Past Analytics Excellence Awards

Excellence Awards 2017: Bill Lyons - Multiple Award Categories

K-LOVE_logo.pngAuthor: Bill  Lyons  - Principal Data Scientist

Team Members: Trudy Fuher, Alana Welz, Arlyn Baggot

Company: Educational Media Foundation

 

Awards Category: Best ‘Alteryx For Good’ Story 

The initial project has the potential to save this non-profit organization up to $2.2 million per year in streaming costs when recommendations are fully implemented. Other use cases improve internal efficiencies, communication, and productivity.

 

Awards Category: Best Use of Alteryx Server for Analytics Deployment

Alteryx Server automatically processes daily file downloads, weekly file downloads with decompression, decryption and bulk insertion, and monthly zip code DMA assignments. Other use cases support self-service imports, exports and reporting.

 

Awards Category: Best Use of Alteryx for Spatial Analytics

Alteryx spatial tools combined with Alteryx data is driving optimization of regional streams associated with DMAs.

 

Awards Category: Best Value Driven with Alteryx

Optimizing regional streams has resulted in at least $500,000 in savings since July 2016, with recommendations implemented so far. When all recommendations are fully implemented, savings could be $2.2 million per year or more.

 

Awards Category: From Zero to Hero

Even though we purchased our first Designer license in June 2015, as of early March 2016, we had not created a single workflow with Alteryx. We were considering not renewing our license. At that time, we got a new rep, Nick Glassner, who arranged for a couple of WebEx sessions with Alteryx Solutions Engineer Ali Sayeed to get us started on a real project. Within a few weeks, I recognized many more potential applications for Alteryx, and was off and running. I changed from a skeptic to an enthusiastic user. Analysis for this project began in mid-April and was completed in mid-May. We acquired Alteryx Server in June, and had the first phase of the implementation of this project running on a daily schedule by August. Other phases came online in November and in January 2017.

 

At that point, I was still the only person using Alteryx heavily in analysis and production. So, I began some internal workshops showing how to solve real-world problems with Alteryx. We now have 3 more internal users becoming productive with Alteryx, and are looking to hire another. Some of these users are also taking advantage of the “Enablement Series” offered by our new rep, Tim Cunningham.

 

Describe the problem you needed to solve

Initial business problem: Recent regulatory changes caused our national internet radio streaming costs to more than double, from less than $1 million to over $2 million annually. The goal was to find ways to optimize our streams to move usage from the national stream to our underutilized regional streams, and thus reduce our costs.

 

Other use cases, including their business challenges, solutions, and benefits, follow the solutions and benefits of this initial business problem.

 

Describe the working solution

Alteryx played a major role in analysis of the streaming data. Some of the regional streams were underutilized, while others exceeded their cost effective limits, so the first phase was to analyze the accuracy of IP address geolocation software to see what would be causing this. The website systems and the log analytic systems used different IP geolocation software (the websites used IP2Location, and the analytic systems used Maxmind) so we needed to know if one was better than the other, or if neither was adequate. However, these system are isolated from each other by firewalls, making direct comparisons impossible. Alteryx Designer allowed me to connect to three different SQL Server database systems and compare their data with a .csv file from another vendor being evaluated (NetAcuity).

 

This analysis made extensive use of Alteryx spatial matching and Alteryx spatial data, visualizing results with Tableau. It revealed some disturbing facts, including that the geolocation was very inconsistent between the systems. As an example, we found that less than half of the listeners to the New York City stream were even in the NYC DMA (Figure 1).

 

Figure 1Figure 1

 

Additionally, we learned that only a little more than half of the listeners in the NYC DMA were listening to the NYC stream. (Figure 2)

Figure 2Figure 2

 

 The analysis also compared actual registered listener locations to the location reported by the various services. This showed that IP2Location was clearly inferior. (Figure 3)

 

Figure 3Figure 3

But Maxmind returned a significantly higher number of unknown locations, both within the US, and even identifying the country. (Figure 4)

 

Figure 4Figure 4

 

The analysis concluded with 16 recommended changes to systems, software, programming and contracts.


One of those recommendations was to unify both the websites and the analytics on the same and most consistently accurate IP address geolocation provider: NetAcuity. Alteryx supports the updates to the NetAcuity database by downloading the data from NetAcuity, decompressing, decrypting, and bulk inserting it into SQL Server. It does this on a weekly schedule in Alteryx Server, each time moving roughly 40 million rows of data in about an hour.

