Demand Planning for Ecommerce Brands

Ecommerce demand planning, part of demand planning and inventory optimization, is the work of forecasting online demand and positioning stock so that orders can be fulfilled quickly and profitably. Online businesses have an advantage over stores: they see every visit, search, basket and purchase, and they can test and adjust within days. They also have challenges. Demand is highly responsive to marketing spend, price and promotions, so it can move quickly and unpredictably. Customers expect fast delivery, free returns and accurate availability, and the cost of shipping from the wrong location is high. This guide covers how to forecast demand using drivers such as traffic and conversion, how to plan around marketing and promotions, how to handle returns, how to allocate stock across fulfillment centers, how to plan for peak periods and what to measure. It includes a checked worked example. It follows food and beverage, leads to specialty retailers and pharma and healthcare products, and connects to how to forecast ecommerce sales and cash flow management.

A driver-based forecast

A transparent way to forecast online demand is to build it from drivers: sessions, conversion rate, units per order and returns. Suppose a store expects 200,000 sessions next month, a conversion rate of 2.5 percent and 1.4 units per order (illustrative). Orders are 200,000 × 0.025 = 5,000. Gross units are 5,000 × 1.4 = 7,000. If 20 percent of units are returned, net units are 7,000 × 0.80 = 5,600. The driver view makes it clear what is being assumed. If marketing spend is increased and expected to lift sessions by 20 percent to 240,000, but the extra traffic converts at a lower rate of 2.2 percent for the blended total, orders become 240,000 × 0.022 = 5,280, gross units 7,392 and net units 5,914 (rounded). Check: 5,280 × 1.4 = 7,392, and 7,392 × 0.8 = 5,913.6. The lift in net units is 5,914 - 5,600 = 314, or 5.6 percent, far less than the 20 percent lift in traffic, because conversion falls. The driver model turns a marketing discussion into a planning input.

Linking the plan to marketing

Marketing is the main lever of online demand, so the plan must be built with the marketing calendar. Agree what campaigns are scheduled, with what spend, and which products are featured. The lift from a campaign should be estimated from past results, by channel and product. Establish a shared view of the baseline, the incremental volume and the uncertainty. When marketing wants to push a product, check stock and lead time first. Many online brands lose money on advertising that drives traffic to out-of-stock items. Equally, when stock is excess, marketing can be asked to feature those products. This two-way link is one of the largest opportunities in the category.

Availability and lost demand

When a product is out of stock, sales history understates demand. If the history is used naively, the forecast will learn the lower figure and create more stockouts. Record the days of stockout or low availability by product and adjust the history, using the pattern of similar products or the sales rate before and after. Also capture demand signals that do not convert: product page views, add-to-cart events, back-in-stock requests and waitlist sign-ups. These show interest that could not be served and help in sizing the next buy.

Returns

Returns reduce net demand and create a flow of goods back into stock. Rates vary by category, with higher rates in fit-sensitive and gift categories. Forecast returns as a percentage of units by product group, with a time lag between sale and return. Check what share of returned items can be resold, how long processing takes and what share must be discounted or scrapped. Include the processing time in availability, since an item that is sitting in the returns queue is not available. Look for patterns in return reasons that suggest product or content issues. See inventory write-downs for the accounting of unsellable returns.

Fulfillment center allocation

If the business ships from several fulfillment centers, the allocation of stock determines delivery speed and cost. Stock held near customers reduces shipping time and cost, but spreading stock across many locations increases the total needed because safety stock is held at each. The trade-off resembles the pooling effect: fewer locations need less stock for the same service, but more locations improve delivery speed. A practical approach is to hold slower items in a central location and faster items in regional centers, allocated by the forecast of regional demand. Monitor the share of orders split across locations, which raises cost, and rebalance regularly. See safety stock formula.

Pre-orders, subscriptions and drops

Pre-orders give early visibility of demand and cash, but they create promises that must be met. Treat pre-orders as committed demand with a delivery date, and plan supply accordingly. Subscriptions provide predictable base demand, with churn and pauses that should be forecast. Product drops, limited releases at set times, concentrate demand in a short window and are best planned with scenarios, with controls such as purchase limits per customer. For all three, communicate realistic delivery dates and keep them.

Peak periods

Peak periods, such as holiday shopping and major sales events, can account for a large share of annual sales for some online businesses. Plan well ahead. Establish the last order dates for suppliers, the capacity in fulfillment and carriers, the staffing and the cash needed. Build the peak forecast with scenarios, and agree triggers for actions such as expedited freight or promotional changes. Beware of promotions that bring forward demand from the following weeks, which leaves a quiet period in which stock is higher than needed. See seasonality and seasonal demand planning.

