Starbucks Is Discounting Strike Math the Market Hasn't Priced — Yet

I read the SBUX desk note at 6:18 Friday morning the way I read most sell-side product around a labor event — by skipping the price target and going straight to the assumption table. The four big-bank notes I had on my desk this week all use a Q1 SG&A walk that bakes in a labor-line variance of plus or minus 40 basis points, which is the standard envelope for a QSR coming off a December comp. None of the four had moved that envelope to reflect the strike. None of them had moved it to reflect the 170 unfair-labor-practice cases sitting at the NLRB. None had moved it to reflect the $38.9 million Fair Workweek settlement that’s already on the New York City public docket. The strike is six weeks old today. The math is a coffee gone cold next to the calculator.

The contrarian thesis I want to put down before the next batch of sell-side updates lands is this: Starbucks is, on Dec 19, a labor-cost story the sell-side hasn’t fully modeled, and the market is going to have to price it twice — once for the direct labor envelope, once for the regulatory comparable the strike puts on the chain’s books. The “less than 1% of stores” frame the company has been running since Day 8 is fine as PR. It’s an analytical trap when you put it next to the NLRB caseload and the Fair Workweek number. And the CTO transition that’s been reported as forthcoming for several weeks now lands on top of all of it, separately, later this month — a second variable the buy-side has not yet put in the same model.

What’s actually on the board, by the numbers

Let me lay out the facts that matter for the model, all of which are knowable today. The strike has covered more than 180 stores in 28 cities for six consecutive weeks and has generated 170 unfair-labor-practice cases at the NLRB, of which more than 130 are filed in the Seattle region alone. (KING 5) Two weeks ago the union escalated with a Manhattan rally outside the Empire State Building that ended in twelve planned arrests and a rare floor appearance by SEIU President April Verrett, which my colleague walked the day it happened. (CNBC) On Day 36 the picket line moved to the Starbucks Support Center on Utah Avenue South in Seattle, which I read separately yesterday as a rhetorical inflection.

The company has held one line publicly: average pay and benefits across the U.S. partner base run roughly $30 an hour, and the strike affects fewer than 1% of the chain’s 17,000 U.S. stores. (KING 5) Both statements survive a fact-check. Neither belongs in an analyst model the way the sell-side is currently using them.

The fourth fact, the one that gets less play in the strike coverage but matters more in the cash flow model, is the $38.9 million Fair Workweek settlement Starbucks signed with the New York City Department of Consumer and Worker Protection. That is a regulatory comparable, not a rumor. It is a priced answer to a question the union is going to ask in every future bargaining cycle: what is the cost of non-compliance at a multi-store operator in a Fair Workweek jurisdiction? The answer is now on the public docket.

The sell-side’s labor envelope is the wrong instrument

Here is the analytical problem. The standard sell-side QSR model treats labor as a single line, walked quarter to quarter against same-store sales and minimum-wage step-ups. That line carries a variance envelope — call it 30 to 50 basis points of operating margin — which is fine for normal weather. It is the wrong instrument for the cost stack a six-week strike actually puts on a multi-unit operator with a unionizing footprint.

The direct labor line is the smallest piece. Run the back-of-the-envelope. If 180 stores are striking and the chain has roughly 10,000 company-operated U.S. stores, the gross exposure is 1.8% of the operating base. Cut that by the share of revenue those stores generate — they’re concentrated in dense urban markets where Starbucks runs its highest-volume locations, so the revenue share is materially higher than the store share; call it 3 to 4% of company-operated revenue, being conservative. Apply a six-week duration, and the direct foregone contribution margin lands somewhere in the $35 to $60 million range depending on how much volume diverts to nearby unaffected stores versus how much walks out the door. That is not a number that breaks an earnings call.

The indirect lines are where the model needs to move.

Line one: the regulatory comparable. The $38.9M NYC settlement is not a one-time event in a free-cash-flow model. It is a benchmark that every other Fair Workweek jurisdiction’s enforcement agency now has on its desk. Philadelphia, Chicago, San Francisco, Seattle, Los Angeles. The exposure-by-jurisdiction math is straightforward — count the stores, multiply by the per-violation schedule, discount by historical enforcement intensity — and the expected-value line item for additional Fair Workweek exposure across the unionized footprint is, on my cut, comfortably above $50M over the next eight quarters even before you assume any escalation. The sell-side has not added a line for this. They should.

