When Demand Has No Shelf: Planning in a Digital Goods World

Demand planning has traditionally been built around a fairly intuitive problem: determine what customers are likely to buy, translate that demand into units, and ensure the supply network can place the right inventory in the right location at the right time. For practitioners in physical goods businesses, the consequences of getting the forecast wrong are visible. Too much demand creates stockouts, expedites, and lost sales; too little demand creates excess inventory, markdowns, obsolescence, and working-capital pressure. As companies increasingly move into digital goods, subscriptions, software, digital services, and hybrid physical-digital offerings, that basic planning question does not disappear. What changes is the economic system surrounding the forecast. The planner is no longer primarily determining how much inventory needs to be positioned. Instead, demand planning increasingly becomes a way to anticipate customer adoption, usage, capacity requirements, revenue, and the operational resources required to support that demand.

The Fundamental Shift: From Inventory Risk to Demand and Capacity Risk

The most important distinction for practitioners is that physical goods planning is constrained by inventory and replenishment lead times, while digital goods planning is generally constrained by capacity, customer behavior, and service economics. In a physical business, a demand forecast of 10,000 units immediately raises supply questions: Where will those units be manufactured? What materials are required? How long will replenishment take? Where should the finished inventory sit? The forecast creates a series of inventory commitments throughout the supply chain.

For a digital product, the marginal unit may not need to be manufactured at all. Selling another software license, digital subscription, media download, or cloud-based service does not normally require the business to hold one additional finished unit in inventory. That can create the impression that demand planning becomes less important. In practice, I would argue the opposite. The forecast still drives decisions; the decisions simply migrate to different parts of the business.

Instead of determining inventory positions, planners may be helping the organization anticipate cloud infrastructure, transaction-processing capacity, customer-support staffing, implementation resources, licensing requirements, sales capacity, or revenue expectations. Forecast error therefore has a different cost profile. An overforecast may no longer leave thousands of units sitting in a warehouse, but it could cause unnecessary infrastructure commitments, excessive staffing, unrealistic revenue guidance, or poor allocation of commercial resources. An underforecast may result in degraded system performance, inadequate customer support, implementation bottlenecks, or missed opportunities to expand capacity ahead of a surge in usage.

The Unit of Demand Becomes More Complicated

Physical products typically give the planner a relatively concrete unit of measure. We forecast cases, pieces, pallets, tons, or some other representation of volume. Digital businesses force a more fundamental question: What exactly are we forecasting?

A digital business can have multiple legitimate measures of demand operating simultaneously. New subscriptions may matter to Sales and Finance. Active users may matter to Product. Transactions or compute consumption may matter to Technology. Support contacts may matter to Customer Operations. Renewals and churn may matter to Customer Success. Revenue may matter to the executive team. All of these measures describe demand, but they answer different planning questions.

This means the demand planning process has to become more explicit about the relationship between commercial demand and operational consumption. One customer does not necessarily equal one unit of workload. Ten thousand customers who use a service occasionally may create less operational demand than one thousand customers who use it intensively. The practitioner therefore needs to understand not only how many customers are likely to be acquired, but how those customers behave after acquisition.

For mature digital businesses, I often think of demand as a hierarchy: prospects become customers, customers become active users, users generate activity, and activity consumes resources. Building those conversion relationships into the planning model can be much more useful than trying to force a traditional SKU forecast onto a digital business.

Demand Signals Become Faster—and Potentially Noisier

One major advantage of digital products is the amount of demand information they can generate. Physical businesses frequently rely on orders, shipments, point-of-sale data, distributor inventory, and promotional calendars. Those remain valuable signals, but they can contain meaningful delays between consumer behavior and planner visibility.

Digital businesses can observe behavior almost immediately. Searches, registrations, trial starts, logins, feature usage, transactions, cancellations, upgrades, and renewal behavior can all become potential demand signals. From a planning standpoint, that creates an enormous opportunity to sense changes earlier.

It also creates a new problem: there is far more data available than there are useful signals.

