How much depletion history does a useful forecast need?

Seasonality is the constraint. A model that has to separate a genuine trend from an annual pattern needs at least two full cycles, so roughly 24 months of history by SKU is where vintage and seasonal effects become readable. With around 12 months you can still get a usable short-horizon velocity read, but you cannot yet distinguish a soft quarter from a seasonal trough. Where a SKU has less history than that, the honest answer is to treat its forecast as provisional rather than to substitute a category average and present it as a prediction.

Can the system handle the three-tier reporting lag common in control states?

Yes. Vintaflow uses sales-order and invoice data — which is available immediately when orders are placed — as a leading indicator of depletion, rather than waiting for state-reported depletion data that can lag by four to six weeks. This approach is particularly valuable in control state markets where official depletion reporting delays are longest.

Where does the working capital actually sit in a distributor book?

In the tail, and that is what aggregate reporting hides. A book of 200 to 500 SKUs will usually have a long list of slow movers that individually look too small to matter and collectively tie up a disproportionate share of capital. Reporting at brand or supplier level averages them away. The practical first step is not a target number for turns improvement; it is producing a ranked list of SKUs by days of cover against current velocity, which is usually enough on its own to show where the money is parked.

What level of SKU segmentation is most useful for California wine demand forecasting?

The more granularly you can segment your demand model, the more actionable the output. For California wine distributors, tracking velocity separately by variety, vintage, price tier, and channel — off-premise chain, independent retail, on-premise — captures the material differences that aggregate reporting misses. Napa, Sonoma, and Central Valley producers have very different velocity profiles at the same distributor, and allocation decisions are much cleaner when those profiles are tracked separately rather than blended into a single SKU average.

Wine & Spirits · Demand Forecasting

Demand Forecasting for California Wine Distributors

California winery inventory sat at roughly 19 months of forward supply at the close of 2025, down from above 20 months in 2024 and still above historical norms. Against a US wine market that shipped about 329 million cases last year — its lowest volume since 2016 — distributors who are still running manual depletion tracking and spreadsheet-based reorder systems are bleeding working capital on cases that will not move at full margin. The math has changed. Forecasting has to change with it.

Key Challenges

Depletion data from three-tier retail accounts arrives weeks late or not at all, so reorder decisions at the distributor level are based on last month's sell-through, not today's pace. In a market declining 2% annually, that lag means you are buying to last year's velocity.
Vintage allocation windows compress every year as wineries clear older stock before new releases. A distributor who cannot model forward demand by SKU has no credible basis for committing to allocation quantities, and routinely either under-commits and loses the premium allocation or over-commits and absorbs the unsold surplus.
Control state and on-premise account mix shifts are opaque in real time. A restaurant closure or chain planogram reset can silently kill 300 to 400 cases of on-premise velocity. Without sell-through visibility at the account level, that loss does not surface until the next quarterly review.
The bottom quartile of US wineries posted a -10.2% sales decline and -10.5% operating margin in 2025. Their distress creates deal flow at the winery tier, but distributors without forward demand models cannot evaluate whether taking on a distressed SKU at a lower FOB price will actually improve their turns or just add to the inventory problem.

Industry Data

Metric	2024	2025	Change
US wine market volume (cases)	335.9M	329M	-2.0%
California grape crush (tons, 2025 preliminary)	2,942,673	2,759,202	-6.2%
Average price per ton, all varieties	$1,016.96 (approx.)	$978.60	-3.8%
California winery inventory (months supply)	~20+ months	~19 months	~1-2 months

Source: SVB State of the US Wine Industry Report 2026; USDA NASS California Grape Crush 2025 preliminary. Months-of-supply figure is trade commentary and is described as such in the body rather than presented as a measured statistic. (2026)

The Inventory Math California Distributors Cannot Ignore

Trade commentary through early 2026 put California winery inventory at roughly 19 months of forward supply, improved from above 20 months a year earlier but still heavy. Treat that as a directional read rather than a measured statistic: months-of-supply estimates are brokerage commentary, not a published dataset, and they move with whichever depletion assumption is used. What is measured, and what matters more, is the direction of the two series either side of it.

