Hello all, can you please share your experience of using Segment?
We currently use Amplitude as a product analytics tool and GA4 for analyzing digital marketing sources like Google Ads, Facebook Ads, TikTok Ads, Mailchimp, SEO stuff and etc. The data is injected into a data warehouse within ETL/ELT pipelines; setting up the reverse-ETL processes and implementing data quality middleware is on the roadmap.
A couple of benefits I learned about Segment
It allows user IDs and session data to be more easily shared across different tools.
Mitigate the vendor-lock risks, being a data-proxy
Two features I found late in the game but liked a lot were “grouping” which is useful especially for B2B where you usually have an “account” with many “users” …it’s an easy way to send that data and be able to look at what both accounts and specific users at accounts are doing in other tools…not just at the user level.
The other is the ability to take from an anonymous ID to a known ID. It’s tricky but you can use this to track the full lifecycle of a user by putting segment on your marketing site when they first go onto your site they’ll get an anonymous ID and when they sign up and you send their ID you can match the anonymous ID to the known ID and then understand what path they took to signing up. …one other tricky caveat to this is the pricing for MAU when you put it on your marketing site.
Hello @Marco, Thanks! Scan you elaborate on the first one please? Don’t see an issue if I assign the account as a user property in Amplitude / Heap / Mixpannel
I’d highly recommend it. I’ve used it at 3 companies over 5+ years now with varying complexity of use cases and it worked well for all of them.
Segment’s key value prop is reducing engineering time spent on analytics implementation and maintenance. In theory, you implement one Segment event and then most tools are a few clicks to set up, and some can be moved server side.
It does this well, is easy to use for marketing ops, simple to implement, and you can use GTM for any edge cases.
Both pros you noted are correct - lock-in wasn’t bad, it does make identity management easier (as long as the spec is followed closely, which is occasionally not obvious)
Only case where it didn’t excel was more complex data mapping use cases - e.g. we need to pass event property x to field y in FB Ads or custom dimensions in GA. That logic can sometimes get tricky enough to require a workaround by using Segment to load GTM to load your desired tag.
Depends. If you’re a consumer app that could be fine where 1 person = 1 user.
For a B2B app a common scenario is where you have a company and that company has many users. Let’s say Company A has 2 users. So when you are looking in amplitude for example and you are looking at user level data. You can see that user 1 did something and user 2 did something but you wouldn’t really have a way to slice that data and answer the question “which companies had active users in the last 30 days” because the data didn’t come with a “group” identifier.
A more concrete example was from my old company where when a new user was created we’d do an identify call with a group call and that would send to our destination for our customer success tool (Custify) and since we were using a group call that user would automatically be associated to the proper company inside of Custify.
Chiming in, as a long-time Segment user at different companies. Your Segment experience will depend on:
how you integrate it
what analytics state you currently have
what reporting level you expect.
I have seen it working well, especially for client-side data reporting supporting marketing. I have seen it not working well, quite often for server-side data integration.
I wouldn’t use Segment to build my reporting for revenue, transactions, payments. I’d use it for activity data for example. But again, depends on your current architecture, data state.
