Demand planning vs forecasting: overrides, last year plus growth and S&OP-lite for DTC

"The forecast" means different things to different people in the same business. To the ops lead, it's what the model says will sell. To the founder, it's the revenue target. To marketing, it's what will sell after the campaign. When these aren't separated, nobody knows which number the purchase orders were based on — and when stock runs out or piles up, nobody knows why.

This guide separates the forecast from the plan, explains how to adjust one into the other without losing track, and lays out a light monthly process that works for a small DTC team.

Key takeaways

• The forecast is an unbiased estimate of demand from history. The plan is the number you commit to buying for, including what you know that history doesn't.
• Keep both. The gap between them is information.
• Overrides should be specific, reasoned and measured. Overrides that don't improve accuracy should stop.
• Last year plus growth is useful but fragile: it repeats last year's stockouts, promotions and one-offs unless you clean them.
• A monthly cycle of data, demand, supply and sign-off is enough for most DTC brands — call it S&OP-lite.

Forecast and plan are different numbers

The statistical forecast is a model's estimate of future demand based on history. It should be unbiased: as likely to be too high as too low. It knows about trend and seasonality in the data. It doesn't know about next month's influencer campaign, the wholesale account you just signed or the price rise in March.

The demand plan is the forecast plus that knowledge. It's the number purchase orders, cash flow and staffing are based on. It belongs to a person, not a model, and it should be explainable line by line.

A third number is worth keeping alongside them: last year's actuals for the same period. Not as a forecast, but as a reality check. If the plan says November will be three times last November, someone should be able to explain why.

Seeing the three side by side, per product and month, is the core of demand planning. Stokrio's demand and supply plan is laid out exactly this way: forecast, plan and last year for each product and month.

Overrides: when and how

An override is any change from the statistical forecast to the plan. Good reasons:

• Known future events: a promotion, a launch, a marketing push, a new retail account, a price change.
• Known supply events: a product being discontinued, a reformulation, a packaging change that resets demand.
• Information the history can't show: pre-orders, waitlist sign-ups, a confirmed bulk order.

Bad reasons: a feeling that the number "looks low", a target that needs to be met, a single unusual week.

Make every override:

1. Specific — which product or category, which weeks or months, how much (a percentage or a fixed number of units).
2. Reasoned — one line explaining why. "Black Friday 25% off" is enough.
3. Time-limited — an override for a November promotion should end in November, not linger.
4. Measured — once the period is over, compare the forecast and the plan with what actually sold.

The last step is the one teams skip. If the plan is consistently more accurate than the forecast, the overrides are adding value. If it's consistently less accurate, or consistently higher than actual sales, the overrides are costing money. Our guide to forecast accuracy covers how to compare them.

Last year plus growth

The simplest plan of all is last year's sales for the same period, multiplied by expected growth. It's popular because it's easy to explain, captures seasonality automatically and is easy to reconcile with a revenue target.

It works when last year was a clean picture of demand. It fails when last year included things that won't happen again. Consider a product that sold 300 units last November, with 15% growth expected:

last year + growth = 300 × 1.15 = 345 units

Now suppose the product was out of stock for 5 of November's 30 days. It sold 300 units in 25 in-stock days, a rate of 12 per day:

estimated November demand = 12 × 30 = 360 units
cleaned last year + growth = 360 × 1.15 = 414 units

The difference is 69 units — about 20% of the naive plan — and the naive version would lock last year's stockout into this year's plan. Before applying growth to last year, clean it:

• Replace stockout periods with estimated demand. See stockouts and lost sales.
• Remove one-off wholesale orders and flag promotions that won't repeat.
• Align moving events — Black Friday, Easter, your own sale dates — rather than comparing calendar weeks.

And be careful with a single growth rate across the whole range. If total revenue is growing 15%, some products are growing 40% and others are declining. Apply growth per product or category where you can.

Reconciling top-down and bottom-up

Leadership usually thinks in revenue; operations thinks in units. The plan has to satisfy both. Suppose next month's revenue target is $120,000 across three products, and last year's mix was 50%, 30% and 20%:

At the forecast, revenue would be:

(1,300 × $40) + (1,500 × $25) + (300 × $60)
= $52,000 + $37,500 + $18,000
= $107,500

That's $12,500 — about 10% — short of target. This is the useful conversation. Either there's a specific plan to close the gap (a campaign on product A, a bundle featuring product C), in which case it becomes an override with a reason, or the target is a hope rather than a plan, and inventory should be bought to the forecast.

What you shouldn't do is quietly raise every product's plan by 10% to hit the number. That turns a revenue target into excess stock.

From demand plan to supply plan

The supply plan answers: given this demand, what has to arrive and when? Roll stock forward month by month:

closing stock = opening stock + receipts − demand

Take a product with 800 units on hand, a planned demand of 400, 700 and 500 over the next three months, a purchase order of 300 arriving in month 1, and a minimum of 200 units you want on hand at each month end:

Each recommended receipt has an order-by date: the month it's needed, minus the supplier's lead time. If the lead time is two months, the month 3 receipt must be ordered now. Changing the plan for month 3 next month will be too late. That's why the plan has to be agreed on a cadence that runs ahead of lead times.

S&OP-lite: a monthly cadence

Sales and operations planning in large companies is a heavyweight process. A DTC brand needs the core of it, run by two or three people in a few hours a month:

1. Week 1 — data. Close last month. Update sales, stock and purchase orders. Measure forecast and plan accuracy. Flag stockouts and anomalies.
2. Week 2 — demand review. Marketing, ecommerce and ops review the forecast for the next three to six months. Agree overrides, with reasons. Reconcile with the revenue target.
3. Week 3 — supply review. Roll the plan into a supply plan. Identify receipts that need ordering, constraints (MOQs, cash, warehouse space) and risks (late suppliers, single-source products).
4. Week 4 — sign-off. One person approves the plan, the purchase orders and the budget. Anything unresolved is written down as a known risk.

The output is one agreed plan per product and month, a list of purchase orders to place, and a short list of decisions. The plan then drives replenishment until the next cycle — and when something changes mid-month, you change the plan, not just the order.

In Stokrio, overrides made on the demand and supply plan flow straight into replenishment and alerts, so the agreed plan and the purchase orders don't drift apart between meetings.