Where Starbucks customers went, and who came back

How we listened

We collected every public post and comment since January 2022 that mentions Starbucks in 14 online communities: Starbucks' own customer and employee communities, communities for Dunkin', Dutch Bros, 7 Brew and McDonald's, home-brewing and independent-cafe communities, frugal-living and personal-finance communities, and one Texas city community. That is 2,527,928 posts and comments.

92,774 of them talk about thinking of leaving, cutting back, leaving, switching or coming back. A language model read each of those in full and wrote down the same five things every time: what the person did, how often they used to go, where they went, why, and what brought them back. Keeping all five together is what lets us follow a departure from start to finish (why that matters). The result is 8,984 customer moments we can count.

Why we count per 1,000 mentions

A raw count rises and falls with how busy a forum is. So every number below is how many customers said something for every 1,000 posts that mention Starbucks in the same month. When that number goes up, a bigger share of the conversation is about leaving, not just more conversation overall. The field note density, not presence explains why this matters.

Most of Starbucks' own forum is employees

About three in four posts in Starbucks' own communities come from people who work or used to work there (Starbucks calls them partners). We sorted every author into employee, customer or neither, and only customer posts count toward the customer numbers. Employees get their own section. We also set aside posts that looked coordinated (new or single-topic accounts, bursts of near-identical text): 50 of the 92,774.

We split the time into four periods, named for what was happening at Starbucks:

data table · who wrote the posts in Starbucks communities, by period

period	employees	customers	neither
before	81%	19%	0%
the break	75%	25%	0%
turnaround push	74%	26%	0%
after first closures	72%	28%	0%

When it broke

From January 2022 to September 2023, about 1.5 of every 1,000 customer posts mentioning Starbucks described leaving. From October 2023 to August 2024 it was 3.1. dfmchn's reading of 2.5 million public posts found that talk of leaving Starbucks doubled, from 1.52 to 3.10 per 1,000 mentions, between the 2022 to 2023 baseline and the period from October 2023 to August 2024.

It has not come back down. During the turnaround push under the new CEO (September 2024 to September 2025) it eased to 2.64. In the year since the first store closures it rose again, to 2.91.

data table · customer talk by period, per 1,000 mentions

period	thinking about leaving	cutting back	leaving	coming back
before (Jan 2022 to Sep 2023)	0.22 (0.19 to 0.25)	0.57 (0.53 to 0.62)	1.52 (1.45 to 1.60)	0.12 (0.10 to 0.14)
the break (Oct 2023 to Aug 2024)	0.30 (0.26 to 0.35)	0.94 (0.85 to 1.03)	3.10 (2.95 to 3.26)	0.21 (0.18 to 0.26)
turnaround push (Sep 2024 to Sep 2025)	0.30 (0.26 to 0.35)	0.84 (0.77 to 0.91)	2.64 (2.51 to 2.77)	0.20 (0.16 to 0.23)
after first closures (Oct 2025 to Sep 2026)	0.32 (0.27 to 0.38)	0.86 (0.78 to 0.96)	2.91 (2.75 to 3.07)	0.29 (0.25 to 0.35)

A slow drift or a sudden break?

If customers were drifting away slowly, the running total of leaving talk would climb in a straight line. If a crowd were tipping over all at once, it would bend into an S. The best description for Starbucks is simpler: a sudden break. Around December 2023 the pace of leaving talk jumped by about two thirds and stayed there. The pace changed, but the crowd has not finished moving, which is the kind of change that a dashboard tends to catch late.

The first shift in leaving talk starts in November 2023, within four weeks of the first boycott calls (mid-October 2023) and the Red Cup Day strike (November 16, 2023). That timing is suggestive, not proof, and this particular shift does not survive a stricter test.

data table · when each series shifted

series	level shift starts	events within four weeks	holds under a stricter test
thinking about leaving	none	none	yes
cutting back	December 2022	none	yes
leaving	November 2023	boycott calls (Oct 15, 2023); Red Cup Day strike (Nov 16, 2023)	no
coming back	April 2024	none	yes

The single worst month was March 2026, at 5.08 per 1,000 mentions. Nothing on our event calendar falls within six weeks of it, so we are not going to guess at a cause.

