Guidelines for applying segmentation to technology companies 

Editor’s note: Alan Nazarelli is president and CEO of Silicon Valley Research Group, San Jose, Calif. This is an edited version of a post that originally appeared under the title, “Applying segmentation to technology companies.”

As a San Francisco Bay Area market research firm with technology company clients, we wanted to address how segmentation applies to companies selling technology products and services. Most of the classical concepts apply to companies selling to consumers, so we won't address those here. But what about those selling to businesses? Here are four guidelines.

1. The variables you will be computing are different. 

The following are important dependent variables to consider: 

  • Company type.
  • Industry classification.
  • Number of total employees. 
  • Number of information workers.
  • Size of IT department. 
  • Size and nature of development teams.
  • Computing and development environments. 
  • Application platforms.
  • Maturity (start-up, growth, mature). 
  • Key customers or industries served. 
  • Purchase criteria and processes.

As you can see, these are different from traditional demographic and psychographic variables used for consumer segmentation.

2. Should basic demographic variables be collected for business segmentation? 

Yes. Respondents age, gender, education, etc., may be relevant to the segmentation map, as well as general demographics of the company's workforce. These can sometimes be collected by asking proxy attitudinal questions – for example, social media savvy of the company’s employees using a numerical or Likert rating scale. 

sample size concept3. Sample size. 

As any business-to-business researcher will tell you, business-to-business sample sizes will be relatively smaller than consumer sample sizes. There are simply fewer businesses in any category. Attempts to over sample often result in a poor quality sample being collected. Researchers frequently end up surveying non-decision makers and influencers whose opinions don’t generally reflect how the company does business, and hence they risk getting erroneous data. There are a two techniques I find useful in these situations.

a. Include plenty of qualitative methods in the research design to enable in-depth insights on the different segment clusters to be gathered. Ideally, these should be placed both prior to the quantitative survey and after to enable validation of quantitative findings. Frequently, quantitatively obtained independent variables need to be explained and the qualitative efforts help explain the "why” behind the observed phenomenon.

b. Design the project flexibly to allow for additional sampling to take place after the initial regression analysis and cluster mapping has taken place. Any interesting subsegments with a small sample size may be beefed up to ensure that the initial hypothesis still holds about that cluster. 

4. What are the ideal sample sizes for B2B segmentation?

The rules vary. I have seen poorly crafted and ineffective B2B segment maps based on larger sample sizes. In contrast, I have also seen well executed and effective segmentation based on qualitative data for certain audiences such as highly specialized neurosurgeons. However, as a general rule, a sample size of 350 to 500 should be considered effective, providing a sampling error of about 5-to-6% at the 95% confidence level. That means that 95% of the time (or 19 times out of 20) the sample results will be within 5-to-6% of the true result for the total population.