Dimension Modeling for the Hotel Industry

About the Business:

To illustrate the concept of dimension modeling, I would like to take the example of the Hospitality industry. And to take a more tangible example, I would like to focus on Hilton Worldwide Holdings (HLT from now on). HLT encompasses a large portfolio of hotels and resorts worldwide. It was founded by Conrad Hilton in 1919 in Cisco, Texas, and later bought over by the Blackstone Group for about $26 billion in 2007. It is the sixth largest hotel chain globally by the number of Hotels.

Performance Metrics:

Now if I was the CEO of Hilton Worldwide (Christopher J. Nassetta), and I wanted to see some performance metrics for HLT I would probably want to see some of the points below:

* Which line of hotels is the most profitable? - To find out which hotel chain is raking in the most revenue.

* Which line of hotels is the most popular? - To find out which hotel chain is attracting the most customers.

* Which location-segment is the most popular? - To find out which locations do customers frequently visit.

* Which market-price segment is the most popular? - To find out segment is the most popular amongst Luxury, Upscale, Midscale, Economy
  
* What is the occupancy of a specific hotel for a particular time period? (Paid Rooms Occupied/Rooms Available) - To find out how occupied is a hotel during a time period.

* What is the Revenue per Available Room for a specific hotel chain? (Total Room Revenue / Total Rooms Available) - To find out the revenue earned per room in a given period.

* What is the Average Daily Rate for a particular hotel? (Rooms Revenue / Paid Rooms Occupied) - To find out the average rental income per paid occupied room in a given time period.

There might be many more metrics but let's focus on these for now. Now to get all the above information we will have to capture a lot of data. Particularly, we will need data of bookings done at hotels, the hierarchy of hotels (which chain do they belong to), the revenue paid by customers for the bookings (and the discount), the number of rooms categorized by class in each hotel, etc. The definition of "revenue" might be different for each business segment but for this article, let's assume the revenue is the amount a customer pays for a hotel room.

Dimensional Modeling:

To accommodate the lowest level of facts, we will be capturing the booking data i.e. transaction data of each customer. Measures like Revenue per available room, Average Daily rate can be derived by the BI tool based on the data captured. Revenue over a time period can be done by adding the revenue obtained from each room of a hotel over a period categorized by date (since all dimension models should have a date dimension). We can also have a periodic snapshot table over a month or over a quarter identifying the revenue and the average occupancy over a period.

The dimensions for such a model would include the following:

• Hotel Chain Dimension
• Hotel Dimension
• Customer Dimension
• Date Dimension
• Room Type Dimension
• Deal Dimension

A high level dimensional model for the hotel industry would look something like this:

Comparison of Business Intelligence and Analysis Tools

Organizations today suffer from the DRIP syndrome. DRIP stands for 
Data Rich Information Poor. There is too much of data coming in from various different sources but there is no central system to extract some useful information from the data. This is where Business Intelligence comes in, to make sense of all this data.

There are quite a few commercial vendors offering Business Intelligence and Analysis products. This article deals with comparing a few vendors based on a weighted criterion. First, let's talk about the criteria.

For the weighted decision matrix, I will be using the following criteria:

Data Sourcing: The first step of analysis using a BI tool is getting the data. Data can be from text files, flat files, CSVs, databases, servers, etc. The more sources a tool supports, the more are the opportunities for getting data. Also, it should be easy for an end-user to connect the data sources to the BI tool.

Usability: No matter how good a tool is at its core functions if it's not easy to use it will most probably be rejected by the end user. A tool should be user-friendly, the UI should not be too complex and the end user should not be lost while trying to figure how to go about using the tool.

Filtering and Visualization: The end user should be able to drill down to the data they are seeking. This can be done using filters, selecting criteria, the intelligence of the software (maybe?), etc. Another function associated with selecting data is visualization. A picture is worth a thousand words and certainly more than 10,000 rows on a spreadsheet! If a tool allows the user to see data in terms of graphs or charts and then allow drilling down, that tool is intuitive.

Reporting: The tool should allow the creation of reports. A BI tool would be marginally useful if it cannot be used to share information. Reports should be customizable based on user selected criteria and the tool should allow exporting reports in different formats. Also, integration with most popular extensions e.g. Words, Excel, PDF, etc. would be a plus.

Cost: Cost is one of the first factors which is looked at while selecting a business tool, especially by the finance division. The cost, of course, is relative. A big brand license would be too expensive to a small scale customer and vice versa. But a lot of cost-to-features can be done while determining the selection of a BI tool.

Added Features - These features are not the core features of a BI tool but have gone from could have to should have or even must-have. Following are the add-on features:

Admin & Management: BI tools are not stand-alone tools anymore. They are full-fledged applications. Thus, these applications should allow administration of users, access rights, security, etc.

Deployment: BI tools should make it easy for users to implement them in their own system. It should be easy for end users to integrate the tool with their current existing systems.

Mobility: A cell phone today is more powerful than the computers on the Apollo 11's mission. And with users being more mobile than ever, BI tools are expected to seamlessly integrate users on mobile devices and offer the same or almost the same experience as a desktop user.

This article will be comparing the following commercial vendors:

• IBM Cognos

• Microstrategy

• Tableau

• SAS

• Microsoft

A weighted scoring matrix criterion of features of all the popular BI tools yields the following result:

From the above matrix, we can see that Tableau scores high on Usability, while SAS scores high on Filtering & Visualisation. Microstrategy scores high on Mobility and Deployment and Reporting. Microsoft scores on Administration while SAS and Tableau tie on Data Sourcing. Overall, Microstrategy ranks ahead because of its higher ranking of features.

Sources:

https://www.gartner.com/doc/2989518/magic-quadrant-business-intelligence-analytics

http://www.pcmag.com/article2/0,2817,2491954,00.asp

http://www.sas.com/content/dam/SAS/en_us/doc/analystreport/ovum-decision-matrix-bi-105875.pdf

http://searchbusinessanalytics.techtarget.com/feature/How-to-evaluate-and-select-the-right-BI-analytics-tool