Solved

How do your product teams manage pushing a new event to an analytics tool?

  • 4 January 2022
  • 3 replies
  • 36 views

Userlevel 1
Badge

Hi all,

 

We all know the importance of anaytics and events documentation for a product. At my current company, the PM and engineering team can pass any events that can be published in our analytics tool ASAP. This has led to often seeing people leading to look at the wrong event or property and less confidence in the quality of events and data we are tracking.

 

To solve this problem, our data team have come up with the following proposed workflow.

1. Any time a new event or event property needs to be added to anaytics tool, PM / EM must inform the data team.

2. Data team will approve the event, add it to the event data dictionary. Their SLA to complete this task is 1 week depending on the request.

 

I want to find a optimised solution where engineers can still be able to push the event directly to analytics tool. Definitely, there should be standard procedures that must be followed but adding a sign off process where data team will push an event may result in decrease in team's productivity. IMHO, the sqauds should be able to function automously to produce better outcomes.

 

Any suggestions that can help to find the middle ground will be really appreciated.

 

Many thanks,

Ashima

icon

Best answer by david.morgan 4 January 2022, 16:36

View original

3 replies

Userlevel 5
Badge +10

Hi there, @Ash

Pretty interesting situation you have here. There are a few choke points that seem to be of concern here - mostly, as you’ve mentioned, the development team not being able to directly push an event to the analytics tool. Here are some thoughts and questions - 

  1. Is there a process that validates whether the data/event being added to the analytic tool is holistically valuable to the product and/or the end-users? 
  2. Is there a scenario where engineering could add a new data point or event to the analytics tool without it being thrown into production (in front of end-users) immediately? 
  3. One concern I have is that if the data team is evaluating the event/data point it gives me the impression that the architecture design of the analytics tool can be a bit flaky (as in, new events/data points can cause adverse impact to existing data modules). Is this the case?

Overall, I agree that there shouldn’t be a strong hinderence from the Data Team if Engineering is simply adding the ability to use an event/data point in your analytics tool. On the same hand, if all of their changes go directly into production without QA and evaluation (from Product and Data Teams) - you’re running the risk of providing too many “unnecessary” data points, which can pollute the reporting for your end-users (which equates to a negative user experience). 

The middle ground here would likely be a branch that would only communicate with the sandbox deployment of the analytics tool. This would give the Data Team time to evaluate and test for inconsistencies or negative impacts to existing data modules before it gets thrown in front of your users. 

 

This has led to often seeing people leading to look at the wrong event or property and less confidence in the quality of events and data we are tracking.

 

This comment right here is the most worrisome one. The absolute last thing you want is for your users to start to question your analytics tool (as they often make business decisions from the data they see). Once confidence begins to wane, engagement/adoption will start to decline as well. So the goal is to allow the engineering team flexibility to open the product up to sharing events with the analytics tool and allow the data team time to evaluate its impact/value. 

Userlevel 1
Badge

Thanks @david.morgan , this will really help to initiate the conversation with data team on how event is validated and if it is validated in sandbox before pushing it to production enviornment. We can then brainstorm on we can improve the process.

I will keep you posted on how we go. :) 

 

Best Regards,

Ashima

Userlevel 5
Badge +10

Thanks @david.morgan , this will really help to initiate the conversation with data team on how event is validated and if it is validated in sandbox before pushing it to production enviornment. We can then brainstorm on we can improve the process.

I will keep you posted on how we go. :) 

 

Best Regards,

Ashima

 

Sounds good!

The main thing to keep in mind is that data analysts will want to keep the data users are interacting with as clean and cohesive as possible. Just because an event/metric can be added to the analytics platform doesn’t mean it should be added. I would suggest the Product Team looking at usage of the analytics application, perhaps do a few user interviews to gain insights into how it’s being used (and by whom), and try to identify some metrics/events that will give your users what they need to be successful. 

From a Product perspective, the users should have the utmost trust in the data that they’re interacting with and feel emboldened to make decisions based on that data. Even if it’s a “soft touch” directional decision, it’s still something that the users are looking for and need. The engineers shouldn’t be overly restrictive in what they can add, but it needs to provide value to the users and/or the platform. :grinning:

Good luck! I’m eager to hear how things work out for your organization.

Reply