A new approach to analytics for improved CX takes off at Finnair

Profile picture for user jtwentyman By Jessica Twentyman October 15, 2019
At the Finnish national carrier airline, a new head of data is helping the company to navigate a new approach to analytics, in order to provide the best possible customer experience to passengers


When Minna Karha stepped into newly created role of head of data at Finnair two years ago, she knew she was taking on a sizeable but exciting challenge. Not only would she responsible for defining the role itself, but also a data strategy for the airline to take it to 2025 and beyond. Along the way, she’d need to design and build a data architecture to support that strategy, as well as assembling a skilled team of data engineers.

As a mark of how far she and her team have travelled since take-off, the national carrier of Finland, which will celebrate its 100th year in operation in 2023, now uses a machine learning-based model to predict flight delays. This combines Finnair’s own historical data about flight operations with third-party data on weather patterns, plus data on runway capacity from its home hub of Helsinki Airport, in order to determine the probability of disruptions and their likely impact on schedules.

With this successful use of machine learning under their belt, Karha and her team are now looking for other ways to apply the technology, not least for increased personalization in the customer experience (CX) that Finnair offers. In fact, CX is a major focus for many of the analytic use cases that will be supported by a new data platform from cloud-based data warehousing company Snowflake, implemented earlier this year. As Karha puts it:

The potential for using data and analytics for business development within Finnair is huge, but in the past, it had been very challenging to combine datasets from across the airline. Everybody was using their own environments, with business units setting up their own [AWS] Redshift clusters or other databases and trying to combine data by themselves.

So what we wanted to do was build a centralized data platform that enables scalable and secure use of data for analysis, so that teams within Finnair don’t need to build their own solutions but can simply turn to a single collaborative place where everybody can work together.”

This makes a lot of sense, when you consider that the customer experience of purchasing and taking a Finnair flight, by definition, is the product of work carried out by employees from right across the business and, in many cases, those at its partners, such as code-share airlines and airports.

Clearance for take-off

Following a proof-of-concept conducted late last year, the Finnair team was satisfied that Snowflake could provide the data platform it was seeking to build. This is underpinned by a data lake based on the Simple Storage Service (S3) from Amazon Web Services (AWS) and a data catalogue that gives Karha’s team an at-a-glance view of all the data assets Finnair holds, as well as their provenance. This, she says, ensures that no potentially valuable source of data for a particular analysis is hidden away or missed. The infrastructure, which went live at the beginning of 2019, is front-ended mostly by tools from PowerBI for dashboarding and other visualizations, as well as some custom-coded reports.

Today, other CX-focused use cases that Karha and her team are working on include analysis of customer feedback, gathered by customer service agents from direct interactions with passengers and from social media streams, in order to determine the types of issues that are most frequently reported by customers and tackle them.

The team has also been able to build unified customer profiles that give marketing colleagues a richer insight into the habits and preferences of individual passengers and thus target their marketing campaigns more efficiently.

A third use case focuses on revenue management and pricing - a huge concern for all airlines, in the face of intense competition between them and relentless downward pressure on price. Here, better analysis will enable Finnair to price flights - and individual seats on those flights - in a way that keeps fares competitive, while maximising profitability:

We’ve already been able to onboard our first users, business analysts, who now rely on this architecture for data analysis - and they’ve been super-happy because the environment is much more scalable and, at the same time, they don’t get the same performance challenges that they experienced in the past.

We have also introduced the role of ‘data asset product owner’, which means that we now have five data domains, representing the core data assets of the airline, and a dedicated product owner for each, who is responsible for making sure that we put our development effort into the most valuable use cases. This helps us to see where we can best deliver business value through data analytics and also assists in identifying new opportunities in areas where we don’t use analytics yet or are not getting full value from the data we hold, but where new insights would assist in better decision-making.

Looking ahead, Karha sees a big role for analytics in Finnair’s overall push to become more sustainable, which has driven a number of recent initiatives at the airline, focusing on electric aviation and biofuels. Says Karha:

We already get a lot of sensor data from aircraft, and I see huge potential in using that data to see how certain weather conditions affect fuel consumption, for example. I’m sure there’ll be an option also to maybe provide pilots with some sort of insight and knowledge into how they can fly in the most sustainable ways possible. I see a lot of possibilities ahead.