At the same time, the SVB 2026 State of the US Wine Industry Report puts total US wine shipments at about 329 million cases for 2025, a 2.0% decline from 335.9 million cases the prior year. These are not independent problems. Excess inventory is a direct function of the gap between supply commitments made at FOB and actual depletion rates at retail and on-premise accounts.

For a California distributor running 200 to 500 SKUs, this means every reorder decision carries meaningful risk. The SKU you bought at 400 cases six months ago based on last year's velocity may now be moving at 60% of that pace because a regional chain has reset its planogram, a restaurant account has closed, or a competitor brand at a lower price point has captured the shelf position. Manual tracking does not surface these shifts in time to prevent surplus buildup.

Book a 30-minute conversation to see how depletion-velocity modelling works with your specific account mix.

What the 2025 Crush Data Tells Distributors About 2026 Supply

The USDA NASS California Grape Crush Report for 2025, preliminary figures released in March 2026, shows a total crush of 2,759,202 tons, down 6.2% from 2,942,673 tons in 2024. This reduction in incoming supply is the market's self-correcting mechanism: smaller harvests mean fewer cases entering the distribution channel, which helps reduce the inventory overhang over time.

The average price per ton across all varieties fell to $978.60, down 3.8% from the prior year. Red wine grapes averaged $1,280.63 per ton (down 4.4%), while white wine grapes averaged $707.12 per ton (down 0.9%). These price signals reflect the degree to which the market is still working through red varietal oversupply in particular, a pattern that has persisted for three consecutive harvests.

For distributors, the practical implication is this: the supply correction is underway on the vineyard side, but the pipeline clearing at the winery and distributor tier will take longer. SVB expects total market demand to bottom in 2027 through 2028 before returning to modest growth. That is two more years of operating in a market where precise demand forecasting is the difference between maintaining margin and absorbing write-downs on excess stock.

What Winery Performance Data Means for Distributor Relationships

The top quartile of US wineries grew sales 8% in 2025 with an 11.9% operating income margin. The bottom quartile saw a -10.2% decline and -10.5% operating margin. The spread between those outcomes is not primarily explained by brand quality. It is explained by how well each winery managed its inventory, its depletion data, and its channel allocation decisions. Distributors who give their winery partners accurate, timely depletion data are better positioned to earn and keep the top-tier allocations.

How Demand Forecasting Works in the Three-Tier System

The structural challenge in California wine distribution is that depletion data — the sell-through from retailer to consumer — is the most important signal for replenishment decisions, but it is also the slowest to arrive. Retailers report sales to distributors on weekly or monthly cycles, state-reported depletion data can lag by four to six weeks, and NABCA reports in control states take even longer.

A demand forecasting system built for the three-tier environment addresses this by using leading indicators: sales orders placed by retailers (which reflect buyer intent before the product moves to the consumer), invoice cadence by account, and historical depletion seasonality by SKU. These signals are available in near real time and provide a statistically reliable proxy for actual sell-through velocity.

Applied to a distributor's book of business, this creates a model that can do several things a spreadsheet cannot. It can flag which SKUs are trending below their forward demand target with enough lead time to pause reorders before excess builds further. It can model allocation scenarios by vintage and variety, showing the working capital impact of different commitment levels at the next portfolio tasting. And it can segment performance by channel — off-premise chain, independent retail, on-premise — so a shift in one channel does not contaminate the aggregate picture.

The goal is not to eliminate all inventory risk. In the three-tier system, holding appropriate inventory is part of the distributor's value proposition to the winery. The goal is to hold the right inventory: enough to service accounts without shortfalls, not so much that slow-moving SKUs tie up capital that should be turning faster.

To see the ranked days-of-cover view against your own account mix, book a 30-minute conversation.

Vintage Allocation Strategy in a Declining Market

The 2025 SVB report identified vintage allocation as one of the highest-leverage decisions a distributor makes each year. When the market is growing, over-committing to allocation has modest downside: you sell through eventually, perhaps at a slight discount. When the market is declining, over-committing at the allocation stage means carrying cases at full landed cost against a demand curve that will not support the price needed to protect margin.

