Tell Apart - Customer Data-Driven Marketing

In my final interview at TellApart, I asked the co-founders if they’d describe the company as data driven. They proceeded to show me an intricate set of dashboards that monitored everything about the business and our fleet of servers in real time. The dashboards ingested Ganglia and Amazon CloudWatch data, but also homegrown metrics for things like:

• The number of third-party ad network bids considered in the last minute.
• The distribution of product prices across retail sites.
• The number of clicks that have occurred since March.

The best teams I’ve worked with see data visibility as an uncompromising necessity from Day 1, and I immediately jumped at the chance to join up. Now, nearly a year later, I found something missing from our internal measurements – web page latency. Fortunately, we found an ally to help us secure this missing piece of data.

First, some background. At TellApart, our two biggest concerns are tag latency and global thermonuclear war (to paraphrase Austin Powers). Our servers handle hundreds of millions of requests per day, many of which originate from end users browsing the web. But while we can easily measure the latency of these requests, we needed a way to measure the total, end-to-end impact of TellApart tags, including the third-party tags we’re responsible for.

There are many different web performance monitoring services out there, and we tried a few of them, but only Catchpoint lived up to our high expectations. Catchpoint has a lightweight, modern, self-explanatory user interface. Many stats are quickly broken down and summarized on waterfall reports that can jumpstart performance investigations. You can also see which page elements triggered third-party requests, giving you a view into who is responsible for different elements on the page.

A few of the services we tried prior to Catchpoint made it difficult to export data into a usable format. Dashboards can be attractive, but not if you want to reformat graphs or provide your own data analysis. Catchpoint offers well-formatted CSV exports that open cleanly in Excel and other statistical analysis suites. Test maintenance is also minimized with a strong inheritance system for setting defaults and an instant test feature that can verify parameters prior to starting long running tests.

Fast testing and well-formatted reports help us maintain a proactive approach with our clients. We catch anomalies faster, minimizing our impact to the user experience and optimizing campaign performance.

Ned Rockson is a software engineer at TellApart (with Ben Snider, technical solutions engineer.)