Article
September 30, 2025 · 5 min read timeAt our Product Owner Meetup, we were excited to host two highly experienced digital product professionals from Finnair: Emmi Sallinen, Digital Product Lead, and Olga Oksanen, Digital Product Manager. We discussed how data guides decision-making at Finnair and how it’s embedded in every part of their product work. In this article, Emmi and Olga share the processes and mindset of data-driven decision making, showing how a structured approach to data helps make sense of complex customer behaviour and drives meaningful, measurable outcomes.
At Finnair, we work at different ends of the customer journey. Emmi focuses on everything leading up to the flight booking, while Olga owns the post-booking experience, including preparing for the trip and check-in. Though our responsibilities and user volumes differ, our approach to product development is the same: a shared mindset, processes, and frameworks enable data-informed decision-making throughout the journey.
Two journeys, one approach
Finnair.com is Finland’s largest e-commerce platform. On Emmi's side of the journey, traffic is massive (we’re talking millions of users), and user behaviour can be challenging to predict. On Olga’s side, post-booking interactions are fewer in volume but very critical in nature. We have large amounts of data from the entire customer journey, and the challenge is to process this data so it is actionable, understandable, and capable of guiding decision-making.
Pre-booking challenges: People have varying needs for their travel, as well as different motivations for choosing Finnair. Customers may browse flights on mobile for a single passenger but end up purchasing flights for their family on desktop. The flight search phase becomes challenging when it differs from the actual need, and we lose the connection between searching and buying when these activities happen on different devices. To offer a more personalised experience, we aim to connect data points across devices and channels. To make the most of this, we recommend that customers log in to the Finnair Plus profile whenever possible.
Another challenge we often face is segmenting people when their needs shift. One day they might be planning a family holiday, and soon after, a solo business trip. Personalisation is expected, yet hard to execute well. That is why we focus on patterns instead of assumptions, and prefer no personalisation over poor personalisation.
Post-booking realities: We see fewer users, but everyone is already a paying customer. Small issues can have a huge impact even if they affect only a handful of customers. For example, a glitch in the flight change process once left some customers confused about whether their ticket was actually updated. This issue was almost invisible in the numbers but surfaced through logs and customer support feedback.
Even though we have high volumes of quantitative data, it requires business knowledge to be put into context. For instance, we need to understand how much of our total customer base is represented in certain peaks. This is why qualitative data and close collaboration with customer service become essential. Quantitative data can give us a trend, but qualitative data helps us understand the story behind it.
Safety and privacy at the core
At Finnair, safety is always our top priority, including privacy and data responsibility. When we discuss how we use data, we start by ensuring our customers' personal information is protected. All data is collected according to our strict security and safety standards, and we make sure dashboards never have any personal information exposure. Respecting privacy is a cornerstone of our work, and data is collected, stored, and used only according to the consent our customers have given us.
We work with both quantitative and qualitative data, and our dashboards are used across different roles: developers, designers, analysts, and product managers all use them. Even our executives have dashboards tailored to support their decision-making. These tools help us visualise behavioural patterns, monitor funnel performance, understand conversion rates, and track live experiments.
Trusting our data enables fast releases
Even the clearest dashboard won’t help if we don’t know how to interpret it. For example, if negative feedback suddenly stops, we treat that as a positive sign. Silence often means things are working as expected. People usually don’t give feedback when everything goes well.
Therefore, when working with data, the first questions we ask are:
Is this meaningful?
Why is this happening?
We release updates to Finnair.com twice a week. Our teams are free to test and launch independently, without unnecessary approvals. This is possible because of the trust we have in our data, people, and processes. It allows us to make decisions quickly and also pull back or correct errors rapidly if needed. Also, we never give release dates or prioritise based on requested deadlines. Customer impact is our main guiding metric, followed by effort and confidence.
Our tips: How we make data work in practice
Even though intuition can guide us toward interesting new areas, we never rely on it alone. Data doesn’t lie, but interpretation can. We let intuition help us ask better questions, and trust the data to answer them. To sum up our approach to data-driven decision-making, here are our key insights:
Ensure data is both reliable and variable: Use trustworthy sources and ensure your data captures a range of behaviours. Diversity in data leads to more robust insights.
Data doesn’t lie, but interpretation can: Numbers are objective, but conclusions are not. Always validate your interpretations.
Choose the right validation approach: Not all hypotheses require direct customer interaction. Match the method to the risk and impact of the decision.
Trust your gut, but be ready to be wrong: Intuition is a valuable starting point. However, be prepared to pivot quickly - fail fast, learn faster.
AI has become a part of our toolbox. We use it to summarise feedback, predict outcomes, and support discovery work. In a recent discovery sprint, we invited Copilot to participate by feeding it open-ended questions. It helped surface unexpected ideas because it doesn’t self-censor. Sometimes, seeing a suggestion from AI helps validate a thought we might have otherwise dismissed too quickly.
The bottom line
At the end of the day, placing data at the centre of our work helps us prioritise what matters most to our customers. It’s a way to “hear” what our customers need, even when they don’t tell us directly. If we could share one thing with other product teams, it would be this: the data itself is rarely the challenge. The challenge is listening to it continuously and building capabilities to read between the lines. That is when we can continuously improve the customer journey.