Segment is becoming a frequently used data source for many clients. Faraday can setup and pull your customer data from Segment but there are some key considerations to review, to ensure our teams are obtaining the best data to support your predictive models, personas, insights, and more.

Key Points:

  • Segment data is delivered through a webhook. (No historical data, only real-time triggered events)

  • Segment event data for the most part fits the build with Faraday for data we expect.

  • Segment, by default, does not have all the required fields Faraday asks for (e.g. - physical address).

  • Stitch Data (ETL) is utilized as an intermediary step between Segment and Faraday for this integration type.

Guidance on each of the above points:

  • Webhook: only data for newly triggered events will make it to the final destination (Faraday's warehouse). However, for Business Tier customers Segment does seem to offer a service called "Replay" which may be something you wish to explore with them before we begin, learn more about it here.

Why is this important?

In order to build effective models, Faraday requires historical data, often times (at least) for the past 2 years.

  • Events: Segment event data comes to us with tables for: group, identify, page, screen, and track --> we'll want to know more about what you're tracking and how to effectively piece together the above standard and customized field names to construct the whole picture once it's in our hands.

Why is this important?

Event data is the fastest way to make insightful predictions about consumer behavior. It is also structured in a manner that allows us to know exactly what someone was doing, and when. Providing summary or aggregated data (e.g. - "account rollup") is only the tip-of-the-iceberg for understanding distinctions and proves to be exhaustively difficult to reverse engineer into event data.

  • Matching: the Faraday Identify Graph is based at the individual level which required us to match based on First Name, Last Name & Physical Address. Oftentimes Segment does not capture all of these fields that are crucial for high match rates into FIG. Learn more about our matching types here.

Why is this important?

Faraday's match algorithms perform best when exposed to first and last name + physical address combinations. You will most likely already know if you have this level of PII on individuals, and from there its a matter of ensuring its in your source data. Let us know if we can offer insight into alternatives if that is not possible through Segment on your end.

  • Integration: because Stitch Data is part of the integration, we'll need to authenticate the Segment connection with Stitch to begin pulling data. We have observed that this is only possible when the Segment user (account) this is done with has sufficient permissions to do so. The easiest option is to temporarily raise the permission level to Workspace Owner for the setup and then lower it after complete.

Getting started with Segment:

What starting the integration process, we recommend a collaborative call with one of our Customer Success Engineers to walk through the straightforward process and cut down on back-and-forth troubleshooting.

You have the option of creating a Faraday-specific user/role for this, or utilizing an existing account in your Segment instance. Our goal on the setup call will be to:

  1. authenticate the connection to your Segment workspace

  2. create the necessary endpoint in Segment for events to begin flowing through.

To prepare, you may visit the Faraday app at https://app.faraday.io/sources

Simply select "Create source", select Segment and give it a descriptive name and proceed to enter credentials and finalize.

Did this answer your question?