When developing new products, should your estimates of future value from these new products be precise or rough?
Here’s an analogy…
You love playing tennis, especially grass court tennis. You and a friend decide to play each other and you find that courts at your local tennis court are easy to book, but you must book in advance and no changes are allowed. You and your friend would like to book a court some time on Saturday, but you’re quite flexible on exactly when.
Before you book, you decide to check on the weather, so the game is not cancelled due to rain. You consult your weather app on your phone and it says it will be raining on Court 3 of your local tennis court on Saturday at 11:37am, with 0.3mm of rain, and a wind from the SE of 3 mph. That is the only information it gives you.
You think to yourself “wow! that’s very precise! I wonder how they do that…”
But does that information from the weather app help you know what time to book your tennis court on Saturday? It certainly helps to know that you shouldn’t book a court time around 11:30, but it doesn’t help you know if it will be raining before or after 11:30. It doesn’t help you know how long the rain will last.
And innovation forecasts are similar in a way.
Sophisticated Innovation P&L?
I frequently see very precise and highly sophisticated volume forecasts and projected P&L’s for these amazing new products. They have taken dozens of man hours, may be even having a cross functional team working feverishly to find out complex and bewildering information, to come out with a very precise answer.
The challenge is that there are so many variables, most if not all that are unknown at the time to the innovation team, that forecasting with accuracy is essentially a waste of time.
I can guarantee with a certainty of 99.9999% that any and every precise innovation forescast will be wrong.
Why can I guarantee that? It only considers one possible outcome of all the different possible variables that contribute to the forecast. It requires everything from the suppliers’ quotes & estimates, suppliers volumes, factory capacity, factory efficiency, distribution costs, number of stores selling, sales per stores, advertising reach, advertising effectiveness, advertising costs, consumer pricing, retailer pricing, competitors’ pricing, competitors’ advertising, competitors’ distribution, competitors’ rate of sale, cannibalisationrates, consumers and so on, to behave EXACTLY as the model has identified them to behave and to intereact in the EXACT way together. If only one of those elements is wrong, then the whole forecast is wrong.
So how can you manage all these variables and still make decisions?
I’m not advocating a rough back of the fag packet approach to innovation forecasting. I’m advocating rigour, but rigour in a way that accepts there is a wide range of possible outcomes and to predict the LIKLIHOOD.
We can outline a range of possible outcomes for each element and when combined together run a model to computate the probability of a specific outcome being reached. We can also outline the probability of a range of outcomes.
So as an example, I recently worked on a project that forecasted they would sell 11 million litres per annum of their new product. They were very happy with their model and so they should. It was accurate in the way it calculated it.
We were able to tell them that given the same elements and the range of those elements, and assuming all things being equal, there was a 66.2% probability of achieving the 11 million litres per annum volume forecast above, with a margin of error of 0.4% at a 95% confidence.
We can also work out how much volume would result from a given probability. Given the same example as above, we can say if you wanted to accept a 50% probablity of success then the volume would be somewhere between 0 and 19 million litres per annum, 75% probability of success then it would be between 0 and 7.8 million litres.
So next time you are running an innovation project and want to better understand the forecast, think about the range of possibilities, not the precise forecast.