Does thinking about leaving come first?

Talk about thinking of leaving peaked in February 2026, and talk about actually leaving peaked one month later. Month to month the two rise and fall together. But the timing gap between them is not reliable: a pattern this strong shows up by chance about one time in seven. Read them as two views of the same mood, not as an early warning.

“I spend, minimum $15 at Starbucks, at about 3-4x/week. bc I need to fill my 24oz thermos for work. AND THAT'S WITH STARS. I just started looking into my own home machine because I did the math for last month and my coffee budget was like...”

“everytime i buy starbucks i feel guilt and regret whenever i buy starbucks, i feel immense guilt and regret. i feel like i had just betrayed a loved one. should i just buy it less often or am i losing it?”

How people describe leaving

Few people announce a clean break. More often they cut back, or they switch to something else and say so.

data table · customer moments by type and period

period	thinking about leaving	cutting back	leaving	switching	coming back
before	219	564	713	791	118
the break	145	451	727	765	103
turnaround push	189	518	818	817	122
after first closures	141	379	683	592	129

The regulars are the ones leaving. Where leavers say how often they used to go (2,286 moments), the most common answers are some version of every day: "daily" (315), "every day" (170), "everyday" (94), "every morning" (69), "all the time" (45), "every single day" (40).

The price people quote went up and stayed up. 1,129 moments mention a specific price. The typical price quoted was $6.00 in 2022 to 2023 and $7.00 in every period since.

“I used to go daily, now it’s 1x a week and sometimes 1x every two weeks. Food and drink for two isn’t worth the money spent”

“I’ve cut waaaaay back on Starbucks. High prices and too many calories.”

“Coffee at Starbucks. Just a plain cup of coffee. Quit going there well over a year ago.”

Where they went

Nearly half of leavers name no destination at all. They just go less. We keep them as their own group rather than dropping them, because it is the largest group in every period.

data table · all leaving and cutting-back moments

period	leaving and cutting back	named a place	went less, named nothing	home brewing	no clear place	local independents
before	2,068	900	47%	28%	9%	6%
the break	1,943	844	48%	27%	9%	6%
turnaround push	2,153	937	48%	25%	9%	7%
after first closures	1,654	694	49%	21%	9%	6%

Among leavers who do say where they went, home brewing is still the biggest alternative by far, but it is shrinking: 65% of named destinations in 2022 to 2023, 51% in the latest year. The fastest-growing destination is Dutch Bros. Among Starbucks leavers who named where they went, Dutch Bros grew from 3% in 2022 to 2023 to 14% in the year to September 2026, according to dfmchn's analysis of public customer posts. 7 Brew went from nothing to 3%. Dunkin' and local independent cafes held steady.

data table · named destinations by period

destination	before	the break	turnaround push	after first closures
home brewing	65% (584)	63% (530)	57% (537)	51% (352)
local independents	15% (133)	14% (122)	17% (155)	15% (104)
Dutch Bros	3% (26)	7% (56)	9% (82)	14% (98)
Dunkin'	9% (78)	7% (60)	8% (72)	8% (55)
other chains	6% (53)	6% (54)	5% (47)	7% (50)
7 Brew	0% (0)	1% (9)	3% (25)	3% (22)
McDonald's	2% (20)	1% (11)	1% (11)	1% (9)
energy drinks	1% (6)	0% (2)	1% (8)	1% (4)

Dutch Bros is a regional story

Dutch Bros has about 1,225 shops against roughly 16,900 Starbucks stores in the US, so it cannot take in most of the customers Starbucks loses. Where it has shops, though, it matters. Among leavers whose region we could tell, Dutch Bros took 10% of named destinations in the West, Southwest and Texas (307 people) against 3% everywhere else (253 people). That gap holds up statistically.

“My espresso machine. I used to get at least two Starbucks lattes per week and now I get none.”

“Good one. I stopped getting Starbucks once I realized it was costing me $7 for my customized matcha latte. Now I buy matcha powder and oat milk from the grocery.”