Marketplaces and multiple channels

Selling through marketplaces adds a channel with its own rules, fees and fulfillment options. Demand may be sensitive to ranking, price and reviews. Forecast marketplace demand separately, and keep rules for allocating limited stock between your own site and marketplaces. If a marketplace fulfills for you, understand its inbound requirements and lead times. Avoid committing all stock to one channel at the expense of another with higher margin.

Pricing and promotions

Online prices are visible and easy to compare, and price changes can shift demand quickly. Use price elasticity estimates where available and test changes in a controlled way. Plan promotions with an estimate of lift, cannibalization and the effect after the event. Include shipping offers in the plan, since free shipping thresholds change basket sizes. See price elasticity of demand and promotion forecasting.

Data and tools

Ecommerce platforms provide rich data: sessions, conversion by source, basket contents, product views and returns. Combine them with inventory and supply data in a planning table. A spreadsheet can work for a small store, and as the number of products grows, specialist tools or a data warehouse with automated forecasts become useful. Ensure that product identifiers match across systems. See data quality.

What to measure

Track forecast accuracy and bias by product group and by horizon, in-stock rate and stockout days for top products, fill rate and on-time delivery, return rate and processing time, share of orders split across locations, days of inventory, aged stock and the cost of fulfillment per order. Review weekly in season and monthly otherwise. See in-stock rate.

Pooling stock across fulfillment centers

The effect of splitting stock across locations can be shown with a small example. Suppose a product has weekly demand with a standard deviation of 60 units in each of three regions, and the regional demands are independent (illustrative). At a service level of about 95 percent (z of 1.65), safety stock per region is 1.65 × 60 = 99 units, and for the three regions 297 units. If the stock were held in one central location, the standard deviation of combined demand would be the square root of 3 times 60 squared, which is 60 × 1.732 = 103.9, and safety stock would be 1.65 × 103.9 = 171 units. Check: 99 × 3 = 297, and 1.65 × 103.9 = 171.4. The central pool needs 297 - 171 = 126 fewer units, a reduction of about 42 percent. The price is slower and costlier delivery to customers far from the center. Between the two extremes lie options, such as two regional centers for the fastest items and one central center for the long tail. The best answer weighs inventory cost against shipping cost and the value of speed to the customer. Real regional demands are often correlated, which reduces the benefit, so use your own data.

Contribution per order and the cost of a lost sale

Online planning decisions should be tied to unit economics. Suppose an item sells for $50 with a product cost of $22 and fulfillment and shipping costs of $8, leaving a contribution of $20 per unit (illustrative). A typical order has 1.4 units, so the contribution per order is 20 × 1.4 = $28. If the cost to acquire the order through advertising is $18, the net contribution is 28 - 18 = $10 per order. A stockout of a product that is being advertised wastes the $18 for each click that fails to convert, so the true cost of a stockout includes the wasted marketing as well as the lost margin. Conversely, if returns of 20 percent each cost an additional $6 in handling, that cost on 1.4 units × 20 percent × $6 is about $1.68 per order, which reduces the net contribution to roughly $8.32. Use these numbers to set service targets by product: high-contribution items with heavy advertising deserve higher availability.

A weekly trading rhythm

Because online demand can change within days, many teams run a short weekly trading meeting. Marketing brings the plan for the next two to four weeks, planning brings the forecast and stock position, and operations bring capacity and carrier status. The group reviews the top products by sales and by risk, agrees actions such as reorders, price changes, promotions or campaign shifts, and records decisions. The meeting is brief and focused on the next few weeks, while the monthly planning cycle handles the longer horizon of supply commitments. The rhythm keeps marketing and supply aligned in a way that no tool does by itself.

Common mistakes

Forecasting from sales without correcting for stockouts understates demand. Treating traffic growth as demand growth ignores conversion. Ignoring returns overstates net demand. Allocating stock to fulfillment centers by size and not by demand creates imbalances. Planning the peak without checking supplier cut-off dates leaves no time to react. Letting marketing run campaigns without knowing the stock position wastes spend.

FAQs

How do you forecast ecommerce demand?

Use drivers such as sessions, conversion, units per order and returns, adjusted for marketing and promotions, and corrected for stockouts in history.

How should marketing and planning work together?

Share the campaign calendar and stock position so that spend supports products that are available and excess stock can be featured.

How do I handle returns in the forecast?

Forecast returns as a percentage of units with a time lag, and plan for how much can be resold and how quickly.

How should stock be allocated across fulfillment centers?

Hold fast items in regional centers by forecast demand and slower items centrally, balancing delivery speed against total stock.

How do I plan for the peak season?

Build scenarios, check supplier cut-offs, capacity and cash early, and set triggers for expedited actions.