As we're all too commonly aware, the commercial aviation industry loves citing acronyms and using a lot of jargon. Unfortunately, while it can sometimes help simplify the communication of long or complex subject matters, it also leads to confusion and misunderstanding.

One such phrase that fits this bill is "real-time," as in real-time data and how it's transmitted between commercial systems, specifically as it pertains to executing revenue management functions.  

We all know that analyzing data over time is valuable. Still, nowadays, with data transaction volumes growing and the market circumstances changing rapidly, we need to stay on top of trends, make decisions and make them fast.

So, let's look at what real-time data means in the context of revenue management and what different real-time options are currently available and how they impact the life of a revenue manager.

Real-Time and Its Role in Revenue Management

Real-time data refers to data that is presented as it is acquired/created. A couple of decades ago, traditional revenue management systems would process the data on a nightly basis by acquiring transactional data from the Passenger Service System (PSS).

Information such as flights, legs, inventory, schedules, overbooking and booking counter information would drop in and be processed nightly to be available for the users at the start of their business day.

In the early days, the revenue management system would only be able to prepare and process the data based on Data Collection Points (DCPs). In this concept, a manager would only be able to see flight data at a pre-defined set of data points. For example, flight information 70 days before departure, 50 days before departure, etc.

This created a challenge as managers would not be able to measure performance between two data collection points. However, over the past decade, most of the systems are now able to provide insights on flight data at any day before departure, without gaps in the flight life cycle.

"Revenue managers find that real-time data allows them to uncover insights faster and improve their capabilities to respond to rapidly changing behavior."  

Nowadays, whether it is due to the increase of competition, flying becoming more of a commoditized product, or a combination of other reasons, we see booking curves tend to move closer towards departure with managers spending increasingly more time monitoring flights close-in.

Additionally, airlines see whimsical patterns which can change instantly and move a flight from 'normal’ to one which is in high demand.

Revenue managers find that real-time data would allow them to uncover insights faster and improve their capabilities to respond to this changing behavior.  

Put into industry terms, real-time data can be considered flight, traveler, pricing, or other information provided to commercial departments virtually immediately to facilitate better decision-making. And again, we‘re specifically looking at these data sets through the lens of revenue management.  

This is a very generalized and basic definition. Therefore, to truly understand real-time, let’s look at the different options and how they each can help the revenue managers do a better job optimizing flights.  

Passive Real-Time

This type of real-time is the most common method implemented in systems today. Effectively, upon opening a flight in the revenue management system, data is requested through an API fetching the latest information of the flight.

"Real-time data is only displayed when manual requests for flight detail information are made. A manager can only detect different behavior to act upon if a flight is opened."

The right APIs are not always available to accomplish this, and systems sometimes need to process a complete Flight Manifest (which includes PII [Personally identifiable information] data).  

Another downside is that real-time data is only displayed when manual requests for flight detail information are made.

A manager can only detect different behavior to act upon if a flight is opened. This type of monitoring is quite passive hence the label ‘Passive Real-Time'.

The process can be slow and doesn’t scale, meaning this information cannot be retrieved for hundreds of flights at scale.  

Scheduled Real-Time

This type of real-time is more adequately explained as intraday updates. In this scenario, data in the database is updated through the backend of systems on a scheduled basis.

For example, every 15 minutes a request is made to update all flight data and store this data in a database.

The advantage of this execution is that data can be retrieved for multiple flights at once; however, it is not truly real-time and requires a lot of batch jobs and computation power, which generates a lot of redundant data transmitting.

For instance, a request can be made for an update on 7,000 flights while only 700 flights have updated bookings versus the last update.

It is quite complex to organize these types of scheduled requests without compromising efforts to avoid performance hits on the other backend systems.  

"Data can be retrieved for multiple flights at once; however, it is not truly real-time and requires a lot of batch jobs and computation power."

Contrary to passive real-time, scheduled real-time allows systems to store the data into the backend at scale easier and presents the user with a more accurate representation of the flight performance at any point in time.

This is advantageous when a manager tries to monitor multiple flights at once; they can query data based on the most recent information.  

Streaming Real-Time

Unlike scheduled real-time, streaming real-time doesn’t require a batch process. Instead, it is a continuous flow of data that can be processed, stored, analyzed, and acted upon as it’s being generated.

"The challenge is that streaming real-time requires a deeper integration between the PSS and revenue management system. Unfortunately, this integration is not always available or prioritized as there are many compatibility challenges to solve."

If organized correctly, streaming real-time requires less overhead to process and scales well for many applications.

The challenge is that streaming real-time requires a deeper integration between the PSS and revenue management system. Unfortunately, this integration is not always available or prioritized as there are many compatibility challenges to solve.

The scalability of this method allows for the application of real-time for the entire network rather than only a limited number of days, which typically happens with Scheduled Real-Time batch processing.

Any additional change is essentially processed and limits the risk of missing out on the latest trends.  

Streaming Real-Time Integration Service

What is particularly interesting in revenue management is the ability to flag or act upon real-time data. In this integration, automation systems are getting new data and can also link data updates instantly to specific services and automatically perform certain actions.

Yet, in the applications above, we potentially have sparked the imagination of retrieving real-time bookings, changes in inventory or flight activity.

Some revenue management departments have taken advantage of streaming real-time services wherein data is immediately processed through optimization data flows utilizing triggers that are linked to automated workflows and/or machine learning pattern recognition which can either notify the manager to evaluate detected anomalies or re-optimize the flight data.

For example, the system could automatically detect changing consumer behavior on specific flights after the announcement of big sporting events or, after an announcement of lifting travel restrictions between two countries.

"Some revenue management departments have taken advantage of streaming real-time services wherein data is immediately processed through optimization data flows utilizing triggers that are linked to automated workflows and/or machine learning pattern recognition which can either notify the manager to evaluate detected anomalies or re-optimize the flight data."

Soon, revenue management tools will be able to process streaming services including intent data (availability requests by customers). Having this information will allow revenue managers to truly understand buying behaviors, painting a clearer picture of seat demand.

By being able to monitor different trends and behaviors, managers can better understand, action and manage their commercial performance.

This provides the foundation to layer on digital retailing functions to increase conversions, leading towards better overall optimization techniques.