Pour & Loss

Gallup published its annual read on American drinking on August 20. Fifty-four percent of adults say they drink alcohol, unchanged from last year and the lowest reading in a series that begins in 1939. Nearly every write-up I saw led with that number and reached for the same explanation. The old drinkers are leaving and the young ones are not arriving.

Three scrolls down the same page is a table that does not support the explanation.

Gallup sets 2023 against 2026 by subgroup. Among adults 18 to 34, the drinking rate fell from 58 percent to 50. Among 35 to 54, from 66 to 58. Among 55 and older, from 61 to 55. Eight points, eight points, six points, against a national decline of eight, from 62 percent to 54. Gallup's own summary is that the decline shows up across all age groups.

Generational replacement does not make that shape. It is slow and it shows at the young end first. A cohort effect looks like the youngest band falling while the middle and older bands hold, with the national figure drifting down a fraction of a point a year as the population reweights. Instead all three bands moved together inside thirty-six months. That is something acting on everybody at once.

The something is not hard to find. Gallup has asked since 2001 whether one or two drinks a day is bad for your health. Twenty-seven percent said yes in 2001, and the number sat in the mid-twenties for more than a decade: 28 percent in 2015, 26 in 2016, 28 in 2018. Then 39 in 2023, 45 in 2024, 53 in 2025, and 51 this year. Belief and behavior bend at the same place.

Split that series by age and the sequence is the argument. In 2018, 34 percent of adults 18 to 34 called moderate drinking bad for you, against 26 percent of both older bands. By 2023 the young had moved 18 points, to 52 percent, while the 55-and-older band had moved 3, to 29. Gallup's phrasing that year was that the increase was owed largely to young adults, and most people stopped reading there.

Then the older bands caught up. Adults 55 and older went to 39 percent in 2024 and 48 in 2025. The 35-to-54 band went to 50. Nineteen points in two years at the top of the age range, after three points in the five years before. Belief did not skip the older drinker. It reached him about five years late, and the behavior series turns in 2023, precisely when it arrived in the bands that had not yet moved.

What matters more than the sequence is that belief has stopped moving. Fifty-three percent last year against 51 this year, on a sample of 1,200 with a margin of four points, is no movement. Eighty-six percent say they have heard at least something about the research, up from 79 percent in 2024, and no group moved significantly in the past year. The message has reached nearly everyone it will reach.

That matters for the shape of the next five years. A category losing customers to a belief shift that has saturated is in a different position from one losing them to a population that is dying. The first stabilizes at a lower level and can be sold to. The second keeps going.

The 2026 BMO Wine Market Report, published May 12, carries two panels built from the NIH and SAMHSA annual drug and health surveys that I have not seen discussed anywhere in the trade. The first sets 2016 against 2025 for Americans aged 21 to 25.

The population of that band is flat. Twenty-two point nine million then, twenty-two point eight million now. The share who drink fell from 83 percent to 77. Reported servings per week fell from 5.9 to 4.3. Total servings per week for the cohort fell from 112 million to 76 million, a drop of 32.5 percent.

Run the two effects apart. Hold participation at 83 percent and let servings fall to 4.3, and the cohort produces about 81 million servings a week. Hold servings at 5.9 and let participation fall to 77, and it produces about 104 million. The observed figure is 76 million. Better than four-fifths of the loss is intensity and about a fifth is participation, and the paths overlap, so they sum to more than the total.

So the cohort this industry has spent three years calling sober is the same number of people, drinking materially less each. A real loss, and a different kind from the one the industry keeps describing.

Now the other end of the age range, from the companion panel, and it is not a snapshot of two age bands today. BMO follows the Boomer cohort itself across nine years, which means it follows them across an age line. Boomer population fell from 74.1 million in 2016 to 65.4 million in 2025, down 11.7 percent, and the average age of the survivors rose from 60.3 to 69. In 2016, inside the 50-to-64 band, 67 percent drank and averaged 5.3 servings a week. In 2025, inside the 65-and-over band, 56 percent drink and average 4.5. Total servings fell from 263 million a week to 165 million, down 37.4 percent. Both ends reconcile: 74.1 at 67 percent at 5.3 is 263, and 65.4 at 56 percent at 4.5 is 165.

Two cohorts, almost the same headline decline, two engines driving it. At the young end, the same people pouring less. At the old end, fewer people every year, and those who remain pouring less, and then at some point not at all. The first is a habit. The second is a census.

A habit answers to price and occasion. A census answers to nothing. Every commercial decision downstream of this diagnosis forks on which one you are looking at, and most of the plans I have read this year are written as though it were all census.

