Every forecasting method rests on history, which leaves new products in an awkward position. The usual response is to ask sales for a number, accept it, and discover several months later that it was optimistic — a pattern so consistent across manufacturers that it’s worth treating as a structural problem rather than a series of individual misjudgements.
Use Analogues
You may have no history for this product, but you probably have history for something similar. Find products with comparable price point, target customer, channel, and positioning, and look at their launch curves — not their eventual volumes, but the shape of the ramp.
Use several analogues rather than one, and look at the spread between them. That spread is a more honest picture of your uncertainty than any single projection, and it tells you what range to plan for. Include analogues that underperformed as well as successes — selecting only from products that did well builds optimism into the method itself.
Adjust deliberately for the ways this launch differs: more or less marketing support, wider or narrower distribution, stronger or weaker competition. Write the adjustments down so they can be reviewed afterwards.
Separate the Questions
A launch forecast bundles several distinct estimates that are better made separately: how many customers will try it, how quickly, how much each buys, and how many repeat.
Breaking it apart makes the assumptions visible and arguable. It also tells you what to watch after launch — if trial is on track but repeat is weak, that’s a product problem, and it needs a completely different response from a distribution problem.
Ask Customers, Carefully
For products sold to a known customer base, direct enquiry beats internal estimation. Customers will tell you whether they intend to buy, though stated intent consistently overstates actual behaviour and needs discounting.
Where you have distributors or major accounts, their opening orders are a real signal — with the caveat that opening orders include pipeline fill, which is one-off. Distinguishing pipeline from consumption is the most common error in early launch reads, and it makes the second month look like a collapse when nothing has gone wrong.
Plan For The Range, Not the Number
This is where launch planning most often fails. A single-point forecast drives a single supply plan, and being wrong in either direction is costly: stockouts during the launch window that you never recover, or a warehouse full of a product that didn’t land.
Better to plan supply in stages. Commit to an initial quantity sized for the conservative case, secure the option to increase quickly, and agree with suppliers in advance what a rapid follow-on looks like. Paying for that flexibility is usually cheaper than either failure mode.
Correct ast
The first few weeks of real data are worth more than any pre-launch estimate. Set a formal review at two, four, and eight weeks with an explicit decision each time — increase, hold, or cut. Launches go wrong when the original forecast keeps driving supply long after actual demand has said something different.
ticktick.ai supports analogue-based launch forecasting with staged supply commitments and automatic reforecasting as early actuals arrive.
