Ai Editorial: XML data processing – how airlines can get it right?

First Published on 26th June, 2018

Ai Editorial: Airlines are investing in infrastructure to make data usable in real-time and at scale for search and analytics use cases. But there are significant challenges in processing and surfacing XML data which needs to be in place to pave the way for actionable business intelligence. Ai’s Ritesh Gupta spoke to Triometric’s Jonathan Boffey about the same. 


Digital organizations count on data as a core asset in their quest to serve their passengers in an earnest manner in today’s connected era. It is imperative to not only act on 1st party data, but also ones emanating from non-company owned sources, including 3rd party distribution partners, to deliver a consistent offering.

Airlines, as an industry, too have been gearing up for the same.   

For airlines, the journey of being in control of their own data and making the most of it hasn’t only been about breaking organizational silos, embracing customer experience management platforms and adaptive IT architecture. Carriers have also acknowledged the need to modernize airline-specific systems and legacy processes. One tangible result that needs to come out from such massive transformation is the understanding of travellers’ buying preferences, what is being searched for, the most popular destinations etc. The available intelligence has to pave the way for improved airline network planning and revenue management, airport route development and travel-related marketing.

Airlines’ data processing capability 

A testament of any successful digital or technology transformation is the intelligent information that comes from transformed data.

As airlines sharpen their ability to store and analyze large volumes of disparate data, how are they looking at understanding the search pattern even from their indirect channels especially in the wake of IATA’s NDC standard that has been around for nearly six years now?

It is vital to assess how airlines are counting on various data sources and focusing on real-time processing. 

“Airlines need to sharpen their respective XML data processing capabilities,” asserts Jonathan Boffey, Triometric‘s SVP for Business Development.

According to Boffey, the capability should allow multiple data sources such as Kafka log queues or packet level network traffic from an NDC system to be captured and processed for key IT and business data content that can then be fed into corporate BI environment such as QlikView or Open Source big data technologies such as ELK.

As for how Triometric is playing their part in preparing airlines for the same, the company promises to deliver via their new offering, Trio Data Engine.

The Trio Engine initiative brings together a number of key technologies that will greatly improve the chances of airlines trying to deliver actionable information from data projects based around technologies such as Kibana/ELK, Kafka. Pentaho and QlikView amongst others.  

“For airlines, it (Trio Engine) can process in real-time the potentially huge volumes of valuable search data from NDC APIs.”

Significance of search data in a digital economy

Airlines need to collect and process the recent search data across all their channels. They also need to have appropriate skills to analyse this data by market segment, formulate offers, set pricing and then adjust booking engine rules to deliver this at point of search. 

“This has to be a continuously improving process of set the rules, analyse the outcome and adjust. In a NDC world this becomes dynamic,” says Boffey.

Boffey also highlighted that by counting on search data, an airline can differentiate their offering for several reasons.

“Firstly for ‘right price’ reasons, everyone is using recent booking history in their RM system to predict the probability of getting another booking.  That has nothing to do with how many people are actively searching and potentially not booking.  Secondly for ‘right product’ reasons, you need to understand who is searching or travelling, what their travel intent is, where they are in their planning/booking process and what might be of interest to them. Most if not all of this information can be gleaned from analysing search data,” said Boffey. Overall, counting on search data, airlines can diligently become a part of a digital economy. By being aware of the intent and what is of interest to passengers, digital teams can embrace rapid learning cycles and work on tailored offering for different channels. 

On how airlines are looking for a deeper insight from the indirect channel, Boffey underlined that is there is low search data availability for GDS-based distribution channels but with the NDC channel share growing that will change and this is the opportunity.

Dealing with ELK and budgeting properly

ELK is acronym for three open source projects: Elasticsearch, Logstash, and Kibana.

Kibana is an open-source data visualization and exploration tool used for log and time series analytics, operational intelligence use cases etc. It is known for features such as histograms, line graphs, pie charts, heat maps etc. Elasticsearch is an open-source, RESTful, distributed search and analytics engine and Logstash is an open source, server-side data processing pipeline that ingests data from log file sources.

As for specific initiatives, Boffey mentioned that whilst some airlines still view NDC analytics as next year’s project, the advanced airlines are starting to deploying open source solutions such as Kibana-based reporting but struggling to get meaningful results.

“Kibana is an attractive reporting interface but the “garbage in” challenge still applies when it comes to what it has available to report,” he said.

Boffey continued, “Logstash is the data collection part of ELK. It provides great support for standard server and website log file formats which is fine for the direct channel.  In the NDC world, you need to pair and convert XML based requests and responses into actionable business intelligence. The ELK stack doesn’t handle XML well.”

According to Triometric, real-time Trio Engine can be used to solve the XML processing problem and feed an ELK stack so Kibana delivers on NDC.

“Open Source technologies such as ELK have apparently low entry costs but the devil is very much in the detail.  Toe-in-the-water style approaches to ELK or other open source technology based projects that lack proper planning are not an appropriate approach to solving this serious data crunching challenge.  Big Data requires heavy lifting.  A clear plan and sizable investment is required - Airlines need to budget properly for it,” said Boffey.

Role of AI in data processing

Boffey summarized and mentioned that today the main focus is still on collecting the right data and getting it processed into a usable form for largely manual decision making process.

“Make the decision and take an action – that’s where the ROI is.  There is a lot of interest in AI technologies.  AI is the automation of processes that in this area will include analytics and related decision making.  AI means more data can be processed and more real-time tuning of systems and that translates into dynamic offer optimisation and revenue.  It all hinges on having the correct data approach. Whilst there is an element of ‘walk before run’, the opportunity for the advancing airline is pretty clear,” concluded Boffey.


How airlines are going about their data strategy?  Hear from senior industry executives at the upcoming Mega Event Asia-Pacific (Ancillary, Loyalty and Co-Brand Conferences) to be held in Bangkok, Thailand (28-30 August, 2018).

Follow Ai on Twitter: @Ai_Connects_Us