A demand forecasting system changes how allocation decisions get made. Instead of committing based on last vintage's sell-through, a distributor can model forward demand at the SKU level using its own current velocity data and account-level trends. The system can also model the impact of a price increase or decrease on projected sell-through, which is critical when negotiating FOB terms with a winery that is trying to hold price in a softening market.

The distributors who are outperforming in the current environment are not making better guesses. They are making decisions grounded in SKU-level depletion data, and they are making those decisions faster than competitors who are still waiting for monthly reports to tell them what happened six weeks ago.

Where to Start if the Book Is Already Heavy

A distributor sitting on surplus does not need a forecasting project before it can act. It needs a ranked list, and the inputs for that list already exist in the order history.

Rank every SKU by days of cover at current velocity, not by case count. A 400-case SKU turning in three weeks is not the problem; a 60-case SKU with fourteen months of cover is, and it will never surface in a report sorted by volume or revenue. Do this at SKU and vintage level, because a 2022 and a 2023 of the same wine are different products with different remaining runway.

Then split the tail into two piles. Some slow movers are structurally slow and always were: allocation obligations, portfolio-completeness SKUs, wines carried to keep a winery relationship. Those are a cost of doing business and the decision is whether the relationship still justifies it. The rest decelerated, and each one decelerated for a reason that is usually visible in the account data: a chain reset, a closed on-premise account, a competitor at a lower price point taking the facing. Those are recoverable, and they are where reallocation or targeted promotion actually returns something.

Finally, take the ranked list into the next allocation conversation rather than the next quarterly review. The commitment that creates surplus is made months before the surplus appears, and a winery partner presented with SKU-level sell-through has a materially different conversation with you than one hearing a request to cut the number.

How Vintaflow helps

Demand Forecasting and Analytics

Vintaflow forecasts demand using seasonality, trends, and historical sales performance that the distributor supplies as xlsx or csv, and provides account-level performance and inventory dashboards so velocity can be read by SKU, channel, and account rather than as one blended number. Reorder points and target inventory are calculated from demand and shipping constraints, and inventory alerts and replenishment suggestions surface SKUs whose cover is running ahead of demand. Selected inventory and sales performance can be shared with participating winery partners. No ERP replacement is required. The distributor supplies the files; there is no automatic data feed from a retailer or distributor system.

Talk through this challenge Prefer to send a message?

Frequently Asked Questions

How much depletion history does a useful forecast need?: Seasonality is the constraint. A model that has to separate a genuine trend from an annual pattern needs at least two full cycles, so roughly 24 months of history by SKU is where vintage and seasonal effects become readable. With around 12 months you can still get a usable short-horizon velocity read, but you cannot yet distinguish a soft quarter from a seasonal trough. Where a SKU has less history than that, the honest answer is to treat its forecast as provisional rather than to substitute a category average and present it as a prediction.
Can the system handle the three-tier reporting lag common in control states?: Yes. Vintaflow uses sales-order and invoice data — which is available immediately when orders are placed — as a leading indicator of depletion, rather than waiting for state-reported depletion data that can lag by four to six weeks. This approach is particularly valuable in control state markets where official depletion reporting delays are longest.
Where does the working capital actually sit in a distributor book?: In the tail, and that is what aggregate reporting hides. A book of 200 to 500 SKUs will usually have a long list of slow movers that individually look too small to matter and collectively tie up a disproportionate share of capital. Reporting at brand or supplier level averages them away. The practical first step is not a target number for turns improvement; it is producing a ranked list of SKUs by days of cover against current velocity, which is usually enough on its own to show where the money is parked.
What level of SKU segmentation is most useful for California wine demand forecasting?: The more granularly you can segment your demand model, the more actionable the output. For California wine distributors, tracking velocity separately by variety, vintage, price tier, and channel — off-premise chain, independent retail, on-premise — captures the material differences that aggregate reporting misses. Napa, Sonoma, and Central Valley producers have very different velocity profiles at the same distributor, and allocation decisions are much cleaner when those profiles are tracked separately rather than blended into a single SKU average.

Sources

Last updated: August 5, 2026