Statistical Sampling Simplifies User Satisfaction Measurement

Share on facebook
Share on twitter
Share on linkedin
Share on email

statistics-graphicHere at Unify Square we firmly believe a truly successful Lync deployment lies in end-user satisfaction. When users are happy with Lync, they improve collaboration and reduce the cost of communications. The user-satisfaction mantra drives everything we do and we think it should drive you, too. Getting your users’ opinion is a good way to measure success and can help focus improvements and investments. But, as soon as you want to start listening to your users you will face the question of who to survey.

If your user base is around 100 employees or less, the answer is usually pretty straight forward: Just ask everyone. This is more of a census than a survey, but the small population size makes it feasible.

For bigger corporations, setting up a survey is a bit more involved. For example, let’s say you have 30,000 users. Inviting everyone to take your survey might not be such a good idea. There are obvious challenges which make targeting a survey rather difficult. Users are in different time zones and locations and they possibly speak different languages. There are also the less obvious productivity costs of running a survey. Take our example of 30,000 users taking a quick, one-minute survey. Just this one survey adds up to a whopping total of 500 work hours.

Luckily, there is no need to survey tens of thousands of users at once. The reason is a statistical concept called “survey sampling.” According to Wikipedia, survey sampling is the process of selecting a sample of elements from a target population in order to conduct a survey. Even if you’ve never heard this term before, you have certainly seen it applied in everyday life. Anytime you hear a story on the news claiming that “48% of the population is in support of [something],” a survey sample was used to get what is considered a statistically relevant result.

So, how did the surveyor arrive at the result? Did they go out and ask every single member of the population about their opinion? Certainly not. Instead, they only asked a few people, and extrapolated the results to be representative of the entire target population. This is a common concept, and a widely used and accepted practice (exit polls after elections are another good example) when it comes to surveys.

Now you are thinking about the number of people you need to survey to get an accurate result. Your number needs to hold up to the scrutiny of your peers and the boss. This is where “Sample Size Determination” helps guide your survey program. You can read all about how this complex formula works on Wikipedia. For the purpose of this blog post, the short version of the equation is that you need two simple data points to choose your sample size:

1. How many people are in the total population?

2. What is the level of certainty you want that the results represent the total population?

The first data point is straight forward (30,000 in our example), but the second data point requires explanation. The level of certainty is called the “Confidence Level.” In regular speak, the confidence level lets you claim “We are 96% confident that the gathered results are representative.” As a surveyor, you select the confidence level you desire. This selection directly influences the number of people in the survey sample size. Let’s take a look at how many of the 30,000 users we’ll have to include in our survey, given a few common confidence levels:

(This table assumes a standard margin of error of 5%)

As you can see, even if we go for the highest possible confidence level, we only need to get the opinion of 650 users to achieve a 99% representative result. Interestingly enough, even for a huge population of 300 million, the sample size only increases slightly to a total of 664 users.

By using a statistically relevant sample of the user population, you can save a lot of time and get your survey results quickly.

Next time you are shopping for a survey tool to get the pulse of your user population, make certain it can help you determine your best sample size. Unify Square PowerSat is a terrific choice to implement a regular program of measuring user satisfaction. PowerSat uses the convenience of Microsoft Lync Instant Message to engage users when they are available, yielding extraordinary response rates. Better yet, when you set up surveys in PowerSat, the system lets you choose your desired confidence level and handles the sample size determination for you.

Regardless of the survey tool, our Lync experts suggest checking user satisfaction monthly and tracking the number as a KPI. Using survey sampling enables your team to quickly and consistently track the KPI, monitor trends and take action when needed. The end result will be happier users and improved ROI for the Lync initiative.

Shopping Basket