 

Primary workflow:KLOVE-5.png

 

Supporting macros:KLOVE-6.png

 

 

An Alteryx Server scheduled app then builds Calgary databases of the IP geolocation data.KLOVE-7.png

 

 

Next, another Alteryx Server scheduled app applies that geocoding to the streaming log data.KLOVE-8.png

 

 KLOVE-9.png

 

 

Alteryx spatial data also supports Server scheduled monthly updates to keep zip to DMA to stream assignments up to date.

 

Describe the benefits you have achieved

4 of the 16 recommendations have been implemented to date, saving over $500,000 since last July, and an estimated $700,000 for 2017. More steps are in development, with a goal of saving $2 million per year.

 

Never before did we have a reliable and up-to-date zip code to DMA assignment process. We previously bought zip code to DMA data from Nielsen, but it was incomplete and quickly out-of-date.

 

Other Significant Alteryx Use-Cases

 

1. Transmitter location identification

  • Business Challenge: Property tax filings must be made with the appropriate jurisdiction for the location of the property. With normal property, the street address easily identifies that jurisdiction. However, radio transmitter sites are frequently in very remote locations where there is no street address, and frequently on tops of mountains, within a few feet of jurisdictional boundaries. Historically, property tax accountants manually used transmitter location geographic coordinates to search maps to identify state and county with which to file property tax forms. This very laborious process took a team of 3 or 4 people up to 8 weeks each year, and was fraught with error.
  • Solution: Alteryx Server scheduled app performs spatial match between transmitter geographic coordinates and Alteryx spatial data, precisely and accurately identifying and coding transmitter location state and county. Run time: about 15 seconds per day, automatically. This simple workflow took only a couple of hours to build and deploy.KLOVE-10.png

     

  • Benefit: Savings of up to 8 man-months of manual labor per year. Reduction in errors (this process identified more than 200 instances where the location was either undocumented or in the wrong jurisdiction; 2 were even in the wrong state).

 

2. Log file FTP download

  • Business Challenge: The system downloading new log files from content delivery network (CDN) daily was very fragile, requiring manual checks and restarts every few days.
  • Solution: Alteryx workflow app, scheduled to run daily, downloads list of available files, compares list to list of previously downloaded files, downloads new files, updates list of files downloaded.K-LOVE-11.png

     

     KLOVE-12.png

     

 

 

  • Benefit: Alteryx job has run without error for 8 months. Saves time (about an hour per week) monitoring and maintaining each week, but it is mostly a huge reduction in the "hassle factor." Time to develop was less than a couple months’ worth of manual corrections.

 

3. User import of Excel into SQL Server

  • Business Challenge: Data files from mobile app vendors come each month in Excel files and need to be imported to SQL Server. This import required a DBA to manually import, and was consequently a year behind.
  • Solution: Gallery app allows users to upload files themselves, automatically removes duplicate data, reports duplicates ignored, structure errors, and data imported.KLOVE-13.png

     

     

 

  • Benefit: Self-service of data import relieves workload of DBAs and allows users to have immediate reporting of data in Tableau. This process also revealed that the supplier had duplicate records that overlapped between months. This had created erroneous data of which we had not previously been aware.

 

4. Tealium reporting

  • Business Challenge: Connecting Tableau directly to Redshift was slow.
  • Solution: In-Db tools query Redshift database, filter, aggregate, and download to Tableau Server Data Source Extract. App is scheduled in Alteryx Server. 

     

    KLOVE-14.png
  • Benefit: Faster Tableau reports

 

5. Studio automation logs

 

  • Business Challenge: Log files have been inconsistent and incomplete, with gaps and overlaps, making downstream reports unreliable.
  • Solution: Download tool connects directly to REST API of studio automation software, parses the JSON, and inserts into SQL Server data warehouse. Scheduled in Alteryx Server daily.KLOVE-15.png

     

     

  • Benefit: Reliable data for reporting.

 

6. Record of donor communication

 

  • Business Challenge: Producers call donors to record their stories, logging that call in Google Sheets. Donors call back, talking to communicators in the Listener Services department who have no visibility to the Google Sheets, and there was no record in the donor system. Awkward conversations ensued.
  • Solution: Alteryx Server app scheduled to run every 5 minutes connects to Google Sheet, downloads the call records and insert records into the SQL Server donor system of record.KLOVE-16.png

     

  • Benefit: Listener Services communicators can now intelligently communicate with donors.

 

2 Comments
LeahK
Alteryx Alumni (Retired)
Status changed to: Inspire 2017 Las Vegas Award Entry
 
andrewdatakim
12 - Quasar
12 - Quasar

Great use case Bill Lyons! We are also heavy users of Nielson information and we have worked out different ways to scrap for zip codes and DMAs to keep our databases up to date.