Importers: forecast demand with data (without drowning in sp

By Équipe EtOH · Published on 13 Nov 2026

Learn how importers can forecast demand using the data they already have: key signals, a lightweight model, lead time integration, examples, and a checklist.

Importing is a delayed equation: you buy today, you sell in 3 to 9 months—sometimes longer. In that window, the market moves, prices shift, and your customers change priorities. The good news: you already have enough data to anticipate demand—provided you structure it and turn it into decisions.

1) Start with a useful data foundation (not “all the data”)

The goal isn’t to build a data lake. It’s to make 20% of the data reliable—the 20% that explains 80% of demand variance.

To collect (minimum viable):

• Historical sales: volumes (cases), revenue, average price, discounting, by SKU (wine + vintage + format), by customer and channel.
• Availability / inventory: importer stock, stock at key customers (if shared), backorders, estimated arrival dates.
• Commercial calendar: planned promotions, wine fairs, local holidays, on-trade trade shows, account openings/closures.
• Actual lead time: time from supplier order → warehouse arrival → on-sale. Measure the distribution (min/median/max), not just an average.
• Price & FX: ex-cellar price, freight, duties, exchange rate, and selling price by channel.

Actionable tip: start with 12 months of history if you don’t have more. Consistency is what matters: same units, same product codes, same discount rules.

Realistic example:

• 250 active SKUs, 60 customers, 3 channels (independent wine shops, on-trade, retail).
• You won’t analyze 250 SKUs at once: prioritize the 50 SKUs that represent 70–80% of volume.

2) Segment to forecast: SKUs, customers, channels… and jobs to be done

A single global forecast “all references combined” hides signals. Demand is managed through stable segments.

Simple (effective) segmentation:

1. By channel: retail (volumes + promo), independent wine shops (mix + newness), on-trade (velocity + wine list).
2. By price tier (selling price): entry-level, core range, premium.
3. By SKU role:

- Core: steady sales, low volatility.

- Seasonal: peaks (rosé/summer, sparkling/holidays).

- New launch: no reliable history.

- Long tail: low velocity.

Actionable tip: apply an ABC rule by segment (not across the whole portfolio).

• A: top 20% SKUs = 70–80% volume
• B: 30% SKUs = 15–25% volume
• C: 50% SKUs = 0–10% volume

You’ll focus forecasting effort on A, do “reasonable” steering on B, and run a minimal-stock/opportunistic ordering strategy on C.

Quantified example:

• Retail segment: 15 A SKUs (mostly IGP + well-known brands) represent 75% of retail volume.
• On-trade segment: 10 A SKUs (by-the-glass + wine list) represent 68% of on-trade volume.

3) Build a lightweight forecast that works: baseline + adjustments

You don’t need a complex model to improve accuracy. A robust approach:

Step A — Baseline by SKU (moving average)

For each A/B SKU:

• Baseline = average of the last 8 to 12 weeks (or last 3 months), excluding out-of-stock weeks.
• If seasonality is strong: use the same quarter last year, then adjust.

Step B — Business adjustments (the “drivers”)

Add simple coefficients:

• Planned promo: +20% to +80% depending on channel/scale.
• Price change: rough elasticity (e.g., +5% price ⇒ -2% volume on core-range retail, to be calibrated).
• Account wins/losses: expected volumes per customer.
• Upcoming stockouts: mechanical reduction if supply can’t keep up.

Step C — Integrate lead time and decide replenishment

A forecast without lead time is unusable in import.

• Target stock coverage = forecast sales over (lead time + buffer)
• Buffer: 2 to 6 weeks depending on volatility and supplier reliability.

Full example (realistic):

• SKU: Rosé 2025 – 75 cl
• Baseline: 1 200 cases/month (3-month avg excluding stockouts)
• Summer season (June–August): +40%
• Independent wine shop promo in July: +25% (on 50% of volume)
• Median lead time: 10 weeks, max: 14 weeks
• Goal: avoid a stockout in July.

