It’s not perfectly up-to-date, as I made this almost 2 years ago now, but this is how we are putting Segment at the center of our world.
Sources We currently have 16 sources, mostly various front and back end engineering systems, python, ruby, js as well as our ios and android apps. In our typical setup, we have both a staging and prod system connected for testing, but do not connect our staging into the Engage portion to keep data clean.
Warehouse & BI All of our data flows into Redshift and we visualize in Looker. This is a work in progress for sure, but every day we add new reports to get ourselves more campaign performance data.
Destinations This is where it gets fun in my opinion. We have everything from FullStory to Braze, Branch to Hubspot, Zendesk to Survicate, Google Ads to Optimizely. Needless to say, its a handful. In total, our roadmap is to get to 20 tools.
SOOOO….What’s in your stack?
Page 1 / 1
Thanks for sharing that Chris. With the implementations I’ve done, we use mostly the same architecture with a few changes/additions to supercharge it.
CRM
We normally will plug in the CRM as a destination (native) and as a source (see below) so we capture full user interactions and can share with your sales/success teams.
Unlocking Attribution
We built a custom solution that sits on top of Segment to provide multi-touch attribution and media mix modelling in a quick and simple way as it completely leverages Segment. It also provides a new set of sources for Salesforce event flow (eg: Lead created, etc) and new paid media sources for Twitter, Capterra, etc (today, Segment only provides soources for Google Ads and Facebook Ads).
Product Analytics
Alongside Amplitude we’ve witnessed MixPanel really improve their offering and pricing. Worth considering them too.
In-app engagement
Often we need to solve for activation or engagement in web apps. In product engagement tools (eg: Intercom, Pendo, WalkMe, AppCues, etc) are great for this and we normally have them included.
Makes a ton of sense. We don’t have a sales team, which is why we have that huge void in CRM, but definitely understand that value of using it as a source when applicable.
Good to know re Mixpanel - I honestly don’t remember the last time I have considered them as viable. Definitely will be on my radar.
Very interested though in learning more about what you are doing in multi-touch attribution. We just updated / or are about to update our Capterra listing, and have a ton of paid ad, and affiliate referrals. Our business model has changed recently from a free product to more of a SAAS model, and so looking for ways to better attribute our marketing efforts. We very much are just last touch - which has its benefits and drawbacks.
Asked our Data Engineer to draft ours up to share!
Here's a simplified view of our customer data stack built on our core Connections, Unify and Engage, plus our warehouse interoperability features: Profiles Sync and ReverseETL. We power hundreds of distinct workflows, but some core internal use cases include:
Lead processing, attribution and marketing automation - capture high-intent behaviors on segment.com (such as requesting a demo), leverage Unify’s identity resolution capabilities to tie them to a known lead profile, and send them downstream for real-time routing to sales or marketing automation tools. Further enrich leads and conversions with additional data points from the data warehouse.
Customer engagement at scale - construct concise and highly-detailed profiles of our customers from a combination of real-time interactions and account-level data from the CRM or data warehouse. Empower the Customer Success team to construct self-serve audiences and segmentations on top of these profiles, and activate them in downstream customer engagement tools
Product analytics - connect live event sources (segment.com, app.segment.com) to product analytics tools, and enrich with additional context from backend databases or data warehouse models. Surface highly contextual insights on customer product interactions to Data Scientists and Product Managers.
Segment’s Data Stack
Segment Source Layer
Leverage Segment’s source libraries and the Segment Tracking API to translate real-time customer interactions on key data sources into clearly-defined “identify”, “track” and “page” events. Events are captured and sent through three parallel streams:
Cloud object data sources, such as Salesforce, are automatically synced to the data warehouse using the Snowflake connector with selective sync.
Segment Composable CDP Layer
The CDP (Unify and Engage) combines data from event sources, cloud object sources and internal data warehouse models. In addition to events, “composability” features such as Reverse ETL and Linked Audiences enable CDP users to access the entire scope of customer data available in the warehouse, including advanced AI/ML models built by Data Scientists. Unify’s built-in identity resolution capabilities ensure that data from these disparate sources can be unified into concise customer profiles.
Internal CDP users leverage Engage to build additional (self-serve, UI-based) layers of processing on top of profiles, such as audiences, computed traits, journeys and “Customer AI” predictions. Profiles are also sent back to the data warehouse for additional processing/analysis using Profiles Sync.
Segment Integrations Layer
The final step, after capturing events at the source, combining data into profiles and building audiences/segmentations, is to activate the data in downstream tools. CDP users map and configure integrations with destination APIs through the Segment workspace UI. Events can be sent downstream through the following paths:
