New product forecasting for startups

Forecasting is an art, especially when you have zero sales.  New product startups face two forecasting questions:  1) How much revenue will be generated?  and 2) How much should be built?  Get your pencil out – time to predict the unknowable.  As someone so famously once said:  You don’t know what you don’t know.

But there are a few places to look, starting with your unique market situation.  Consider the following products(with launch year shown):

  • Google Glass  The new, new thing.  (2012)  Wearable computing.
  • Vuezone personal video network. (2009)  Home video monitoring on mobile devices.
  • FitBit fitness tracker. (2009)  Personal activity monitor.

These are new, new products – the next big thing.  No market exists.  Can’t find anything like them anywhere.  Good luck forecasting! You are launching an entirely new category of products and creating a new market at the same time. The customer need is real, but few realize what they might do with your bright shiny object.

Then consider this set:

  • NEST learning thermostat. (2010)  A net-connected smart thermostat that saves energy.
  • HP PaintJet color inkjet printer. (1987)  Color PC printing at a consumer price.
  • HP PhotoSmart digital camera. (1997)  Take photos digitally to get them online.

At the time these each launched, a well-understood market existed but not these categories.  Everyone knew of thermostats, computer printers, and (analog)cameras.  But smart thermostats managed at a distance, color printing at home, or images ready to email were all revelations!  These products created entirely new categories within existing markets.


  • Big Zoom  digital cameras. (2010)   Big zoom brought down to an affordable price.
  • HP OfficeJet All-in-one printers. (1995)   Printing, scanning, and copying in one integrated bundle.

Here the category exists and the technology is understood, but each adds unique new benefits to the mix.  You might have dramatically lowered price, delivering the same features as devices costing twice as much.  Or your new feature greatly enhanced an existing application.  Do so and you have carved out a new sub-segment in an existing category.  

Forecasting one of these?  First pick your market situation.

What is your market situation?

This issue of market identification is critical to making an effective forecast.  And each type offers its own challenges.  You get some flavor of that from the examples above.  Consider these 3 market situations:

Market situation # 1: New category and new market.  

This is tough.  It is also the situation faced by many startups.  Until seeing your new offering, a consumer did not realize such a capability existed.  You are starting from scratch.  First you need to educate them on the benefit you are providing, and only then can you present your product as the solution.

Since there is no market, any forecasted estimate will be bound based on external and internal factors.  Use two complementary approaches to bound the opportunity and create your forecast:

Take a tops-down approach.   The idea is to estimate your TAM: Total Addressable Market.   Start with the number of consumers or businesses that could use your product.  This is a maximum – start with  whatever constitutes your target in the broadest sense.   Next, estimate their interest in your product – either though guessing, running a suvery, or looking for third-party sources for judging interest.  You may have several winnowing factors.  Finally, in your first year, how many will you convert to sales?  Estimate your market share.

As an example, take VueZone – a remote video monitoring product viewable on your smartphone.  I use this general framework of:

Estimate market size -> Target Interest Level -> Your Market share w/ First Year factors: = Forecast.

Estimating the market size:

  • Reachable households or businesses.  For a consumer product we use the number of HHs in the US, as all would benefit from remote monitoring:  115M US households
  • But they must have broadband and use the internet.   HH’s with internet access:  85%
  • And the product is of little use without a smartphone.  HH’s with Smartphone 45%
  • Target demographics.  You might want to add another level if your product appeals only to a subclass of HHs – in this case income level is often higher for new tech products.  Assuming HHs with greater than $75K in income:  30%

The Vuezone TAM would be:  115M x 85% x 45% x 30% = 13M US Households.  So that’s still a big number.

Estimate the consumer interest in your product:

  • One good source, augmented by your own survey if possible, are analyst reports on technology trends. In this case, we find that a Parks & Associates survey estimates 17% of consumers are interested in remote video:  17%

Estimate your Market Share in year 1:

Consider how you will inform and convert customers.  Factors include the funding behind your communications and outreach programs, and how easy it might be to reach your target.  You will find being a category of 1 very difficult for many reasons!

  • Reachable in year 1 with message:  20%
  • Of those interested, how many can you convert to a purchase: 2%

Forecast for year 1 = 14M HH x 17% x 20% x 2% = 9500 units in year 1.

This of course is very crude – not even really useful.  Starting with any large number and multiplying with small factors can lead to huge misses.  And the straight multiplication ignored the likelihood of correlation across the various factors.  The resulting forecast is almost always too large; managers involved tend to be very optimistic.  But the process focuses your team on what will drive  sales.

Now try the bottoms-up approach.  Focus on proxies.  Some good questions to ask are:

  • How are the consumers doing this today?  Are there other ways to accomplish the task?
  • What other category are you going leverage – how big is that category (e.g you get traffic by associating in retail with an existing category). 
  • What are you going to do on Amazon?  A very complex place for unknown categories.
  • How many retailers can you sign up in year 1 – what will each deliver?
  • How have similar products fared.  For example, VueZone looked at the sales of WiFi cameras and the  sales profile of the early digital camera market.

Try and anchor the numbers against something similar, even if it’s not a product.  This is not always possible.

Getting to an estimate of year 1 market share is the lynchpin of the analysis.  You learn the most by digging deep and defining how you will escort customers through this launch phase.  Start by describing your target customer in as much detail as possible. There might be 50M Moms in the US, but only 5% will ever be in position to use your product frequently.  Of those, other factors will prevent purchase – too expensive (for them), too techie, waiting for smart friend to evaluate it, simply not yet aware, etc.  Not available where they shop.

Look for customer insights, not absolutes.  Focus groups give good flavor, but can be misleading.  Use quantitative research to understand your targets’ buying behavior.   Make smart assumptions, build a screener, and run a survey to put numbers around your model.   Research is useful for multiple reasons, but don’t kid yourself.  People are notoriously unaware of their actual behavior.   Respondents wanting to please are not always “brutally” honest, and companies in turn will ask leading questions:  “Would you buy this” seldom predicts anything.

As shown by this example, be wary of any forecast for a new, new product!

Market situation # 2: New category of product, but an existing market.

In this case a market exists.  Consumer readily understand the application – words exist to describe it.  Gaining an estimate of the number of potential customers – the market size – is straightforward.  However, your benefit is different and unexpected.  You will create a new category and estimating the product potential within the larger market is difficult.  Your share will depend on the strength of your channels, educating your prospects, and the length of their purchase process.  Still hard but doable with much more confidence than that of creating a new market.

Taking the NEST learning thermostat and using the forecasting framing from above:

Estimate market size:

The size of the thermostat market is will understood.  All HHs have a thermostat today – their replacement sales  is well understood as is the new home market.  You can generate a reasonable TAM without much effort.

Gaging consumer Interest:

Now estimate demand.  Some credible factors might be:

  • HH Broadband and SmartPhone penetration, upper income demographics (expensive product)
  • Customer psychographics:  Use an early adopter profile, and find DIY-type consumers (important as consumer will need to install a thermostat).

Market Share Year 1:

Factors behind a share estimate for year 1:

  • Channel capacity (and timing) – Number of Lowes and HomeDepot and expected orders from each.  Your ability to educate them in the store.
  • Purchase likelihood for the first generation of the NEST product.  Survey shoppers at Home Depot – they can qualify themselves for connectivity, smartphone, installation ability, and can express a likelihood to purchase.
  • Purchase process – what steps will be taken prior to purchase.  Can you impact those?

This is not a case of multiplying all these factors together.  Make a forecast and judge it against these factors. With care you can get a reasonable estimate.

Market situation #3:  Category extension.

Your product offers something new, but you are marketing against an existing set of competitors. Unlike the above two situations, analyst sizing estimates for the category exist, although for smaller markets there often are none.  You know where sell your product (eg. what retailers, which Amazon category, etc) and the sales capacity of each.  You have an idea of what you are willing to spend with each to promote.  Estimate the share you’ll take of the total sales volume you can reach.  Many category extensions simply shift market share to the innovator’s advantage.  More importantly, estimate your innovation’s ability to expand the market.  For example, a startup  offering dramatically cost-reduced capabilities might now find 4 customers for every one prior.  

For non-retail products other barriers arise.  Channel players may prevent sales, even if the buyers find your offering very exciting – dealers sell what makes them money or minimizes their pain – not what customers might want.  New stuff usually causes pain.  

But in this market situation the math is easy – estimate what share you’ll have – see if your spend justifies getting there.

Seek understanding, not certainty

While hard, an organization must align on a forecast.   Manufacturing needs it, finance needs it, channel partners all ask for it, and teams must be held accountable to a goal.  A good forecasting process forces you to encapsulate your marketing assumptions into an analysis, providing you a means to understand what went wrong later.  Building a forecast makes you think about all kinds of business issues (customer, pricing, competition, BOM, marketing spend, schedule, channels) and then proactively react as you enter the market.   But take caution against being too aggressive too early.  The best approach is to build a reasonable number and start selling them.  Reality quickly becomes apparent.

Do not seek a certainty that is not there – certainly in the market situation of new, new products.  Forecast smartly, not hopefully.   If you have too few units, you can quickly build more and will be thrilled to do so.  If you have too many, every thing gets screwed up.

Whole generations of products have been slipped under the need to sell off existing inventory.   Slips make you compete with less competitive products, and prevents your product teams from learning by building new ones.  You will not miss an opportunity in truly new market by under producing.  “New” does not explode that fast.

Forecasting is an exercise in seeking knowledge and understanding. For more about the forecasting process, see 7 Rules for forecasting your new product or service.



Want more on the future of marketing?  Follow me @gilbertdw.

Image courtesy of

One thought on “New product forecasting for startups

  1. It is such a clear and comprehensive article that I can follow to start a new product in the market. I will concentrate on the device that can adjust the temperature automatically or remotely – thermostat. Thanks for sharing each obvious detail. I’m extremely excited right now.

Leave a Reply

Your email address will not be published. Required fields are marked *