Accelerate Your Analytics Maturity w/Segment

In June of 2015, I started at WayUp as the Head of Product and sole product manager. The company was then known as Campus Job and when I met the team, they had just ended a stint at Y Combinator and were showing promising progress towards their mission of democratizing the hiring process for college students and recent graduates. Previously, I had consulted with many startups and I always found myself in this weird quasi-surrogate parent role where I was helping businesses launch their products but not exactly getting to see them mature and evolve, so to say that I was excited to start at WayUp is the understatement of the century.  

The business was still in its infancy, so there were product extensions yet to be built, like our mobile app, and lots of fun challenges to tackle. One of those was establishing the beginnings of a data pipeline where we could measure and analyze feature behavior and roll that back up into our top level KPIs. Having always been a Mixpanel fan, my first instinct was to use their platform to measure everything, but just using Mixpanel was extremely limiting. We had sales, marketing, data, support, engineering  and product teams that cared about this data too, and we needed to be able to send it to a number of different analytical tools as well as to our own data warehouses. We had already ruled out adding the tracking code for each new analytics tool individually, so we needed to find a better solution that would work for everyone — engineering team and other stakeholders included.

Our Process

We had a hodgepodge of analytics events—some backend, some frontend, some pageviews, some event-based analytics—and we were at an inflection point. We only wanted to establish our pipeline once, so we needed a solution that could power the entire company. The right solution for us had to:

• Allow us to measure our top line KPIs - our key metrics were being tracked too informally, and we needed to establish a single source of truth.
• Work across business units - if the data team wanted to build dashboards they needed be able to do it, and if product team wanted to do funneling in Mixpanel they needed be able to do that too. We really wanted to democratize the data.  
• Provide support beyond pageviews - we had an single page app (Angular) at the time, and relying just on the pageview tracking that’s baked into most analytics tools by default would often require (even more) custom configuration to get things working correctly.
• Support backend as well as frontend analytics - as our technology changed, we knew we would have to adjust our implementation on the fly.
• Work cross-platform - we were building a native iOS mobile app at the time, so both web and native mobile support were essential.

Finding something that could fit the bill on all of the above was going to be a challenge. We didn’t want to build it from scratch ourselves, but we were a small startup and we didn’t have a lot of resources or budget to dedicate to our data pipeline, so whatever tool we used had to be able to be run by a skeleton crew.

Enter Segment

Segment circa 2015 was much different than Segment in 2017. There was no Warehouses product yet, and data sources were organized in a fundamentally different form. However, the concept that Segment boasted, and still leads the space in today, was something very simple, but novel and truly powerful—one consistent universal source to capture our analytics and send them to whatever tools we desired to integrate  with: CRMs, product management tools, email tools, error-tracking tools, and more.

Segment additionally had the following benefits:

• Phenomenal documentation
• The SDKs supported Javascript, Python, and Swift, among others, allowing for consistent implementations across our web, backend and mobile apps
• We no longer had to support and maintain libraries for every tool we needed because Segment took care of loading the integrations we turned “on.” This was done by loading the libraries asynchronously at runtime — literally by flipping a switch in the Segment UI. This actually meant we could even add or remove third-party analytics libraries without deploying new code.
• They supported multiple event types with “track” and “page” types.
• The JS and Swift libraries had a very small footprint, which was key for us
• We have a number of different digital properties that exist on subdomains, so it was nice that Segment supported cross subdomain tracking right out of the box.
• Out-of-the-box debugging tools

The implementation was seamless. We outfitted our iOS mobile app with Segment first, and once we iterated a bit on when and how to alias and identify users, we then verified all of our events (over 300) and moved to implementation on our website. From implementation to verification, we completed the retrofit in a little under 2 weeks, which was only possible because of the seamless and robust featureset that Segment offers.

So where are we today?

Well, something great happened during all this….Segment’s platform got even better. They launched two features which have been game-changing for us:

• Warehouses - Warehouses stores all the analytics data for you, making it easy to connect your BI tools and run SQL queries. It has out-of-the-box support for BigQuery, Redshift and Postgres. We set up a Redshift connection and we now have custom dashboards that marry this Redshift data with our underlying Postgres database. (The flexibility of these options is great — we’ve even considered switching our connector from Redshift to Postgres and using the Citus Data extension for Postgres instead. If we ever decide to do that, it’ll take us all of five minutes to switch.)
• Sources - We use a lot of different tools at WayUp for keeping track of and emailing our users, and it used to be impossible, and then prohibitively difficult, to get all of that data in the same place for querying by our Business Intelligence Analyst. Segment’s Sources product connects with many of our third-party tools like Mandrill, Stripe, Salesforce, and others, and dumps that data into our data warehouse. Now, we’re able to run direct queries and custom funnels across over 10 distinct datasets.

Segment sits at the center of our data pipeline, allowing our teams to quickly get access to the data that matters to them. The most important value our business users have drawn from Segment is that all of our data can be seen and tied together. For our account management team, for example, the ability to track client application yield in our Postgres database and tie it to account information in Salesforce has allowed us to benchmark our own performance towards achieving an industry-leading cost per hire for our customers hiring college students and recent graduates. Because of Segment’s numerous email Source options, our marketing team has an advanced ability to track results from their email campaigns and build out conversion funnels that go way further than opens and clicks. (Think questions like: “What is the average spend of a business who clicks this button on the website after having received re-engagement email #2?”) As the support beam of a number of business intelligence tools and platforms used across WayUp, Segment serves as the backbone of our company’s ability to make data-driven decisions.

Looking forward

Segment already contributes to the success of our team, but there’s also something to be said about the company’s determination to keep innovating. Most recently, they launched a feature to deploy analytics on AMP pages, which we use heavily across our digital properties. Segment truly has become a partner for us in every sense of the word and we are excited to continue our journey with them through 2017.