Revenue Forecasting: Strategies for Uncertain Times

Revenue forecasting in a hardware-software business is harder than it looks from the outside, and the difficulty is structural. Inventory decisions are made months before the demand is known. Component lead times, manufacturing slots, and supply chain commitments all require a volume commitment before the market has confirmed it. The uncertainty is not a problem to be solved with better tools. It is a fundamental condition of the business.

The most common forecasting failure I have seen in this context is not bad data or insufficient analysis. It is over-optimism about the adoption curve combined with underestimation of how long it takes customers to integrate and deploy a new product. A hardware-software product does not start generating revenue the moment it ships. It starts generating revenue when the customer has installed it, tested it, trained their staff, and resolved the integration issues that almost always arise. That cycle is longer than most revenue forecasts assume, and the gap between shipment and revenue recognition has derailed more than a few otherwise reasonable plans.

What good forecasting practice looks like

The most useful forecasting models I have worked with treat the forecast as a set of explicit assumptions rather than a single point estimate. The question is not “what will revenue be in Q3” but “if adoption follows this curve, and customer integration takes this long, and the supply chain delivers on this schedule, then revenue in Q3 will be approximately this.” The assumptions are in plain sight. When reality diverges from the forecast, the team knows which assumption was wrong rather than simply knowing that the number was wrong.

Scenario planning helps, but only if the scenarios include a genuine downside rather than an optimistic case and a base case wearing a slightly darker coat. For a hardware-software business, that downside needs to include supply chain disruption, certification delays, a key customer pushing deployment back by a quarter, and an unexpected competitive move in a core segment. None of these are edge cases. All of them happen.

Updating the forecast

The other discipline that separates useful forecasting from unhelpful forecasting is the cadence of updates. A forecast that is built in January and reviewed in December is a plan rather than a forecast. A forecast that is reviewed monthly against actual pipeline progression, customer deployment status, and supply chain signals is a live management tool.

The signals worth tracking are not just the obvious ones. Customer integration timelines in the existing base indicate something about how long new customers will take. Distributor inventory levels indicate something about near-term demand. Competitor pricing moves indicate something about margin pressure. So do certification progress, factory yield, field failure rates in the first deployed cohorts, and the time it takes service partners to become genuinely operational rather than merely contracted. None of these appear naturally in the standard revenue model, but all of them affect whether nominal demand becomes recognised revenue on the timetable the plan assumes.

The goal of forecasting is not accuracy in the sense of producing the right number. It is accuracy in the sense of making the assumptions explicit enough that they can be challenged and updated before the decisions they inform become irreversible.

The behaviours that undermine forecasting

Even with good process, forecasting fails predictably for human reasons. The most common is optimism bias: the tendency to assume that adoption will happen faster than the evidence suggests, that the customer who said “probably Q2” will not push to Q4, and that the supply chain will deliver on the initial promise rather than the revised one. The optimism is often genuine; the people involved are not being dishonest. They are applying the same cognitive shortcuts that make daily decision-making manageable to a problem that requires calibrated uncertainty instead.

Sandbagging, deliberately underestimating revenue to make targets easier to hit, is the mirror image. It is less common in early-stage businesses and more common once a bonus structure is attached to the numbers. The result is a forecast that the finance team does not trust, which triggers a renegotiation that produces a forecast nobody trusts, which makes the planning process increasingly theatrical.

Both problems have the same root: forecasts treated as commitments rather than as estimates with explicit uncertainty. When the number becomes a target, the incentive to produce an accurate number is replaced by the incentive to produce a number that can be defended. The fix is to make the uncertainty explicit and legitimate; to ask not just for the forecast but for the range and the assumptions underlying it, and to treat a well-reasoned downside scenario as a sign of rigour rather than pessimism.

Forecasting and board trust

The board’s relationship with revenue forecasting is often one of accumulated disappointment. Numbers are missed, explanations are offered, the forecast is revised, and the cycle repeats. Over time this erodes confidence not just in the specific numbers but in the leadership team’s understanding of its own business.

The way to rebuild that trust is not to produce more accurate forecasts, that is genuinely difficult, but to produce more transparent ones. A board that understands why a forecast was missed, and can see that the underlying assumptions were reasonable at the time and have been updated, retains confidence in the management team even when the numbers disappoint. A board that receives a revised forecast with no explanation of what changed or why has no basis for that confidence.

The most useful shift a leadership team can make on this is to share the assumptions before the period rather than the explanations after it. That change, from defending past performance to explaining current thinking, changes the nature of the board conversation entirely.