Practitioners should resist the assumption that additional data automatically improves the forecast. Digital businesses can produce hundreds of behavioral metrics that correlate with demand without actually improving predictive accuracy. The planning challenge becomes determining which signals are leading indicators and which simply describe activity that has already occurred. A small number of stable measures—customer acquisition, conversion, usage intensity, retention, and churn, for example—may provide significantly more planning value than a dashboard containing dozens of rapidly moving product metrics.

The planning cadence can also become faster. A physical supply chain may have monthly planning cycles because material procurement and production decisions operate over weeks or months. A digital business may discover meaningful changes in customer behavior within days or even hours. That does not mean every organization needs an hourly demand forecast, but it does mean planners should reconsider whether traditional monthly forecasting cycles remain appropriate for every decision.

Segmentation Moves Beyond the SKU

Physical-goods planners are accustomed to thinking about demand through product and location hierarchies: SKU, product family, channel, distribution center, geography, and customer. Those dimensions remain relevant in hybrid businesses, but digital planning often requires a different type of segmentation.

Customer behavior becomes particularly important. New customers may have completely different usage patterns than mature customers. Enterprise clients may create substantially different workloads from individual users. Free users may behave differently from paid subscribers. Customers on different pricing tiers may have very different usage intensity, renewal rates, and support requirements.

Consequently, segmentation increasingly needs to explain how demand behaves, not simply where demand belongs. Cohort-based forecasting becomes especially valuable. Rather than forecasting total subscribers as one homogeneous population, the planner might separately model new customer acquisition, existing customer retention, expansion, contraction, and churn. Those components can then be combined into a more transparent demand outlook.

This approach has another advantage: it makes forecast conversations more actionable. If demand is below plan because acquisition is weak, the response may belong to Marketing or Sales. If acquisition is healthy but churn is increasing, the response may belong to Product or Customer Success. Demand planning begins to move from explaining a number to explaining the mechanics producing that number.

The Role of the Planner Expands

Perhaps the largest change is organizational. In a traditional physical-goods environment, demand planning often sits at the intersection of Sales, Marketing, Finance, and Supply Chain, with the forecast ultimately feeding supply, inventory, and production decisions. In a digital business, the number of stakeholders expands. Product, Technology, Customer Success, Revenue Operations, and cloud or infrastructure teams may all depend on some representation of future demand.

The practitioner therefore becomes less of a “unit forecaster” and more of an integrator of demand assumptions across the enterprise.

That requires different conversations. Instead of asking only, “How many units will we sell?” the planner may need to ask: How many customers will we acquire? What percentage will activate? How quickly will usage ramp? What will retention look like? How much infrastructure does that level of activity consume? What happens if a new product feature materially changes usage? Which customers are likely to upgrade? What operational bottleneck appears first if demand exceeds the plan?

These questions are familiar in spirit to experienced demand planners. They are still questions about uncertainty, assumptions, drivers, and constraints. The difference is that the planning model increasingly connects demand to a service-delivery system rather than a manufacturing and inventory system.

What Does Not Change

Despite these differences, practitioners should be careful not to throw away the discipline developed in physical-goods planning. Many of the fundamentals remain exactly the same.

Forecasts still need clear ownership. Bias still matters. Accuracy still needs to be measured at the level at which decisions are being made. Overrides should still have an explicit rationale. Assumptions should be documented. Commercial teams and operational teams still need one reconciled view of demand. Scenario planning remains more valuable than pretending a single forecast represents certainty.

Most importantly, the objective of demand planning remains unchanged: help the organization make better decisions under uncertainty.

For physical products, that often means protecting service while minimizing inventory and supply-chain cost. For digital products, it increasingly means aligning customer growth and usage with infrastructure, service capacity, commercial investment, and financial expectations. Hybrid companies have to do both simultaneously.

That is ultimately where I believe the profession is headed. Demand planning is becoming less defined by the presence of inventory and more defined by the discipline of translating market behavior into operational consequences. Practitioners who understand that distinction will find that their core skills—forecasting, segmentation, exception management, scenario analysis, and cross-functional alignment—remain highly relevant. The opportunity is to apply those skills to a broader definition of demand.