Skip to main content

Hey there!

What are best practices for distinguishing “identical” events in amplitude? E.g. we have the same object and action (user-account-create) on several pages. We are currently seeing two options:

(1) naming convention: add prefix to event name to indicate where in app event happened
(2) track as property

What are best practices here, and why?

Very much depends on your use case/ scenario. Key thing to keep in mind is that the user pathing features in Amplitude (“Pathfinder”, “User Journeys”, etc) will display unique event names in the paths...so if you only have a common event name everywhere that’s what you will see in the pathing. If you have a good event schema, you’ll be able to see the page/ screen it occurred on in the path from the previous event (which might be something like “view sign-up screen”, etc).

For things like log-in, account creation, etc, you’ll likely find yourself having a single event name with an event property passed to grab the page/ screen it’s happening on. These articles are pretty useful to help you define your schema:

https://help.amplitude.com/hc/en-us/articles/206404618-Step-1-Instrumentation-pre-work

https://help.amplitude.com/hc/en-us/articles/5447814048795-Data-taxonomy-playbook-part-two-Events-and-properties


Hey @Gerrit Mewes this is a great question, and @dangrainger‘s response is a valuable way to consider which route to go. 

Either option requires data education so everyone with access to your org knows how to properly use the data.

When I consider taxonomy naming conventions, I consider 2 things

  • How will the data be used - analysis in select chart types, event streaming or cohort syncing to a downstream destination, etc. 
  • How intuitive will the data be for all others in the org to consume it

Data ease of use is really important. The more concise your event naming conventions are, the less confusion you’ll likely cause downstream. Having multiple “identical” events with a different prefix might cause blindspots in how end users apply the data to their insights / targeting. Unless if your team plans to upkeep an “official” Custom Event that combines all the identical events into 1 aggregate event for analysis. 

If you’re tracking the same exact action across multiple sources, I would recommend the simplicity of 1 consolidated event with some event properties to identify the source of the event. Sample spec below

"event_type": "account created",

"event_properties": {

"account name": "xxx",

"account id": "xxx",

"source page": "home page",

"source url": "homepage.com",

"source CTA": "Apply now" },

 

2 more considerations:

  • there are event_type limits per project to keep in mind
  • all event based charts, including pathfinder, enable you to group by / break down events by their properties, so you’ll continue to benefit from both the high level aggregate analysis, as well as the granular source analysis

Reply