Purpose of Your Survey: Identify the information
needed and from whom it will be collected. Typical objectives
include compare awareness of your product to your competitors’
conducted among your prospects; quantify attitudes and preferences
about your products among your customers; and measure employee
||Use Plain English: When
writing the survey and invitation, use language that anyone can
understand, such as "a survey" or "a few questions"
instead of "research study." Avoid industry buzzwords
and MBA- and tech-speak. They're useful as shorthand, but they
exclude people who don’t happen to know the terms.
When sending an email inviting participation, make your subject
line short (30 - 40 characters including spaces, or 5-8 words),
and include a specific benefit that accurately reflects the nature
of the survey. For example, “Your Opinion Matters to (Your
addition, strive to personalize the invitation. When a recipient’s
name is not available, substitute a term such as colleague,
subscriber or some other appropriate word. Make the invitation
brief but be sure to explain why you’re doing the survey,
how the recipient was selected and how long it will take to
complete. Assure confidentiality and include a deadline.
||Determine Your Sample and Size:
Online surveys can be conducted by self-selection (advertising
on Web site pages, newsgroups or traditional media to find participants)
or randomly (in which the sample is emailed a survey because it’s
representative of the population being surveyed). For random surveys,
you can use your own internal lists or buy them from a broker.
When you use the self-selecting technique, it’s very important
that your questionnaire screen for the type of person you need
The larger the sample, the more reliable the data. For example,
a sample size of 100 has a 10% chance of error; a sample size
of 1,000 reduces that margin to 3%. In addition, if you plan
on segmenting data, you’ll need a larger sample.
||Should I identify the Sponsor of the
Survey or Not?: This is a very basic question. Your
answer will be determined by whether the respondents’ knowing
the sponsor will bias the results or improve the response rate.
||Write a Questionnaire That Gets Answered:
Make sure questions are clear and unambiguous. Requiring respondents
to answer a question can be a real turn off -- require answers
only on rare occasions. Open-ended questions take more thought
and time to answer, so limit their number. A survey of no more
than 20 questions works best.
||Organize the Questionnaire:
Ask general, easier-to-answer questions first. This will prepare
respondents for more detailed questions that follow. Hold personal,
more sensitive questions (such as age and income) until the end.
Organize questions into sections that flow logically.
Response Rates: Response depends on many variables,
including the sample, sponsor and questionnaire design. But
other factors can also come into play. Use an incentive –
offer to include the respondent in drawing for $100 or a Palm.
For each survey completed, you can also make a contribution
to a charity. Offer to share results with the respondent. When
sending by email, send Tuesday through Thursday between 10 AM
and 3 PM. Send follow-ups to non-respondents.
||Take Advantage of What Technology Has
to Offer: Provide some indication of survey progress.
With a mail survey, you can easily skip to the end of the survey
to know how much you’ve completed. Not so with an online
survey, which can seem like an endless stream of questions. So,
give your respondents some indication of how much they’ve
completed and how much more they have to go. In addition, allow
them to interrupt and re-enter the survey. By having access to
results while it’s in progress, you’ll be able to
know if respondents are having trouble answering any of the questions.
to the Details: Before you send, be sure to check
the logic and skip patterns to ensure that the respondent can
complete the survey easily and successfully. For example, make
sure respondents are not asked about a product or feature when
they had indicated no interest. Most commonly encountered errors
in logic and skip pattern can easily be avoided.