Splurgy, a San Francisco-based start-up, is a social marketing platform where brands reward loyal customers with incentives in exchange for social engagements — on your website (rather than your Facebook page).
Tables of Contents
- Treasure Data for Both Ad-hoc Analysis and Reporting
- How We Put Our Data on Treasure Data: Pixel Tracking
- Treasure Data: The Best of Both Worlds
Treasure Data for Both Ad-hoc Analysis and Reporting
Splurgy uses Treasure Data for both ad-hoc analysis and reporting.
Ad-hoc analysis
As a data-driven, fast-moving start-up, Splurgy is constantly mining data to improve our service. Which social media channel is growing fastest? Is any particular brand getting more value out of their platform than others? Or, simply, where does our web traffic come from?
These ad hoc analyses are very important but also hard to predict: you never know what question you want to ask three months from now. This is one of the reasons we chose Treasure Data: unlike other “web analytics” services that only let you ask a restricted set of questions, Splurgy can write SQL queries to get the answers to whatever questions we have.
The “just add one line of JavaScript” school of analytics has its place. But it is not for everyone. I knew that Splurgy needed something more powerful and versatile than that, and I am happy that Treasure Data has that flexibility (without the development and management cost of a homegrown solution!)
Reporting
We provide our customers with stats and metrics via administrative dashboards. For example, our customers would be interested in how their engagement rates are trending. We use Treasure Data to compute many of these stats and metrics.
Again, the key here is Treasure Data’s flexibility. As our platform evolves, the desired set of stats and metrics will inevitably change. Because we have transparent access to our data with Treasure Data, we know that we will be able to tweak and add queries to meet any future requirements.
How We Put Our Data on Treasure Data: Pixel Tracking
We use pixel tracking extensively to collect data. Pixel tracking, aka web bug, is a well-known method of collecting user event data. For example, many email newsletter services use pixel tracking to calculate a campaign’s open rate.
At Splurgy, we wrote a little in-house library to generate event-tracking pixels on the client side. On the server side, we run a standalone service that listens to pixel-tracking HTTP requests. That service then parses URL parameters, converts them into JSON events, and posts them to Treasure Data. A pretty simple setup, right?
Since we didn’t have to build out a data store behind the scenes, we only had to write a bit of code initially to get this system going. And going forward, there is little to no overhead for our engineers. They just have to add one line of code whenever they want to log user events. The best part is that we don’t have to manage our data. We just need to send user events to Treasure Data which takes care of the rest.
We are excited to hear that Treasure Data is thinking of adding this pixel-tracking feature into future versions of td-agent. That means that it will be even easier to set up your pixel-tracking subsystem: just install td-agent!
Treasure Data: The Best of Both Worlds
As a start-up veteran, I can tell you this: time is your most important resource, but whenever you take a shortcut, carefully consider what you are giving up. I think we made the right decision at Splurgy because Treasure Data offered the right balance of flexibility (“we can query anything”) and abstraction (“we won’t get paged at 2AM because our data analytics system borked”). I am very pleased with their service and highly recommend it to any company that needs a simple and flexible data analytics service.
