Although conversations with customers are always informative, anecdotal customer input might not be systematic enough for decision-making. Why don't you carry out your own primary market research? It is exciting to conduct a customer survey on your own. Here are some pragmatic suggestions.

1. Define the survey objectives. When a survey is being carried out to collect information for resolving specific marketing or business issues, the research objectives are logically apparent.

   On the other hand, when you conduct a survey to gain insights into customer behaviour, you still need to define the objectives. You need a focus.

   Choose a few areas that are very pressing and important to you. Decide, for instance, if it is going to be a survey on product usage, purchasing practices or customer satisfaction, or if it is going to be a combination of all of them.

2. Establish a research budget. As a business undertaking, a survey must have a budget. Please, don't underestimate the value of research by setting an artificially low budget. If the budget is so small that it becomes difficult to execute the survey properly, you would get poor results. Beware of a self-fulfilling prophecy here.

   Look at your advertising and promotion budget, and set your research budget as a factor of it. It could be a factor of one or more!

3. Set a sample size. How many customers are you going to include? It is much more manageable to survey a few hundred.

   When your customers number in the thousands, develop your own random sample. For instance, choose every fifth (one-in-five) on your customer list. If you have 3,000 customers, you will have a sample of 600.

   Depending on your survey objectives, you may wish to base the sampling criteria on other factors. If you are interested in determining the needs and wants of new customers, new customers would be your sample. Then, you have to define what "new" is.

4. Consider whether you should offer incentive. Not every customer will cooperate. Those who like you and those who don't like you will very likely respond to the survey. For everyone else, some incentive would often increase the rate of response. You might wish to offer a draw for prizes.

   Incentive often raises the response rate by as much as 15 percentage-points. On average, a customer survey yields a response rate of one-in-three (33%). With incentive, you might see it rise to 48% or 50%. For a sample of 600, you would get almost 300 completions.

5. Choose a survey method (fieldwork) best suited for your customers. How do you communicate with your customers? Do you contact them by telephone, fax or mail? Survey your customers through the most commonly used communication channel. Use what you have successfully used for customer communications as your survey technique.

   Telephone interviewing is still the most effective data gathering method. Mailing and faxing yield lower rates of response. E-mail and Web-based techniques only work with certain types of respondents.

   Each method has its own associated cost. Telephone interviewing and mailing are more expensive than the others.

   The Five Don'ts of Questionnaire Design 1. Don't ask too many open-ended questions. Keep them to a maximum of three. Many respondents skip such questions. When conducted over the phone, you would just get dead air. 2. Don't ask a yes-and-no question, unless it is truly a black-and-white situation. Consumer behaviour is in different shades of grey. You use research to find these shades. 3. Don't combine two questions in one. "Which toothpaste do you use or do you change brands often?" is a good example of a bad two-in-one question. 4. Don't ask factual questions that require respondents to look up records and invoices to report exact amounts. A questionnaire is not an inventory tally sheet or an income tax return! 5. Don't use jargon and slang. Customers might not be familiar with the language you use in your organization.

6. Design a questionnaire directly related to the survey objectives. Don't ask everything under the sun; only ask questions that directly relate to your survey objectives. Only ask those questions relevant to the survey objectives. Besides, the longer the questionnaire, the poorer the response rate.

   Although questionnaire design is more complicated than generally recognized, it is possible to design a functional one that serves your purpose. Follow these steps:

   • Identify information items that address the survey objectives.
   • Compose the questions with simple and precise words.
   • Arrange the questions in a logical sequence.
   • Use a page layout that is easy and intuitive to follow.
   • If feasible, pretest* the questionnaire.
   For a telephone survey, write the questions in such a way that the respondents can understand (by listening only) what is being asked.

7. Look for patterns, exceptions and segment differences. After collecting the responses, the last step is data analysis. Spreadsheet and database applications are not ideal for tabulating survey data. With in-house expertise, a statistical program such as SPSS delivers the processing power of the tool of the trade. Or, SumQuest that interacts with the user in plain English will meet your needs very well.

   In data analysis, while averages, mean ratings and frequency distribution are essential in spotting patterns and trends, please actively look for exceptions. Deviations from the norms represent opportunities and flag areas where attention is warranted.

   You might wish to classify customers by segment or subgroup. Use criteria such as a customer's regional location, frequency of buying, annual consumption, length of relationship, and industry type. Segment analysis reveals much more about the variety, subtlety and complexity of product usage, preferences and attitudes.

   Be sure to interpret the findings and consider their implications through the lens of experience; use your business insights and industry knowledge. For anything surprising, look further into it.

Should You Seek Outside Help?
Sometimes, there is just so much you can do on your own. When outside assistance is considered, use a researcher for questionnaire design and data analysis. In these two areas, the right skills help tremendously.

Final Words
Document the results. Believe in them, and implement what you judge is sensible. After two months, review the findings. You will be pleased at how well you did.