Past Analytics Excellence Awards

Author: Alexandra Wiegel, Tax Business Intelligence Analyst 
Company: Comcast Corp

Awards Category: Best Business ROI

 

A Corporate Tax Department is not typically associated with a Business Intelligence team sleekly manipulating and mining large data sources for insights.  Alteryx has allowed our Tax Business Intelligence team to provide incredibly useful insight to several branches of our larger Tax Department. Today, almost all of our data is in Excel or csv format and so data organization, manipulation and analysis have previously been accomplished within the confines of Excel, with the occasional Tableau for visualization. Alteryx has given us the ability to analyze, organize, and manipulate very large amounts of data from multiple sources.  Alteryx is exactly what we need to solve our colleague’s problems.

Describe the problem you needed to solve

Several weeks ago we were approached about using Alteryx to do a discovery project that would hopefully provide our colleagues further insight into the application of tax codes to customer bills. Currently, our Sales Tax Team uses two different methods to apply taxes to two of our main products respectively. The first method is to apply Tax Codes to customer bill records and then run those codes through software that generates and applies taxes to each record. The second method is more home-grown and appears to be leading to less consistent taxability on this side of our business.

 

Given that we sell services across the entire country, we wanted to explore standardization across all our markets. So, our Sales Tax team tasked us with creating a workflow that would compare the two different methods and develop a plan towards the goal of standardization and the effect it would have on every customer’s bills.

 

Describe the working solution

Our original source file was a customer level report where the records were each item (products, fees, taxes, etc.) on a customer’s bill for every customer in a given location. As it goes with data projects, our first task was to cleanse, organize, and append the data to make it uniform.

 

 

The next step was to add in the data from several data sources that we would ultimately need in order to show the different buckets of customers according to the monetary changes of their bills. Since these sources were all formatted differently and there was often no unique identifier we could use to join new data sources to our original report. Hence, we had to create a method to ensure we did not create duplicate records when using the join function. We ended up using this process multiple times (pictured below)

 

 

And so, the workflow followed. We added tax descriptions, new codes, and other information. We added calculated fields to determine the amount of tax that should be owed by each customer today, based on our current coding methods.

 

 

After we had layered in all the extra data that we would need to create our buckets, we distinguished between the two lines of business and add in the logic to determine which codes, at present, are taxable.

 

 

For the side of our business whose taxability is determine by software, you will notice that the logic is relatively simple. We added in our tax codes using the same joining method as we did above and then used a single join to a table that lists the taxable codes.

 

 

For the side of our business whose taxability is determine by using our home-grown method, you can see below that the logic is more complicated. Currently, the tax codes for this line of business are listed in such a way that requires us to parse a field and stack the resulting records in order to isolate individual codes. Once we have done this we can then apply the taxability portion. We then have to use this as a lookup for the actual record in order to determine if a record contains within the code column a tax code that has been marked as taxable. Or in other words, to apply our home-grown taxability logic is complicated, time consuming, and leaves much room for error.

 

 

Once we stacked all this data back together we joined it with the new tax code table. This will give us the new codes so that the software can be used for both lines of business. Once we know these new codes, we can simulate the process of the software and determine which of the new codes will be taxable.

 

 

Knowing whether or not codes are taxable helps us hypothesize about how problematic a geographic location may end up being for our team, but it does not tell us the dollar amount of taxes that will be changing. To know this we must output files that will be run through the real software.

 

Hence, once we have completed the above data manipulation, cleansing, and organization, we extract the data that we want to have run through the software and reformat the records to match the necessary format for the software recognition.

 

 

We created the above two macros to reformat the columns in order to simply this extensive workflow. Pictured below is the top macro. The difference between the two resides in the first select tool where we have specified different fields to be output.

 

 

After the reformatting, we output the files and send them to the software team.

 

 

When the data is returned to us, we will be able to determine the current amount of tax that is being charged to each customer as well the amount that will be charged once the codes are remapped. The difference between these two will then become our buckets of customers and our Vice President can begin to understand how the code changes will affect our customer’s bills.