“I stopped getting Starbucks and went to Dunkin…I don’t love their coffee but Starbucks is being ridiculous with prices, challenges and the change in star structure. Also hate the way they treat their employees so Dunkin for me”

What pushed them

We did not start from a list of complaints. We grouped posts by what they actually said, named each group from its own examples, and only then matched the groups to known issues. Recent posts count more than old ones: a post about thinking of leaving loses half its weight in three months, a post saying "I left" loses half in twelve (why complaints get a half-life).

Price is fading as the reason. Drink quality is taking its place. In dfmchn's analysis, price fell from 49% to 31% of the reasons Starbucks customers gave for leaving or cutting back, while drink quality and consistency rose from 26% to 37%. In the year since the first closures, drink quality is the most common reason people give. Labor disputes rose from 7% to about 11% after late 2023 and stayed there.

data table · reasons for leaving by period, with ranges

reason	before	the break	turnaround push	after first closures
price	49% (47% to 52%)	48% (46% to 50%)	41% (39% to 43%)	31% (29% to 34%)
drink quality and consistency	26% (24% to 28%)	24% (22% to 26%)	26% (24% to 28%)	37% (34% to 39%)
labor disputes	7% (6% to 9%)	10% (9% to 12%)	10% (9% to 12%)	11% (9% to 12%)
Rewards changes	7% (6% to 8%)	5% (4% to 6%)	7% (6% to 8%)	9% (8% to 10%)
other new reasons	5% (4% to 6%)	5% (4% to 6%)	7% (6% to 9%)	6% (5% to 7%)
menu changes	3% (2% to 4%)	2% (2% to 3%)	3% (3% to 4%)	3% (2% to 4%)
boycott and politics	0% (0% to 0%)	3% (2% to 3%)	2% (2% to 3%)	1% (1% to 2%)
loss of cafe feel	1% (1% to 2%)	1% (1% to 2%)	2% (1% to 3%)	2% (1% to 2%)
waits and mobile orders	1% (0% to 1%)	1% (0% to 1%)	1% (0% to 1%)	1% (0% to 1%)

Boycott and political reasons peaked at 3% of reasons during the break and have fallen since. We report that share as measured and do not characterize the positions people took. It barely changes when posts that looked coordinated are put back in. About 118 posts said they had stopped going without giving any reason; they sit in "other new reasons".

data table · boycott and political reasons, with and without coordinated posts

period	share, coordinated posts removed	share, coordinated posts kept	posts flagged as coordinated
before	0%	0%	3
the break	3%	3%	0
turnaround push	2%	2%	1
after first closures	1%	1%	0

“I switched from Starbucks to Dunkin once the prices jumped again. I love their cold brew with cold foam. I usually get a medium with 1 caramel and 2 French vanilla or a shaken espresso with the same swirls”

“I used to really enjoy Starbucks and would visit 3 times a week. However as costs kept going up I just couldn't justify the cost anymore. Now I go much less frequently about twice a month. There is only so much I am willing to pay for a coffee. =(”

Who came back

Talk about coming back rose to 0.29 per 1,000 mentions in the latest year, the highest of the four periods (it was 0.12 in 2022 to 2023). But the people coming back are few, and most of them did not come back because the thing that drove them away was fixed.

We followed 1,001 people who posted about Starbucks at least twice with dates we could place. 825 stayed away. 120 cut back and stayed cut back. Only 33 left and came back. The 33 were gone for about four months (the middle half were back within 2 to 11 months).

data table · what happened, and reasons for coming back by period

what happened (people with two or more dated posts)	people
stayed away	825
cut back and stayed cut back	120
left and came back	33
never really left	14
came back, leaving not seen	9

period	price or deal	new drink	service got better	dispute settled	habit or convenience	something else	did not say
before	11	5	10	1	44	24	21
the break	17	6	1	2	37	28	12
turnaround push	12	10	3	2	48	27	20
after first closures	12	15	10	0	51	23	18

Only about 15% of returning Starbucks customers who gave a reason said the problem that drove them away had been fixed, according to dfmchn. That is 4 of 27. The rest came back out of habit or convenience, or for reasons that had nothing to do with why they left. Leaving and coming back are different doors: fixing what went wrong is necessary, but it is not what brings people back. The return needs its own reason (why winning someone back costs more than keeping them).

