Most successful companies reinvest 3-6% of net sales into Research and Development. The amount reinvested is not really related to the size of the company, but it is often related to the type of industry. Generally manufacturing (electronics included) is at the low end of the scale with 3-4%, while pharmaceuticals, for instance, typically revinvests 15% of net sales in R&D. To put some scale on these numbers, in 1998 Proctor and Gamble had sales of $38US billion, and reinvested 4.5% in R&D - that is: $1.7 billion, a rank of 21st largest investor in the US (Source: "Jagers gamble" Economist magazine October 28, 1999)
The amount reinvested into Research and Development is often an indicator of where the company is (or believes they are) on the "technological wave". Companies that are at the forefront of exploiting new technology, the leading edge, usually invest more. More often though, the spend is part of company culture - companies in precisely the same industries may each invest double or half the "normal" amount in R&D.
The quality of the spend probably accounts for just as much as the actual amount. Some companies do not have the best environment to foster creativity among their own staff (see our comments on creativity and inspiration). Other companies have cultures that stifle innovation and prevent new ideas from getting to market quickly. Other companies have a significant aversion to risk - rather than a calculating acceptance that some new products while fail, while others succeed. There is a whole consultancy business built around telling directors where their companies have gone wrong, and it is quite certain that there is no single simple way to "do it right".
The more original a discovery, the more obvious it seems afterwards --Arthur Koestler
Perhaps the best approach is to focus less on the company, and more on each specific project. Each project will have requirements which may be best met by in-house personnel, contractors, subcontractors or a mix of these. The project may be best handled within the established, tiered R&D department management, or as a seperate "innovation team" - perhaps even housed in a physically seperate location to the existing company. The project risk and return on investment should be evaluated in isolation from other company activities.
Some of the most interesting small companies are those that have one core product that has made them successful (unkindly, they might be labelled one hit wonders). Moving on from this position is quite a challenge. Usually the best that can be done is to better exploit the existing market with new products, or move the old product into a new market. The worst that can happen is the company owner believing that he "has what it takes" and that every product he tries will be just as successful as the first one.
Some small companies avoid developing new products that they see as competing with what they already make. There is certainly a case for reducing duplicated effort, and also for ensuring that a new project does not take too many resources away from an existing project - however, to make your own "competing" product can be a most successful avenue for some small companies. It's a lot better than your real competitors developing the same thing!
Choosing R&D on a project by project basis rather than trying to steer an overall company direction makes other decisions much easier, also. Larger companies continuously have to decide if new developments are part of their core business, - and if it fits company policy of specialization or not. Years ago, Nokia used to make many products, including rubber boots and toilet paper.
Everything that can be invented has been invented --Charles H. Duell, Commissioner, U.S. Office of Patents, 1899
The myth of sunk costs. Electronics R&D projects (and even more so, programming) are very difficult to budget for. The engineers joke that the management set wish-based deadlines, and the management joke that the engineers will always take one (arbitrary) unit of time longer than the deadline, no matter how much time they are given. However, in some projects the (lack of) progress gets beyond a joke. It's not always (or even often) the fault of the people involved. Unfortunately there is risk involved in any R&D project, and it seems that the main risk encountered is not that the project will fail to work, but that it cannot be made to work within a reasonable length of time. The best approach is an honest, pragmatic re-evaluation as if a new project was being started to build on top of the work already completed. It is usually not opportune to either start a witch hunt, or, conversely, to reason "we can't stop now, look, we have $xxx,000 invested already".
There's no point in putting whipped cream on manure -- John Madden