 

Describe the benefits you have achieved

Although this project took several weeks to build in Alteryx, it was well worth the time invested as we will be able to utilize it for any other locations. We have gained incredible efficiency in acquiring insight on this standardization project using Alteryx. Another benefit we have seen in Alteryx is the flexibility to make minor changes to our workflow which has helped us easily customize for different locations. All of the various Alteryx tools have made it possible for the Tax Business Intelligence team to assist the Tax Department in accomplishing large data discovery projects such as this.

 

Further, we have begun creating an Alteryx app that can be run by anyone in our Tax Department. This frees up the Tax Business Intelligence team to work on other important projects that are high priority.

A common benefit theme amongst Alteryx users is that Alteryx workflows save companies large amounts of time in data manipulation and organization. Moreover, Alteryx has made it possible (where it is impossible in Excel) to handle large and complicated amounts of data and in a very user friendly environment. Alteryx will continue to be a very valuable tool which the Tax Business Intelligence team will use to help transform the Tax department into a more efficient, more powerful, and more unified organization in the coming years.

 

How much time has your organization saved by using Alteryx workflows?

We could never have done this data discovery project without using Alteryx.  It was impossible to create any process within Excel given the quantity and complexity of the data.

 

In other projects, we are able to replicate Excel reconciliation processes that are run annually, quarterly, and monthly in Alteryx.  The Alteryx workflows have saved our Tax Department weeks of manual Excel pivot table work.  Time savings on individual projects can range from a few hours to several weeks.

 

What has this time savings allowed you to do?

The time savings has been invaluable.  The Tax Department staff are now able to free themselves of the repetitive tasks in Excel, obtain more accurate results and spend time doing analysis and understanding the results of the data.  The “smarter” time spent to do analyses will help transform the Tax Department with greater opportunities to further add value to the company.

Author: Brett Herman ( @brett_hermann ) , Project Manager, Data Visualization 

Company: Cineplex

 

Cineplex Inc. (“Cineplex”) is one of Canada’s leading entertainment companies and operates one of the most modern and fully digitized motion picture theatre circuits in the world. A top-tier Canadian brand, Cineplex operates numerous businesses including theatrical exhibition, food service, amusement gaming, alternative programming (Cineplex Events), Cineplex Media, Cineplex Digital Media, and the online sale of home entertainment content through CineplexStore.com and on apps embedded in various electronic devices. Cineplex is also a joint venture partner in SCENE – Canada’s largest entertainment loyalty program. 

 

Awards Category: Most Time Saved

 

Describe the problem you needed to solve 

Incremental/Uplift Modelling is a popular method of evaluating the success of business initiatives at Cineplex. Its effectiveness at measuring the change in consumer behavior over time creates a high demand to produce this kind of analysis for various departments in the organization. Due to the large amount of requests we receive, the ‘Incremental Lift Model’ was developed to take in user-defined inputs, and output the results within a short period of time.

 

Describe the working solution

Our solution works through a four step process. The first step is for the client to populate the ‘study input form’ in order to define their study parameters and the type of study they want to run.

 

Visual 1: Study Input Form

 

The second step is to update/materialize our loyalty data that’s inputted into the model (yxdb format). We do this so that the model doesn’t put stress on our SQL Server databases, and to increase the model’s performance.

 

Visual 2: Update/Materialize Alteryx Input Data

 

The third step is the core of the incremental lift modelling. A macro processes one study at a time by pointing to the user defined inputs made in the first step.

 

Visual 3: Study Numbers are selected and passed through the incremental lift macro, and saves the output to SQL.

 

The data will then be passed through one of several macros depending on the study type, and filtered down based on the inputs defined by the user in the study input form. All data sources are joined together and lift calculations are made, which are then outputted into a master SQL Table ready to be visualized.

 

Visual 4: Incremental Lift Modelling based on study type selected.

 

The results are visualized using a Tableau Dashboard in order to share and communicate the results of the study back to the business clients.

 

Visual 5: Tableau Dashboard to explain to the business how the incremental lift model makes its calculations.

 

 

 

Describe the benefits you have achieved

The overarching goal of this project was twofold; to minimize the amount of work required to process business requests while maximizing the output generated, and to develop a means of delivering the results in a consistent manner. Both of these goals contribute greatly to our ROI by virtually eliminating all time spent executing requests that come in, and by minimizing time spent meeting with business users to explain how the incremental lift model works and how to interpret the results.