Demand Forecasting for Michigan Craft Breweries

Who this is for: Michigan breweries running a taproom alongside a distributed book, deciding batch sizes eight to twelve weeks before the beer reaches a shelf.

The Brewers Association reported that US craft production fell 4% in 2025, to 22,034,000 barrels. Most coverage stopped there. The more useful number in the same report is the split: 60% of breweries reported declines, 39% reported growth, and 1% held steady. A 4% decline is not what happened to any particular brewery. It is the midpoint of two very different years happening at once.

Michigan breweries are running that experiment in unusually favourable conditions. The East North Central census division, which includes Michigan, posted the strongest regional trend in the country at +0.4%, the only division to grow while national volume fell. Michigan itself has 410 craft breweries, eighth most of any state, and the Brewers Association put the state's 2025 craft brewing economic impact at 1,655 million dollars.

Put those together and the conclusion is uncomfortable. If a Michigan brewery lost volume in 2025, the regional market did not take it. Another brand did.

The split that matters is by business model

The 2025 data separates cleanly by how a brewery reaches the drinker:

• Brewpubs: −1.7%
• Taprooms: −3.9%
• Regional breweries: −4.1%
• Microbreweries: −8.9%

Microbreweries, the model built on wholesaler distribution, lost more than five times as much volume as brewpubs. It is tempting to read that as consumers preferring taprooms, and some of it is. But there is a mechanical explanation that matters more for planning.

A brewpub knows what sold last night. The person who drank the beer bought it from the brewery. The feedback loop between demand and the next production decision is roughly one day long.

A microbrewery sells to a wholesaler. That transaction is a shipment, not a sale to a drinker. The actual sale happens days or weeks later at a retailer, and the brewery finds out when a depletion report arrives. Every production decision in that model is made against a picture of demand that is already stale, and the staleness compounds: brew against last month's depletion, and the beer that results arrives in market a further eight weeks on.

That is a forecasting problem, not a market problem. It is also where a Michigan brewery running both channels is most exposed, because taproom sales look immediate and healthy while the distributed book quietly drifts.

Michigan's self-distribution rule creates two different data sets

Michigan allows a Micro Brewer licensee to self-distribute to retailers provided it sells no more than 2,000 barrels a year. Beer sold for consumption in its own tasting room does not count toward that threshold, and all brands of the licensee are combined when calculating it, including brands brewed outside the state.

The constraint that shapes planning, though, is the territorial one: a licensee cannot self-distribute into a sales territory where it has already granted a wholesaler exclusive rights to those brands.

For a growing Michigan brewery, that means the same SKU frequently exists in two channels with two entirely different data qualities. In self-distributed territory, the brewery generates its own sales records and knows exactly what each account bought and when. In wholesaler territory, it sees whatever the depletion report contains, at whatever cadence the wholesaler sends it.

Most breweries forecast these together, because the beer is the same beer. That is a mistake. The self-distributed data is a leading indicator with a one-day lag; the wholesale data is a lagging indicator with a two to six week lag. Averaged into one series, the accurate half gets diluted by the delayed half, and the resulting forecast is worse than either channel alone would produce.

The better approach is to forecast the channels separately and use the self-distributed accounts as an early read on how a style or format is performing before the wholesale numbers confirm it. If a new hazy IPA is decelerating in your own accounts in week three, that is real information about week seven in wholesaler territory.

What a forecast has to produce before the brew schedule is set

A forecast is only worth building if it changes a decision, and in a brewery the decision is batch size and timing. Three outputs are enough:

Velocity by SKU, package, and territory. Not total cases moved, but rate of movement, calculated over a rolling window so a single heavy order does not read as growth. Package matters as much as brand. A brand can be healthy in 16 oz four-packs and failing in 12 oz twelve-packs, and the shelf decision the retailer makes is at package level.

Where committed volume exceeds projected depletion. This is the number that prevents beer aging in a wholesaler's warehouse. If 900 cases are already brewed or committed and the projection says 500 will deplete inside the freshness window, the gap is visible before it becomes a discount.

Where pre-sell commitments exceed planned output. The opposite failure, and the more expensive one in a market reallocating shelf space between survivors. Nationally, openings fell to 300 in 2025 from 518 in 2024, while 481 breweries closed. That space is being redistributed among the brands still standing, and a brand that cannot supply a placement it won does not usually get offered it twice.

Where Vintaflow fits

Vintaflow forecasts demand using seasonality, trends, and historical sales performance that the brewery supplies as xlsx or csv. That matters here because wholesaler depletion exports already arrive as spreadsheets, and no ERP replacement is required to start using them. The brewery supplies the files; Vintaflow does not pull data out of wholesaler systems on its own.

From that data, Vintaflow provides account-level performance and inventory dashboards, so depletion reads by SKU, package format, and account rather than as one monthly total per brand. Where a wholesaler participates directly, Vintaflow can report inventory and sales performance between partners, which replaces the familiar argument about whose spreadsheet is right with a single shared view.

What it does not do is worth stating plainly. Vintaflow does not receive automatic feeds from wholesaler systems, does not assign code dates, and does not schedule production. It produces the demand picture; the brewer still decides what to brew.

To pressure-test a forecast against your own depletion history, book a 30-minute conversation and bring one wholesaler export and one season of self-distributed sales.

Practical steps

1. Separate the channels. Build one series for self-distributed accounts and one for each wholesaler. Do not average them until you have a reason to.
2. Convert shipments to depletion. If you are currently planning from what you shipped, stop. Request depletion by SKU and account from each wholesaler, weekly if they will provide it.
3. Fix the window. Choose a rolling velocity window that matches your order pattern, commonly 30, 60, and 90 days, and read all three so a single large order is visible as an anomaly rather than a trend.
4. Set the review before the brew meeting. A forecast reviewed after the schedule is locked is a report, not a decision input.
5. Check package mix separately from brand. Run the same velocity calculation at package level and compare. Divergence between formats is usually the earliest warning that a shelf set is about to change.
6. Watch the 2,000-barrel line. If self-distribution is approaching the threshold, the transition to a wholesaler in that territory will change your data quality overnight. Plan the forecasting change at the same time as the commercial one.

The 39% of breweries that grew in 2025 did not find a different market. They were in the same contracting category, frequently in the same states, selling to the same tightening retailers. What separated them was knowing which of their own SKUs was working, in which format, in which territory, early enough to act on it.

If you want to talk through how your current depletion reporting would translate into a forecast, contact Vintaflow with a description of your channel split and reporting cadence.