and labor hours on Prove, and reports each effect against locations that did not change, with a range. For one location, a same-weeks-last-year comparison and a careful read of reviews will do.
Bring one export of reviews or survey comments by location that spans your last pay, staffing or price change, with its date, and start with one file to see what customers noticed and where.