Measuring user satisfaction with rolling NPS
Incorporate NPS surveys into your project to have maximum insight
tl;dr Our latest project (django-nps) allows you to run continuous, rolling NPS surveys, and to segment NPS by any user attribute.
NPS (Net Promoter Score) is a well-known methodology for capturing user 'satisfaction' with your product. I won't go into the details here - there's a wealth of info elsewhere on the web - but suffice to say it involves asking your users a single question ("How likely is that you would recommend our product to a friend? (0-10)"), and then categorising their response into one of three groups - "detractor" (0-6), "neutral" (7-8) and "promoter" (9-10). The difference in the % of promoters and detractors is your "Net Promoter Score".
There are dozens of polling services online that support NPS - like other survey services you can set up a poll, run it for a period of time, and they will do the calculation for you. These services are great for running a single poll, but the power of NPS really comes from regular surveys - whether your NPS is 10, 20, or 50 is less significant that whether it's moving up or down over time. If you have a low score, are you addressing the issues, and if you have a high score are you maintaining the experience? (NB there is no such thing as an absolute "good score" - the theoretical range is -100 to +100 - but there are some benchmarks for different industries on the Net Promoter Network website.)
If you mean to run regular surveys, you then run into the problem of how to prevent spamming users, and how to segment the results. This is only really possible with deep integration into your own project / user base.
We recently started polling our users with our own NPS app (django-nps) on a continuous, rolling, basis. By regulating the rollout of our survey (using the excellent django-waffle feature flags), we can get a statistically significant sample of our users surveyed every month (possibly every week); no user is polled more than once every three months, and because we are polling our own users, we can segment the results any which way we want - employers, freelancers, designers, developers, and so on. We never need to remember to "run a survey" as it's running all the time - and the results are continuously updated.
The project is over on our Github page - pull, fork, comment etc.
Making Freelance Work