We do something similar to what you are asking, with a slight twist (just a little logic because we don't want surveys 100% of the time, obviously it would be easy to just send all the time). Because my answer is rather long, I have just copied what I have posted before ().
(we are using assessments and not legacy surveys, but a solution below should be able to be adapted)
1) Upon closure of an Incident or RITM, we determine if a survey is required. For example, we do not want to survey students and there are certain assignment groups that do not receive a survey. We use a flag on the TASK table that we fill in if a survey is required.
2) Our notifications are two-tiered. We send a normal "Your Incident has been resolved" if our "Survey Required" flag is false. This comes off the Incident table. If "Survey Required" is true, we send a notification off of the Assessment Instance table. Because the Assessment Instance has "Trigger ID" (a field that contains the sys_id of the record calling it), we use a Mail Script to lookup the associated Incident or RITM information based on the sys_id. We then send out a "Your Incident has been resolved" email similar to above, however it is from the Assessment Instance table. We include our survey specific information in it (we use some cool graphics to entice them to click on the survey) and the Incident/Task information, which is obtained through the mail script.
3) Client takes the survey, we then do a custom table to transform the results into a format that is much easier for management to handle (a table where each row contains the questions/answers, rather than the ServiceNow format of each row being a question). We have some reporting we need to do based on the results of a specific question, and this kept it the "prettiest" for management and the most functional for our user base. The table just updates upon a survey changing to "Complete". Because we have the "Trigger ID" value (which again is the sys_id of the incident/task that called it), we can reference who the analyst was that did the ticket, the business service, etc.
This survey is limited to the client themselves (only they can fill it out), and clicking on the face translates what they selected into the first question (whether you are happy, neutral or upset determines what questions you are asked). Our response rate was 12% last time I checked, but I have not checked in awhile. We have been told by our research people that 12% is a pretty good response rate for electronic surveys that are unsolicited.
We have found that the survey module required more work than we assumed it would, so I would be happy to share any information with you to help you get this up and running. You can feel free to message me and we can chat over email/phone - I have helped out a couple other people who wanted this sort of functionality in their environment. For some reason SN didn't make it as easy to do as I think it should have been to have this work in many real-life scenarios.
Below is a copy of our email, I believe this is what you are ultimately looking for?