Line two: the NLRB caseload. 170 cases, most of which will settle and a handful of which will produce remedial orders, do not blow up an income statement on their own. They do create a litigation-and-remediation cost band that runs alongside the labor line for two to three years. The closer comparable here isn’t a previous QSR strike — there isn’t one of this scale — but the FedEx and Amazon NLRB caseloads from the early 2020s, which produced a sustained 10-to-30 basis point drag on operating margin for the units involved. Apply that drag to Starbucks’ unionizing-store cluster, and the model needs another line.

Line three: the rhetorical bill. This is the line the buy-side will say isn’t quantifiable, and they’ll be wrong. Hiring costs at the unionizing-store cluster are already running above the chain average by my back-channel reads, the elected-officials letter of support is now at 180-plus signatures, and the company’s communications strategy has been to issue statements rather than to produce visuals. In a six-week labor story, that’s a measurable brand-cost line. It shows up in two places in the model: the next-twelve-months hiring-cost-per-store, which I’d mark up 8 to 12% versus a non-strike baseline at the affected stores, and the cost of capital on the next round of organizing, which the union is now going to run cheaper because elected officials have given them the air cover.

Add the three indirect lines together and you get a labor envelope that’s two to three times what the standard 30-to-50 basis-point variance contemplates. That’s the wedge the buy-side hasn’t put in yet.

The CTO transition lands on top, separately

The other variable the model has to absorb, and it lands on a separate clock, is the CTO transition that has been reported as forthcoming for several weeks now and is anticipated to formally close before year-end. I want to be precise about how I’m framing this on Dec 19: the transition is forthcoming, the timing is anticipated, and the right read for an M&A reader is not that this is a strike-driven exit but that it adds an execution-risk variable to a quarter that’s already carrying the labor variable.

The reason this matters analytically is that the CTO seat at Starbucks sits at the intersection of the loyalty-program economics, the mobile-order infrastructure, and the labor-scheduling stack — and labor scheduling is the technical substrate underneath Fair Workweek compliance. A clean handoff is a non-event. A rushed handoff during an active strike, with 170 NLRB cases and a $38.9M jurisdictional comparable freshly on the books, is the kind of org-chart move that adds two to four quarters of execution-risk discount to the multiple even when the underlying technical leadership is strong.

I’d advise reading the transition announcement, when it lands, against the strike timeline rather than as a standalone tech-org story. The sell-side is going to want to put it in the tech-product bucket. The right bucket is the labor-and-systems bucket.

What I’d be doing if I covered SBUX

A few specifics, for the readers who run the actual model.

First, move the labor envelope. The standard 30-to-50 basis-point variance for Q1 is too tight. I’d open it to 80 to 120 basis points and explain the move in the assumption table by reference to the strike duration, the NLRB caseload, and the Fair Workweek comparable. The buy-side that complains about the looser envelope is the buy-side that will be marking the same line down on its own a quarter later.

Second, add a discrete Fair Workweek exposure line to the FCF walk. Don’t bury it inside SG&A. It is large enough to deserve its own row, it is jurisdictional enough to model store-by-store, and the existence of the $38.9M NYC number means the assumption is no longer speculative.

Third, treat the CTO transition as an execution-risk overlay on the Q1-Q2 walk, not as an HR footnote. Discount the FY26 EBITDA bridge by 50 to 100 basis points until the handoff is observable in the operational metrics — average order time, mobile-app uptime, scheduling-system error rate. If the handoff is clean by the Q2 call, take the discount off. If it isn’t, the discount widens on its own.

Fourth, model the strike’s end as a discrete bargaining event, not as a return-to-normal. The 170 NLRB cases will not vanish on the day the picket lines fold. They will be the bargaining chips in the settlement window, and the settlement window will produce structural changes to the labor scheduling and Fair Workweek compliance posture that show up as a permanent step-up in the labor line, not a one-time charge.

The bottom line

Six weeks. 180-plus stores. 170 NLRB cases. $38.9M of priced regulatory comparable sitting on the public docket. A CTO transition landing separately, later this month, on top of the same operational base. The sell-side is writing a coffee story. The model needs a labor-cost story.

I’ll write the after-action piece the day the strike ends. The math by then will be easier — duration becomes known, the settlement framework becomes legible, and the labor-line step-up becomes a number rather than a band. What I’d flag now, on Dec 19, is that the band itself is the model. The market is currently pricing the midpoint of an envelope that was sized for normal weather. The weather is not normal.

The coffee gets cold either way.

— Marcus edits The Bottom Line for TableTransfers. Tips: [email protected].