The best innovations don't start with ideas, regardless of what you think or have been told. They start with noticing evolving and emerging trends, and understanding customer needs. Good innovators who can gain insight from these two activities in a purposeful way can identify emerging opportunities and needs and create meaningful new products and services. I call this "creating context" for innovation--that is, gaining understanding about where markets and customers are heading and the challenges and problems they have, before trying to generate ideas.
This approach isn't necessarily radical or new, but few entrepreneurs or innovators take the time to create context. For those who are interested in trend spotting and scenario planning, however, another question emerges: how far into the future should I look for emerging customers or trends? This is the right question to ask, because too many innovators don't know how far into the future to look for meaningful insight.
There are several ways to answer the question of how far into the future you should look. One way to answer depends on the type of innovation you want to create. A small, incremental change to existing products doesn't warrant looking more than 6-12 months into the future. On the other hand, creating an interesting, disruptive product or service based on new technologies or emerging trends may warrant a 3-5 year look into the future. As a general rule of thumb, the more disruption you hope to create or introduce, the deeper into the future you should look for emerging trends or evolving customer needs.
A More Definitive Metric
Another answer I often use when this question arises is based on the company's product development cycle. If, for instance, a company has a product development cycle of 18 months, then it should look at least 24 months to 36 months into the future. Why? Because we know that if it spots any emerging trend or opportunity that may emerge in less than 18 months, it probably won't be able to respond (since its own product development cycle is 18 months long). With this in mind we can say that if you are doing trend spotting or future forecasting, you should look at a future from 1.5 to 2 times your product development cycle.
There are caveats to this of course. In an industry with aggressive competition and shrinking product development and product life cycles, you may want to look further out to skip generations. In other, longer life cycle industries you may want to consider the life expectancy of your product. For example, the US Navy often considers the useful life of a ship to be up to 30 years. In this case they should be thinking about emerging needs and threats for at least a 30 year period as they construct a new vessel.
Getting the Timeline Right
Of course simply doing trend spotting and scenario planning is important, but if you fail to get the timeline or time period correct you'll identify opportunities you simply cannot fulfill. Also, looking too far into the future (for most companies this is anything beyond 5 years) seems like magical prediction rather than purposeful investigation. Which is both true and unfortunate. Constantly investigating and practicing the future makes you much more capable and much more informed about the possibilities than your competitors will be. Alan Kay, a giant figure in computing, was fond of saying "the best way to predict the future is to invent it". You can invent your future by identifying trends and emerging needs and acting before your competitors do.