Those populations are often in defined sectors, such as industry. Surprisingly, though, there is no universal sample frame, or even comprehensive ones, for most sectors.
Directories, although widely used, miss out many smaller businesses and include names of businesses that are in unrelated fields. Also, where research is by sector, the definition of the sector relevant to the research objectives may not match the classifications used in sample frames.
While this may suggest that reliable sampling is not possible, in business-to-business research, samples are taken from directories and often built from several sources. Whilst not completely free from bias, these samples are good enough for most purposes.
There are other issues in business-to-business research too - particularly those relating to the fact that the companies in the population are not of equal size. Take for example the chemical industry. At one level there could be a small company employing one or two people and mixing chemicals with a bucket and a stick, through to the DuPonts, the BASFs and the Dows.
In most business-to-business markets the distribution of companies fits the 80:20 or Pareto rule; that is 20% of the units account for 80% of the market being studied. The domination of most business markets by a small number of large companies means that it is crucial they are included in the sample. In fact, the researcher may seek (so far as co-operation levels allow) to carry out a census of the largest companies, and a sample of the rest.
This singling out of specific companies to interview is judgmental, although there will be some guiding principles such as the number of employees, the output of chemicals, turnover and the like.
Not only does the business-to-business researcher have to contend with wide differences of size of company, there are also complications caused by the intricate nature of the decision making unit (DMU). Whereas decisions in most households are made by the partners and the children (or even by just one person), in businesses there are inputs from purchasing, technical, production, and possibly finance and marketing. These inputs change over time. At the point where a product is being specified or evaluated for the first time, the technical team is likely to have a strong voice. Once the product is in the routine re-ordering stage, the purchasing and production people have more influence. So who do you select to interview? In theory you may set out to build up a picture across the board by selecting people from each group. However, since many business-to-business samples are small (200 interviews is quite a respectable size), the sub-cells of respondents in a particular job responsibility end up being too small to analyse them separately. For these reasons the business-to-business researchers may compromise and decide to concentrate the interviews on just one group such as the key decision maker - thankfully there is usually one person who holds most sway in the decision.