Calculation:

• July forecast = 1 200 × 1,40 = 1 680 cases
• Promo effect (on 50% of sales) ≈ 1 680 × 0,25 × 0,5 = +210 cases
• Estimated July demand ≈ 1 890 cases

Stock decision:

• Coverage = 10 weeks + 4 weeks buffer = 14 weeks ≈ 3,5 months
• Need = 1 200 cases/month × 3,5 = 4 200 cases (excluding peak)
• Adjust with seasonality: if your window covers June–July–August, include the peak over the period.

Result: you move from gut-feel replenishment to an order you can justify and defend with the supplier and your teams.

4) Spot weak signals before your competitors

Anticipating demand isn’t just extending the past. It’s detecting inflection points early.

Internal signals (quick to capture):

• Quote-to-order conversion rate (by SKU/channel)
• Reorder rate (customers who buy again within 60 days)
• Substitution: when a customer replaces one SKU with another after a stockout
• Discounts requested: a sudden increase = price friction

External signals (simple):

• Search trends (Google Trends) for “organic rosé”, “non-alcoholic”, “Prosecco”, etc.
• Public on-trade/tourism data (seasonal peaks by region)
• Grocery retail promotional activity (leaflets, fairs)

Actionable tip: a weekly “alert dashboard”

Create 5 indicators, thresholds, and associated actions:

• Quote conversion < 25% over 4 weeks ⇒ review price/value story or target different accounts
• Reorder rate < 30% ⇒ product fit, perceived quality, or velocity issue
• Stockout > 7 days on an A SKU ⇒ trigger substitution plan + expedited order if profitable

To industrialize consolidation and save time on data quality work, you can rely on EtOH Data.

5) Turn the forecast into a buying plan and a commercial plan

Data only has value if it changes your decisions.

Set up a “light” S&OP (monthly)

One 60–90 minute monthly meeting is enough if it’s structured:

1. Sales review (forecast vs actual): top 10 variances, why.
2. Supply review: what’s arriving, what’s slipping, stockout risks.
3. Trade-offs:

- Which references to push (margin + availability)?

- Which references to slow down (low stock, allocation)?

1. Action plan: promos, substitutions, allocation priorities.

Examples of quantified trade-offs

• You foresee a stockout on a core-range white (A SKU) in 5 weeks.

- Option 1: expedite (more expensive freight): +€0,40/btl logistics cost, margin -€0,25/btl.

- Option 2: substitution: push a B SKU with +€0,60/btl margin.

- Decision: substitution for independent wine shops + on-trade (team training + product sheet), limited expediting for the 10 key retail customers.

Allocate intelligently under constraints

When stock is tight:

• Prioritize customers with high reorder rates and low price sensitivity.
• Protect the channel that builds the brand (often independent wine shops/on-trade) even if retail drives volume.

Operational checklist (copy-paste)

• [ ] 50 priority SKUs identified (by channel) + ABC rules documented
• [ ] Cleaned sales history (stockouts excluded, units standardized)
• [ ] Lead time measured (median + max) and updated quarterly
• [ ] Demand baseline calculated (8–12 weeks or 3 months) by A/B SKU
• [ ] Standardized adjustments (promo, price, new accounts, seasonality)
• [ ] Target stock defined (lead time + buffer) and replenishment thresholds set
• [ ] Weekly alert dashboard (5 KPIs + thresholds + actions)
• [ ] Monthly light S&OP ritual (variances, risks, trade-offs, actions)
• [ ] Substitution plan ready (2 alternatives per A SKU)

Conclusion

Anticipating import demand doesn’t require a data team or sophisticated models. With pragmatic segmentation, a simple baseline, a few business drivers, and strict lead time integration, you’ll reduce stockouts, protect cash flow, and improve margin. The key is turning your forecast into decisions (purchasing, allocation, commercial plan) through a monthly ritual and a weekly alert dashboard. Your competitors are still guessing; you’re steering.