The Wine Market Council released its 2025 U.S. Wine Consumer Benchmark Segmentation Survey on December 4, built on nearly 5,000 adults over 21, weighted to census. Two figures traveled everywhere. Millennials are now the largest wine-drinking generation in America. And the country has about 76 million adult wine consumers, roughly nine million fewer than in 2023, with the share of legal-drinking-age Americans who drink wine at least every two to three months down from 34 percent to 29.

Both figures are accurate. Read together they say something other than what the headlines said.

The full generational series sits in the appendix of the 2026 BMO report, where Christian Miller aligned the Council's historical waves. In 2023 the shares were Millennials 29.7 percent, Gen X 24.6, Boomers 34.1, Gen Z 6.0, older 5.6. In 2025 they are Millennials 30.0, Gen X 26.0, Boomers 26.0, Gen Z 14.0, older 4.0.

Now put those shares against the base. Seventy-six million wine consumers today, and 76 plus the Council's nine gives roughly 85 million in 2023.

Millennials go from 29.7 percent of 85 million, which is 25.2 million people, to 30.0 percent of 76 million, which is 22.8 million. Down about 2.4 million.

Gen X: 20.9 million down to 19.8 million, a loss of about 1.1 million. Gen Z runs the other way, from 5.1 million up to 10.6 million.

Boomers fall from 29.0 million to 19.8 million. That is 9.2 million people gone in twenty-four months.

The older cohorts drop from 4.8 million to 3.0 million.

Those five changes net to nine million, the Council's own figure. That checks my shares and my denominators. It is not a discovery: two sets of shares that each sum to 100, applied to two stated totals, can only net to the difference between them. What it shows is where the nine million sat, and it was not where the coverage put it.

Gen Z is the only cohort in the country that added wine drinkers over the past two years. Millennials were crowned the largest wine-drinking generation by the same survey that shows them shedding two and a half million people. Their share rose because the room emptied faster than they walked out of it.

Be exact about this, because the share language will keep circulating for another year. A share is a ratio and the denominator lost nine million. A cohort can gain share, lose millions of people, and collect a favorable headline in the same week.

Three source notes, because someone will check. The Council's December release gives Boomers at 32 percent for 2023 and Millennials at 31 for 2025, where the appendix gives 34.1 and 30.0 for the same waves. I use the appendix because its waves sum to 100 and the release's 2025 figures sum to 101. On the release's numbers the Boomer loss is about seven and a half million rather than nine, which changes the size and not the direction. The two also differ on Gen Z in 2023, 6.0 percent against 9, and Gen Z gains on either set. Last, the Gen Z move is partly definitional: that cohort had six birth years over 21 in December 2023 and eight by December 2025. Roughly a third of the gain is the calendar. The rest is not.

One more figure from the BMO report, and it is the one that pays the bills. The group that drinks wine two or more times a week peaked near 37 million people in 2012, was about 30 million by 2018, and stands near 24 million today. Wine drinkers as a share of the legal-drinking-age population were 37 percent in 2018 and are 29 now, and the once-a-week-or-more group went from 19.4 percent of that population to 16.0. The category lost its best customers, and it lost most of them at the top of the age range.

Here is the series nobody quotes. Gen X was 19.4 percent of American wine consumers in 2017 and 19.4 in 2018. Then 20.7 in 2019, 23.8 in 2021, 24.6 in 2023, 26.0 in 2025. Six and a half points across six readings, never once giving any back, straight through the worst stretch this category has had since the 1980s.

A word on the base year, since that is the check I run on everybody else. The appendix carries the series back to 2010, and the rows before 2017 are printed in a layout I could not read with confidence, so I am not quoting them. The run above starts where the table becomes unambiguous, not where it flatters the argument.

No other cohort in the table does it. Millennials peaked at 36.5 percent in 2018 and have given back six and a half points since. Boomers gave back eight points in the last two years alone. Gen Z is rising off a base that grows by definition.

In headcount, Gen X lost about 1.1 million wine drinkers out of the nine million the category lost. Roughly one-eighth of the damage, from a cohort now tied with Boomers at 26 percent and sitting four points behind Millennials.

A second instrument points the same way. In 2014, 38 percent of Gallup's drinkers aged 55 and older named wine as what they drink most often, against 32 percent who named beer. In 2026 that band is 37 percent wine. In those twelve years the leading third of Gen X aged into it, and conversion at the top of the age range is holding at the rate it ran for the generation before. Only part of the cohort has crossed the line, so that is an observation and not a proof.

On the driver doing the most damage at the young end, Gen X does not behave like the young. The Council found 11 percent of Gen X changed the type or amount of alcohol they drink to improve mood, sleep or energy in the past year, against 24 percent of Gen Z, 21 of Millennials and 5 of Boomers. On wellness-driven moderation, Gen X sits with the Boomers.

