Customer Story

Center for Open Science Provides Embedded Analytics to 250,000+ Researchers with Keen




Charlottesville, Virginia

"Keen was the best solution for us because it offers full customization and the ability to ship arbitrary metadata and to structure tracking however we want."

Fitz Elliott
Senior Software Engineer for the Infrastructure Team

With Keen, Center for Open Science illustrates the value of their platform and prioritizes key development decisions.

Center for Open Science (COS) is on a mission to increase the openness, integrity, and reproducibility of research. The challenge? In the present scholarly culture, these values are not standard practice across disciplines.

COS enlisted the help of Keen to propel their mission forward. Keen powers the embedded analytics in their Open Science Framework (OSF), a free and open-source project management tool that supports researchers throughout their entire project lifecycle.

Embedded Analytics provided in the OSF powered by Keen
Embedded Analytics provided in the OSF powered by Keen

The OSF gives researchers valuable insight by offering visibility into their downloads, popular pages, and traffic referrals. Providing evidence for the effectiveness of open research is critical to establishing trust with the greater research community and shifting the culture to one that prioritizes rigor and transparency.

We asked Fitz Elliott, Senior Software Engineer for the Infrastructure Team, to tell us a little more about how Keen was able to help:

How did you initially approach offering embedded analytics?

Originally we used a self-hosted project called Piwik (now Matomo), but we were limited to tracking the most common page-visitation metrics. With our limited resources, we couldn’t devote the time necessary to expand these metrics or to improve our hosted infrastructure.

Out-of-the-box analytics services didn’t offer the degree of custom tracking that we needed, and customizing Piwik required time and effort beyond our resources. The need for greater flexibility and support through embedded analytics quickly became evident.

Why did you choose to switch to Keen?

We were looking to get away from self-hosted solutions and take more advantage of the SaaS offerings available in 2016. We wanted to stop wasting bandwidth on maintaining our own analytics software and trust the process to a third party. We also wanted to start tracking the metrics we wanted instead of what Piwik’s developers thought we wanted. And we wanted to do it without having to allocate a developer to becoming a full-time analytics specialist.

Keen was the best solution for us because it offers full customization and the ability to ship arbitrary metadata and to structure tracking however we wanted. It’s Data Explorer enables us to construct robust private queries on the metadata that we ship. Keen also offers a strong charting library that allows us to implement strong data visualization options for OSF users.

What other ways do you use Keen beyond embedded analytics?

We strive to provide useful web-renderings of many different file types. We used Keen to power our internal analytics to figure out which currently-unsupported file types are the most requested so that we can devote internal effort to supporting these.

Additionally, we have used Keen for internal growth tracking to understand user behaviors, the institutions we currently support, and their user engagement.

How was the process of getting started with Keen?

Keen itself was very straightforward to implement. Our main hurdle was in figuring out how to structure public vs. private analytic payloads and in learning what was useful to track.

Two elements that led to straightforward application on the OSF infrastructure: 1) the ability to ship arbitrary data and send anything we wanted without predefined structures, 2) filters that could extract common data from what we were sending. We didn’t need to build our own code, but could instead rely on code that Keen provided.

How has Keen helped your team?

Keen’s Dashboards and Data Explorer have given our Product teams access to a powerful, rich interface to make queries without having to become experts in HTTP / JSON / python, etc. Generated metrics (e.g. adoption behaviors) help surface insights that have fueled our future work and infrastructure improvements. And it has supported our efforts to provide researchers with insight into the impact of their work.

We also support integrations with a number of different external storage services. With Keen, we have been able to identify which integrations are most popular among our audience and triage our support efforts accordingly.

Keen has helped us save time, track progress, and measure success by enabling the automation of monthly key performance indicators. It provides metrics that are not stored or easily obtainable from OSF’s database (e.g. on a given date, what percentage of users have signed in within the last 7 days.)

When would you recommend Keen?

We recommend Keen’s services for organizations or groups searching for a fully customizable analytics experience, and can particularly speak to the value of its robust internal analytics to help order the prioritization of development work, and its support for arbitrary analytics needs.