The good news mostly comes from people who never left. Of the 1,451 positive customer posts in the latest year, only 8% came from people we had seen leave or cut back earlier.

Protein drinks, launched in late September 2025, show up in 4% of coming-back posts in the latest year, about the same as their 6% share of all customer posts. They are not pulling people back on their own.

“Haha, I had quit Starbucks recently, but the banana protein cold foam snatched me back in as a customer.”

“I tried a double on ice today after not having Starbs for a while and it tasted like chemicals. Really a shame that tens of thousands of dollars in equipment and hundreds of hours of training can’t produce a halfway decent espresso.”

“It actually is pretty good! I used to only make my own with tazo but reverted back to starbucks because it's easy”

What employees saw

Employees describe strain in their own forums: too few people on shift, cut hours, floods of mobile orders, drive-thru time targets. If that strain is what customers later feel as slow service and inconsistent drinks, it should show up before customers talk about leaving.

We counted strain posts per 1,000 employee posts each month and checked how well they line up with customer leaving talk at gaps of zero to nine months. At gaps of zero to eight months the link is weak or absent. At nine months, the longest gap we tested, it reaches 0.42.

This is a signal worth watching, not a proven early warning. It shows up only at the edge of what we tested, and when you check ten different gaps, one can look strong by luck. Longer gaps need testing before anyone relies on it. For what it is worth, employee strain has been climbing again since spring 2026.

data table · correlation by gap

gap in months	0	1	2	3	4	5	6	7	8	9
correlation	-0.12	-0.03	0.03	-0.09	0.03	0.18	0.03	0.21	0.20	0.42

Checking the talk against Starbucks' own numbers

Talk is not sales. But if the talk means anything, it should move in the same direction as the visit numbers Starbucks reports. Its filings show US comparable transactions (visits to stores open at least a year) falling for eight quarters in a row, through fiscal 2024 and 2025, and rising from the first quarter of fiscal 2026, which is October to December 2025. Starbucks' fiscal year ends in late September.

For each quarter we took leaving talk minus coming-back talk. The direction agrees with the filings in 9 of 11 quarters. The disagreement is at the turn: Starbucks' visits turned up in October to December 2025, but net leaving talk kept rising, peaked in January to March 2026, and only fell after that. The talk turned later than the visits did. The latest quarter, July to September 2026, is not complete.

data table · net leaving talk by fiscal quarter

fiscal quarter	months	net leaving talk per 1,000	reported US visits
FY2023-Q1	Oct to Dec 2022	1.52 (1.25 to 1.80)	no figure
FY2023-Q2	Jan to Mar 2023	2.41 (2.07 to 2.74)	no figure
FY2023-Q3	Apr to Jun 2023	1.26 (1.00 to 1.52)	no figure
FY2023-Q4	Jul to Sep 2023	1.13 (0.88 to 1.38)	no figure
FY2024-Q1	Oct to Dec 2023	2.82 (2.46 to 3.18)	fell
FY2024-Q2	Jan to Mar 2024	2.52 (2.17 to 2.86)	fell
FY2024-Q3	Apr to Jun 2024	3.43 (3.01 to 3.86)	fell
FY2024-Q4	Jul to Sep 2024	2.51 (2.14 to 2.88)	fell
FY2025-Q1	Oct to Dec 2024	2.10 (1.77 to 2.44)	fell
FY2025-Q2	Jan to Mar 2025	2.80 (2.44 to 3.17)	fell
FY2025-Q3	Apr to Jun 2025	2.38 (2.06 to 2.70)	fell
FY2025-Q4	Jul to Sep 2025	2.56 (2.22 to 2.90)	fell
FY2026-Q1	Oct to Dec 2025	2.66 (2.27 to 3.04)	rose
FY2026-Q2	Jan to Mar 2026	3.68 (3.18 to 4.18)	rose
FY2026-Q3	Apr to Jun 2026	2.23 (1.81 to 2.66)	rose
FY2026-Q4	Jul to Sep 2026	1.88 (1.50 to 2.26)	no figure	not complete