The standing objection is that Gen X is small. It is the smallest of the three large cohorts, and that has been the reason to skip past them for twenty years. It is also the wrong test. Population tells you a cohort's ceiling ten years out. It tells you nothing about the return on a mailing this quarter, which depends on how many are in your file and how they convert when you write to them.

For about five years, since 2021, I have been giving owners the same three-part read. Boomers are going to keep shrinking as wine buyers. Gen X is still buying, and you cannot write them off for being smaller than the cohorts on either side. And the cohorts underneath Gen X will come around to wine the way every cohort before them did, once they reach the life stages that produce wine drinkers.

The first two hold up better than I stated them. I said Boomers were declining. The number is 9.2 million people in twenty-four months, larger than any other movement in the table and larger than I was describing it. I said Gen X still mattered. They have held or gained share in every wave for nine years.

The third part is where I was wrong, and wrong in a particular half. Millennials are 30 to 45 today. That is the life stage. Marriage, homeownership, children, the first real earnings. It is the window I told people to wait for. Their share of American wine consumers peaked at 36.5 percent in 2018, three years before I started giving this advice, and had already fallen to 30.1 by 2021. It stands at 30.0 today, and they have shed two and a half million people in the last two years. The turn I told owners to wait for had already run the other way, and it was in the table when I said it. They arrived on schedule and did not convert.

I had been treating the delay as a financing problem, on the reasoning that a cohort carrying student debt and renting into its thirties would buy wine once the balance sheet turned. The Council's survey says a preference problem sits underneath it. Thirty-five percent of Millennials say they do not like the taste of wine. Thirty-nine percent prefer the taste of something else. Those do not resolve when somebody closes on a house.

The Gen Z half is holding so far. They are the only cohort in the table adding wine drinkers, and they added them in a window when every other cohort lost. BMO also reports that 45 percent of Americans between 21 and 28 are drinking more wine, against 38 percent of Millennials, though the report does not say whether that is behavior or stated intent. The headcount is the part I trust, and on it I would give that half of the advice again tomorrow.

What the wrong half cost was time rather than money. An owner who took it held a club structure and a release cadence in place for a cohort that was going to show up, and spent five years not building for the cohort already in the file.

The Council found one more thing that has nothing to do with who drinks and everything to do with when. More than 40 percent of consumers now say wine is what makes an occasion feel special. In earlier waves the leading answer was relaxing at home. Liz Thach's summary was that weekday dinner and relaxing-at-home occasions have both declined.

That is a loss no cohort table will show you, because the person is still counted. They still drink wine. They drink it a handful of times a year at things that feel like events, and the rest of the calendar the bottle sits in the rack behind other bottles bought under the same assumption.

The national numbers agree on direction. Gallup's drinkers now report an average of 3.2 drinks in the past week, against 3.9 across the five readings before the decline began. The people who stayed are lighter users than the people who were here five years ago.

Follow that into a club P&L and it does specific damage. A twelve-bottle commitment sold to a household that now has four wine occasions a year is not a retention problem that shows up at renewal. It is an inventory problem sitting in the member's own closet. They cancel with thirty-one bottles on the shelf and a shipment coming. The tell is the member who opens every email, comes to the pickup party, and has not placed an incremental order in eighteen months.

That is the link between the two halves of this issue. The Boomer losses are a census and they will happen on schedule. The Gen X and Millennial households are still there and still counted, and what they lost is the weeknight. Those are the customers a winery can do something about between now and the end of the year.

Three moves before the year closes. None of them needs capital.

Sort the buyer file by birth year instead of by tenure. Count the band born between 1965 and 1980 against the band born before 1965, by headcount and by trailing twelve-month spend, and set that split next to the national one, where the older band is losing people fast and the middle band is close to holding. If your file leans older than the country does, part of next year's decline is already booked and no campaign work will unbook it. That tells you how much of your 2027 plan is a forecast and how much is arithmetic you have already lost.

Pull last year's lapsed members and sort the lapses two ways. First by age at lapse. Second by whether the member's last three purchases landed on a release or between releases. If they cluster by age, you are losing people and the repair is acquisition, which is slow and expensive. If they cluster among members who still open the mail and still show up but stopped buying between releases, you lost the weeknight, and the repair is a cheaper bottle, a smaller format, or a reason to open one on a Wednesday. Most wineries I see are funding the first while suffering the second.

Split one release mailing. Write one version for the band born between 1965 and 1980, send the standard version to everyone else, and measure click and conversion rather than opens. The flat aggregate open rate everyone complains about is an average taken across cohorts that want different things from the same bottle, and an average built that way describes nobody. If the segmented version pulls ahead, you have found your segment. If not, you learned that for the cost of one send.