Background
Networks don’t just publish content on their own domain anymore.
A single show airs across the network’s own platform, a roster of MVPD partners like Comcast and Verizon, and OTT providers like Roku.
Every one of those partners reports performance differently, if they report it in a structured way at all.
The Challenge
The network’s content distribution team needed answers a spreadsheet couldn’t give them. Answers like:
- how each show performed on each partner’s domain
- how that performance measured against the terms in each partner contract
- how it compared to performance on the network’s own platform
They also wanted to see the pattern over time, such as quarter over quarter, not just a single snapshot.
Getting there meant pulling viewership data out of a dozen different partner systems, most of which had no shared format and no shared vocabulary. Some partners offered an API.
Many didn’t, which meant building a custom collection method for each one. The data that did come in was inconsistent — show titles spelled differently across partners, manual entry errors, and no authoritative ID for a given show or episode. On top of that, the network wanted one dashboard that worked for two very different audiences: frontline managers who needed to drill into individual shows, and senior leadership who needed the same data rolled up.
THE SOLUTION
Pulling data from every partner.
DAS42 wrote custom Python scripts to pull from partner APIs and use web-scraping scripts to collect the necessary data when a custom API wasn’t available.
- Centralizing at scale. All of that data landed in a Hadoop Hive environment inside the network’s private cloud, built to normalize and analyze data at volume while meeting the network’s security requirements.
- Building a standard nomenclature. Content distribution through third parties wasn’t going away, so DAS42 built a standard naming convention for shows and episodes that held regardless of whether the data came from the network or partner.
- Cleaning and normalizing everything. With the nomenclature in place, DAS42 normalized the raw data from every partner against it: collapsing naming variants, correcting mis-categorized records, and fixing formatting inconsistencies across thousands of datasets. The cleaned data moved into the network’s enterprise database so other departments could use it too.
- Building the dashboards. DAS42 built persona-based interactive dashboards in Tableau Server, giving more than 400 users access to the reports and the ability to download their own copies directly through a browser. Deep filtering lets content-distribution and research teams answer their own questions: how a show is trending period over period, whether viewership is up or down quarter over quarter, and what the net impact on a show’s audience looks like. Senior managers saw the same data rolled up to a higher level.
THE RESULT
The network saw its first analysis within six months of engaging DAS42.
This was a fast turnaround given the typical pace of that enterprise and the scale of data cleanup involved.
For the first time, the content distribution team had a complete picture of their digital and VOD business, including which partners were top performers by show, and whether contractual obligations were actually being met.
More than 400 users now access the dashboards directly through their browser. The research team uses the same data to track the ebbs and flows of viewership and report findings to sales and editorial teams. The client was impressed enough with the initial build to retain DAS42 for continued work.
We work as an extension of our clients’ teams, solving the data problems that stand between them and a business decision.
Real clients, real results.
