Generating insights with partial data

  • 28 April 2020
  • 3 replies

Hi all.


We develop a desktop application. Some of our users will run our product behind a firewall and therefore will not be able to submit events to Mixpanel. We don’t want to prevent these users from using our product. However, having an unknown number of users who don’t submit any events to Mixpanel will create a bias in the data.


Is there someone here that defined an analytics plan for such a product?

Are there any best practices / methodologies to mitigate this missing data?

I want to understand if/how I can infer something on features/users if I have only partial data.


Thanks in advance, Eran.

3 replies

Userlevel 4
Badge +4

Hi @eranro

It sounds like your customers' firewalls are either blocking all outbound requests to non-whitelisted websites or the firewall is preventing the Mixpanel CDN from loading.

The best option is either to implement a completely server-side integration or ask your customers to whitelist Mixpanel.  I also wanted to share a helpful article on how to track geo-location server side in the case that you do end up implementing a complete server side integration on your app. 


Hope this helps!


Hi @stephanie .

In some cases, our entire application (both client and server) are placed behind a firewalls that blocks outbound requests to non-whitelisted websites.

My question is, given that fact that some users use the product and don’t submit data to Mixpanel (or any other analytics tool), how are we supposed to make data-driven-decisions based on partial data?


Thanks in advance.

Userlevel 4
Badge +4

@eranro What you have describe is a common challenge of tracking within the analytics space. You are not alone – these limitations affect everyone that cares about product analytics! Firewall, ad-blockers, opt-outs, and GDPR give users the power to choose if they want the data to be collected or not, and respect that right.

I do think it is worth mentioning that usually not every bit of data is needed to make data driven decisions. As long as you have a good amount of data that is representative of your user base, you can run reports to help answer your questions. With this in mind, I am curious to know what are some of your KPIs? Maybe with some knowledge of your KPIs and events you are tracking, we can suggest some reports to use to help you answer your questions.