Ideal Panel Size Q&A
Posted by Jeffrey Henning on Fri, Mar 12, 2010
Q: What is the ideal panel size?
A: 100% of your target audience.
Q: Um, thanks. What's the ideal panel size if you have 300,000 customers?
A: 300,000 panelists.
Q: You're not very good at this Q&A thing, are you?
A: No.
Q: What's the ideal panel size if you can't get every customer to join your panel?
A: I thought you were an idealist?
Q: You're not allowed to ask me questions.
A: Sorry. The "ideal" panel size is a function of many factors, including:
- How many separate studies you need to conduct each month
- How many times a month you are willing to survey customers
- How many responses you desire per study
- Your survey participation rate (if you don't know it, assume 20%, which is an optimal response rate)
- Your completion rate (assume 90% to be conservative)
Let's work through an example:
- Studies per Month = 4
- Panelist Study Limit per Month = 1
- Desired Sample Size = 1,200 responses (400 responses per segment, with 3 segments)
- Participation Rate = 35% of panelists invited to a survey start a survey
- Completion rate = 90%
Plug those answers into this handy formula:
To get 1,200 complete responses per study, you need to invite 3,810 panelists (Desired Sample Size ÷ Participation Rate ÷ Completion Rate). Multiply that by 4 (the number of studies a month) and divide by the number of times you are willing to invite panelists each month (in this case, 1). Your ideal panel size is 15,238 panelists.
Once you have built a panel, you will find increased demand for surveys to be fielded to that panel. As a result, you will want to grow the size of the panel over time. Some months you may need to settle for smaller response sizes per survey in order to handle all the surveys that you want to field. Resist the temptation to increase the number of surveys you send out a month, as that will produce diminishing returns and will ramp up the rate at which people leave the panel. (In other words, increasing the panelist study limit per month will decrease the observed participation rate, something not modeled in the formula above.)
Q: You're enjoying this, aren't you?
A: Don't you think it's a happy day when you get to use the Equation editor in Word?
Q: Again, you're not allowed to ask me questions.