People Power – Drilling through Data Layers for Faster Insights

Scalable self-service data exploration and visualisation is currently a hot topic. In two articles we  explore how new and easier methods of data discovery is helping organisations on the tough journey to becoming more data-driven.  This article Part 2 describes how the Trio platform can be used for a drilling deeper into search and booking data, to be able to report on the data set that is meaningful and relevant to a B2B distributor.  Part 1 gives an overview of some of the exploration techniques.

Train of Thought Analysis

As an end user, if we want to use data to support our business endeavours then we need to be able to display data in a very visual way that we can interact with.  The objective is to perform a visually directed exploration of the data to identify clear information that we can action.  That might be solving a problem, identifying an opportunity or just [dis]proving a theory.

There is a significant amount of hype around “big data” but the reality is that whilst an AI system might process huge amounts, humans naturally prefer to use “little data” for decision making.  That means crunched big data that has been processed into an aggregated dataset that fits on a screen or page which one can realistically review.  When we explore data we also want to able to keep focused on a small subset of data which is maintained as we follows the data trail to a goal.  This is what we mean by train of thought analysis.

Unravelling the Aggregated Data

The aggregation process can be a very simple arithmetic processes such as summation, averages or percentages. These are usually chosen so that the key patterns of interest are preserved despite that the fact that the data might have been aggregated from many millions of rows of data into a maybe as less than a hundred rows.

Continuing our example above, the search and booking data may be based on tracking hundreds of millions of transactions which have been added up a single value = total number of searches.  When we add the Hotel Property dimension we are starting to unravel the aggregated data into something which is becoming more actionable.  By actionable, we mean the data will likely indicate a particular course of action(s) e.g. we might see information about a specific hotel’s search /book values that stands out and needs attention.  The focus of the investigation is now shifting to that specific Hotel Property and finding out more about the searches to just that property.

If we just add another dimension such as check-in date then we will get a report which has many more rows – one for every possible combination of Hotel Property and Check-in Date over probably the next 365 days.  That could be quite a few rows and we would start to lose focus.  We need a visualisation technology to automatically reduce the scope of our view to just the hotel property and report on all its searches and booking by check-in date.   This process, often referred to as “slicing and dicing”, relies on data techniques are called drilling and filtering.

• Data filtering is a using value ranges to limit the size of the dataset being considered.  In example we would want to have a filter which limits the data being considered to be just the hotel property we are interested in.
• Data Drilling is adding an extra dimension to the current dataset to get more information – a more detailed breakdown e.g. adding Check-In date to the Property report.

In our example, if we apply both of these two approaches in a single combined action and re run the report, the new report will show searches and bookings sub totalled by Check in dates only for our chosen hotel property.

This mirrors exactly what the Triometric reporting system offers to enable the user to perform rapid interactive reporting using data driven drilling to get to the answers.  Drilling by clicking on a chart bar or row in a report updates the data filtering and permits new dimensions to be selected as a combined activity before a new more detailed report is automatically drawn.

The starting report is just a summary of Search volumes by Property for just a one hour period:

When the user clicks on the left most block for Hotel THB Molinos Class, the user can add the dimension of Check-In Date for just that property form the dialog box as follows:

When the user then completes the drill process by clicking on the blue Drill button the report redraws with just the data for the Hotel THB Molinos Class (this gets added to the report filter) and the Search volumes for Hotel THB Molinos Class are now broken out by Check-in Date:

All of this occurs within the same screen/report area so the user maintains their focus and can successfully use highly desirable train of thought analysis concept to traverse the data.

This self-service concept does rely on the data, or specifically insights, being easily understood by the end user. This is where some business knowledge can assist while any interactive guidance that is on offer needs to be intuitive.

Keeping it Simple

The ability to sift through large amounts of shopping and booking data and draw conclusions is still challenging, but the evolution of analytic tools is making it easier. Today’s online travel executives need analytic platforms that are:

• Easy to use, without needing deeper analytic skills
• Agile and flexible, to run reports and get easy answers to ad hoc questions
• Swift at giving the right insights via user-friendly charts and dashboards.
• Able to handle high volumes of data at speed

These design principles are at the core of the Trio platform, as it handles unobtrusive data capture, cleansing, aggregation and preparation for reporting. Commonly used business reports can be pre-configured for easier reporting and available at the push of a button.

So, where is your company on its data journey? Do you have forward-thinking data strategy? Are your business analysts empowered to explore unknown data and questions to discover insights that create faster actions and better outcomes, especially from your API traffic?  Analytics is getting easier, but more and faster insights is not enough. When opportunities are identified, action must be taken. Without action, there will be no value.

If you would like to explore how Trio Analytics can equip your distribution team with relevant and actionable insights, get in touch with Tamas Krajecz

Part 1 gives an overview of some of the exploration techniques,

Trio is a scalable XML/JSON analytics platform (on-premise and cloud)  that can be readily deployed to analyse and apply API search and booking insights for a travel organisation’s distribution operation.