4 ms·

> You can talk to your users in person and ask them doesn't scale. > poll them via email doesn't work. > default-spy on their customers analytics is not sp

by kmlx 3y ago

> You can talk to your users in person and ask them doesn't scale. > poll them via email doesn't work. > default-spy on their customers analytics is not spying.

quickthrower2 3y ago
you don't need scale, you need good data
- charcircuit 3y ago
  Good data requires statistical significance which needs scale
  - throwaway71271 3y ago
    you only need 5 people to test poison
pydry 3y ago
You can poll them via pop ups. Ive seen this done a couple of times. The sites took care to make it unobtrusive, sensibly placed and with short, meaningful questions. Frankly I think the data they got drove better decisions than analytics. I liked that they let you write a message too.
- GordonS 3y ago
  I like this method. Earlier today I filled in some gov forms online, and at the end it asked if I'd like to do a short survey on my experience - I was happy to oblige, because I have a personal interest in service improvement.
tourmalinetaco 3y ago
>doesn’t scale Yes it does. Have you never heard of a survey? >doesn’t work It absolutely can. For instance, polling via newsletter is rather effective. >analytics is not spying. If it contains any personally identifiable information that is not required by law, then yes. Yes it fucking is.
- NegativeK 3y ago
  I'm not in agreement with the top comment's cookie needs, but no. Surveys are not comparable to using analytics/tracking to determine user behavior. Surveys measure what people say and are sporadically filled out. Cookies measure what people do with granularity and accuracy and will have near 100% coverage. People aren't concerned about spying via survey because they're not even in the same realm of effectiveness.
- jobs_throwaway 3y ago
  > If it contains any personally identifiable information that is not required by law, then yes. Yes it fucking is. wahwahwah
vharuck 3y ago
>doesn't scale. Much of the math behind surveys is based on finite samples taken from infinite populations. It scales really, really well. You just have to make sure your samples aren't totally biased, which is a little more difficult. But there are ways to reduce those problems, if you're aware of how the bias is introduced.

by kmlx 3y ago

> You can talk to your users in person and ask them doesn't scale. > poll them via email doesn't work. > default-spy on their customers analytics is not spying.

quickthrower2 3y ago

you don't need scale, you need good data

charcircuit 3y ago
Good data requires statistical significance which needs scale
- throwaway71271 3y ago
  you only need 5 people to test poison

pydry 3y ago

You can poll them via pop ups. Ive seen this done a couple of times. The sites took care to make it unobtrusive, sensibly placed and with short, meaningful questions. Frankly I think the data they got drove better decisions than analytics. I liked that they let you write a message too.

GordonS 3y ago
I like this method. Earlier today I filled in some gov forms online, and at the end it asked if I'd like to do a short survey on my experience - I was happy to oblige, because I have a personal interest in service improvement.

tourmalinetaco 3y ago

>doesn’t scale Yes it does. Have you never heard of a survey? >doesn’t work It absolutely can. For instance, polling via newsletter is rather effective. >analytics is not spying. If it contains any personally identifiable information that is not required by law, then yes. Yes it fucking is.

NegativeK 3y ago
I'm not in agreement with the top comment's cookie needs, but no. Surveys are not comparable to using analytics/tracking to determine user behavior. Surveys measure what people say and are sporadically filled out. Cookies measure what people do with granularity and accuracy and will have near 100% coverage. People aren't concerned about spying via survey because they're not even in the same realm of effectiveness.
jobs_throwaway 3y ago
> If it contains any personally identifiable information that is not required by law, then yes. Yes it fucking is. wahwahwah

vharuck 3y ago

>doesn't scale. Much of the math behind surveys is based on finite samples taken from infinite populations. It scales really, really well. You just have to make sure your samples aren't totally biased, which is a little more difficult. But there are ways to reduce those problems, if you're aware of how the bias is introduced.