Code Across Portal Engagement: A Retrospective

During the 2016 Code Across, dozens of groups around the world used open data to create solutions to challenges in their local communities, and the event is generating an increasing amount of data portal activity.

In this post, I will dig deeper into the numbers and explore the past four years of data portal activity during Code Across events in Austin, Kansas City, and San Francisco. These cities each held Code Across events during the past four years and have publicly available site metrics.

The graphs show data portal page views and data portal downloads during each hour of a Code Across weekend (71 hours total), starting on Friday at midnight and ending on Monday at midnight (EST).

I collected the data using the Socrata Site Metrics API and created the graphs using the ggplot2 package in RStudio. You can collect this data for your portal at any time using our data portal analytics tool, which is currently in the early development stages.

Code Across Page Views

The “Page Views” metric “tracks every time any page on your site has been loaded in a web browser.” During Code Across events, people are scouring open data portals for interesting datasets to use in their projects. We see from the data that page views in each city greatly increased in 2016.

Code Across Total Rows Loaded

The “Rows Loaded Total” is an aggregate of the six methods people can use to receive data from a Socrata data portal (api, download, print, rss, website, and widget.) We see from the data that this metric has also increased year after year in Kansas City and San Francisco, but not Austin.

It appears that 2015 was an outlier year for Austin because of  two very large requests for data. The similarity between the two spikes suggest that a very large dataset was downloaded twice in the same day.


When looking at Code Across weekends in isolation, we are seeing a huge increase in usage in both the number of portal page views and rows of data downloaded.

These trends provide some evidence that data portals and civic hacking events are successful ways to engage the community around open data.

But looking at these metrics alone does not tell the whole story about how people are engaging with a data portal. The numbers don’t tell us what kind of data people downloaded or who was downloading it – this is just a limitation inherent to the data. It is also useful to compare these figures to other civic events, seasonal trends, and changes in data portal content – these are topics I have planned for future blog posts.

To gain better insight into the bigger picture, we would love to hear from those who participated in these Code Across events. Do the graphs reflect what you saw at your event? Do the large spikes in downloads correspond with particular projects? Do spikes in page views correspond with higher in-person attendance? Let us know.

We hope this information is useful in understanding Code Across from a data portal perspective and that it gives organizers useful information when planning future civic events.

Austin Code Across Page Views

Austin Code Across Total Rows Loaded

Austin, TX

Year Average Page Views / Hour

And % Δ

Average Rows Loaded / Hour

And % Δ

2013 28 93,136
2014 82 (66%) 52,545 (-77%)
2015 125 (34%) 814,727 (94%)
2016 801 (84%) 429,577 (-90%)


KCMO Code Across Page Views

KCMO Code Across Total Rows Loaded

Kansas City, MO

Year Average Page Views / Hour

And % Δ

Average Rows Loaded / Hour

And % Δ

2013 15 860
2014 22 (32%) 7204 (88%)
2015 116 (81%) 242,422 (97%)
2016 2124 (95%) 344,399 (30%)


SF Code Across Page Views

SF Code Across Total Rows Loaded

San Francisco, CA

Year Average Page Views / Hour

And % Δ

Average Rows Loaded / Hour

And % Δ

2013 93 81,917
2014 323 (71%) 315,821 (74%)
2015 656 (51%) 3,455,629 (91%)
2016 2,798 (77%) 2,831,147 (-22%)


Appendix: Code Across Dates

2016 – March 4-6

2015 – February 20-22

2014 – February 21-23

2013 – Feb 22-24