- Build a priori segmentation dimensions. Follow your business's conventions. For example, if a cut-off point in defining SMB is companies less than 500 employees, use this. Keep things as industry standard as possible. Don't use industry classifications that aren't used by the Department of Labor. Keep it to three dimensions max. Three that work are Size * Geography * Industry. You want 30 cells max. So, three cells for size, three for geography, and 2-3 for industry work great. Once you have your grid of all possible cells, write in each what you want to know. For example, # of PCs purchased per year; typical buyer title; total IT budget; etc. This will function as your question list when you get to step 3.
- Estimate total # of companies in each cell. You have your 30 cells, now you need to know how many companies there are in each. I'd use publicly available resources to get to this. Bureau of Labor is best; also analyst reports on specific industries cite data and these can usually be gotten free (e.g. if you have a brokerage account, these days they'll give you free research.) You'll never get exact, so the key is triangulation. Get more than 1 data source and build ranges for each cell. Make sure you write down all your sources and assumptions for each in an appendix.
- Qualitative interviews. You don't really know anything until you talk to people. Call into some accounts that meet the criteria for each cell. Get creative, use a phone book. In SMB this is pretty easy; in Enterprise it's not as hard as you think. Just tell the truth, say you're calling from X company and you're doing informal research to help improve service for their company. If the person isn't the right person, get the name of who is. You'll get a 1 in 5 hit rate. This was my first job out of college--it's amazing how willing professionals are to talk to other professionals. Ask them the questions from your grid, 1o max. Get 2-3 (or more if you're having fun) responses per cell. You now have a rough market sizing.
- Quantitative (optional). At this point, you can engage a market research firm to add more quantitative depth to the insight gained from step 3. I'd do 25-30 responses per cell. So, 900 completes max. Survey should be 5 minutes max. You can do this for $30,000. This will get you much better precision to your market sizing exercise, and managers will likely pay for it once they see the insight coming out of step 3.
Once this basic analysis is done, execs will be much more willing to fund more in-depth projects on customer needs; researching how decision making units (DMUs) work; understanding future trends; etc., etc. But you can't do the really interesting work until you've sized the market.