Adjusting responses by page/section | XM Community
Skip to main content
Being a postal website we have certain sections of the site where we get a ton of visitors and therefore a ton of survey responses for our site-wide pop-under customer sat survey (i.e. 9K a month), meanwhile other pages we only get a trickle of responses coming in (i.e. 100 a month). Is there a way to even this out a bit that wouldn't necessarily affect our scoring? The only way I can think of is by setting separate intercepts pointing to the same survey for each of the sections and having their own sample rate dictated. Is there a cleaner/easier way that I'm missing? We're not looking to necessarily have an even rate of distribution but just to lessen the gap a bit. This survey is used for KPI scoring so we are worried anything too drastic might affect our results. Any thoughts or suggestions?
Hi @KristyP - You basically described the same solution that came to my mind, except I think you are envisioning multiple intercepts to manage. In reality you could throttle the main page where you are getting so much feedback by applying page-specific (or pages if that is the case) sampling logic.



To do this, create a new action set (just copy one you already have built) and add 2 things:

1. Page specific logic

2. Sample logic (you will need to find out what sample works for you based on the responses you receive, but you can probably ball park it. If you want 900, instead of 9000 responses, set the sample to 10%)



Below are some images to show you where you can add these things. The font is a little small on the display logic, but I selected Current URL ID "Your Specific URL". You can expand this with OR statements if you have multiple pages.



!



!
Thanks @Akdashboard this is very helpful and will have to be something we toy around with on some of those peskier busy pages. We have the chief concern of how these things will affect our scoring...we've had issue in the past so the team is a bit hesitant on any changes as we don't want to see a complete 180 in scoring. My guess is we should be ok as the initial sample group remains the same (i.e. randomly intercept XX% of all users) so pairing the page down with an additional sampling shouldn't change our core sample group, just lessen it.



I think anyway 😃

Leave a Reply