User research is the foundation of good design. Any successful product you can think of is driven by user insights. And, while all UX designers tap into the same pool of tools and techniques, you’ll find that every team has their own unique approach.
Now, with the rise of AI-powered products, user research is constantly expanding. Studying real-world UX research examples helps us understand how the field is evolving, and how some of the best products are created.
Are you curious about how some of the biggest brands conduct UX research? Then keep reading. In this post, we take a deep dive into four UX research case studies:
- Airbnb: The power of observing behaviour to uncover design opportunities
- Google Gemini: The importance of designing trustworthy AI experiences
- Spotify: The value of human perspectives in a data-driven world
- Spotify: Making personalisation scalable with agentic AI
Each of these case studies teaches us a valuable lesson about UX research — lessons you can apply to your own design projects. So let’s jump in!
UX research case study #1: Airbnb and the power of observing user behaviour to uncover design opportunities
Oftentimes, user research is planned in advance and conducted within a controlled setting. Think user interviews, or analysing how people interact with your website over a specific period of time.
But sometimes, user research occurs organically, like an accidental light shining on a major design opportunity. That’s exactly what happened at Airbnb, leading to the design and launch of a new global check-in tool.
Vibha Bamba, Design Lead on Airbnb’s Host Success Team, writes:
“The decision to design the tool was informed by an intriguing host behaviour. We noticed that about 1.5 million photo messages were being sent from host to guest each week, the majority of them to explain location and entry details. Photos of the home were juxtaposed with maps, lockbox locations were described, and landmarks were called out.”
Observing these behaviours over time, the Airbnb team realised that there was a huge opportunity to make the exchange between hosts and guests much more seamless and consistent. This kicked off a year-long project to design a global check-in tool for the Airbnb platform.
The result? An integrated check-in tool that enables hosts to create visual check-in guides for their guests. They can upload photos and instructions which the tool will translate depending on the guests’ preferred language, and the guides can be accessed both on and offline.
And, after launching the tool, the team continued to observe how hosts used it. They were able to flag issues and further design opportunities, adapting and evolving the check-in tool to better meet hosts’ needs. That’s the power of observing user behaviour!
The takeaway
User behaviour provides us with incredibly rich insights. Don’t rely solely on planned or periodic user research. Continuously observe how people interact with your product in the wild, too. You don’t know what you don’t know, and this approach will help you to uncover design opportunities you may not have even thought to look for otherwise.
UX research case study #2: Google Gemini and designing trustworthy AI experiences
Generative AI has completely changed how people interact with digital products. Instead of clicking through links or navigating structured interfaces, users can now ask questions conversationally and receive instant AI-generated responses.
But designing AI-powered experiences comes with a new set of UX challenges. How do users decide whether to trust an AI response? What happens when the system sounds confident but gives incorrect information? And how should conversational interfaces behave when users ask vague or complex questions?
These are some of the challenges UX researchers are now exploring as they develop AI-powered search experiences.
Discussing the future of UX research in the age of AI, UX researcher Jess Holbrook explains that generative AI products are fundamentally changing how researchers study user behaviour and interaction. Unlike traditional interfaces, AI-powered experiences are dynamic, conversational and often unpredictable, requiring teams to rethink how they evaluate usability, trust and user confidence.
Take Google Gemini for example. As users increasingly interact with conversational AI tools, UX researchers are paying closer attention to how people interpret AI-generated responses, navigate conversational interfaces, and decide whether information feels trustworthy or helpful.
These evolving UX challenges are already influencing how products like Gemini present information and guide user interactions. Features like source linking, follow-up prompts, and conversational refinement tools are all designed to help users better navigate and interpret AI-generated responses.
The result? A growing shift in UX research methods for AI-powered products, where researchers must evaluate not just whether users can complete tasks, but whether they feel informed, confident, and in control while interacting with AI systems.
The takeaway
AI-powered products require UX researchers to think beyond traditional usability metrics. When designing for conversational AI, it’s important to study how users interpret, trust and collaborate with AI-generated outputs; not just whether the interface is easy to use.
As AI becomes more integrated into digital products, UX researchers will play an increasingly important role in helping teams design experiences that feel helpful, transparent and trustworthy.
UX research case study #3: Spotify and the value of human perspectives in a data-driven world
Data is a powerful research tool. It enables you to gather and analyse broad and vast user insights, to make evidence-backed decisions, and to track and measure important UX KPIs.
But, as Nhi Ngo, Insights Manager, User Research & Data Science at Spotify will tell you, it’s important not to become over-reliant on data when conducting UX research. Sometimes, making the best design decision boils down to a human perspective.
Nhi Ngo came to this realisation when developing and launching a feature called “Shortcuts” on the Spotify Home tab. Powered by machine learning, Shortcuts is a dedicated space that showcases the user’s current favourites, as deduced by Spotify’s algorithms.
The feature was developed based on data collected through a variety of research methods, including longitudinal user studies and A/B testing.
So far, so good. But when it came to deciding on a name for the feature, A/B tests came back inconclusive.
In the end, the name was decided based on the product designer’s instinct to go with the name that would create the most human and personal experience. Nhi Ngo explains:
“A few candidates that were tested were ‘Listen Now’ (the objective that the model optimizes for), ‘Shortcuts’ (the user-facing functionality), ‘Quick Access’ (a UX goal of this space), and last but not least, a daypart greeting, ‘Good morning’ (that would change with the time of day to ‘Good afternoon’ or ‘Good evening’).
We were counting on the AB test to help us make this important decision. The test returned neutral. Our designer recommended we go with the daypart name, much to my reservations.
Indeed, participants were most often positively surprised in our interview sessions whenever they opened their phone and saw the greetings. Convinced by our designer’s humanistic approach and recognising the intangible benefits of providing users with this joy of being ‘greeted by Spopify’, we decided to go with our perspective-taking as humans to humans, and chose the daypart name.”
The result? A new product feature that evoked delight in Spotify’s users and led to further improvements, such as incorporating more time-based features in the model so that the recommendations changed depending on the time of day (for example, showing sleep music playlists at night).
The takeaway
Data-driven research is an extremely powerful tool, but it may not always give you the full picture or a conclusive answer. Whenever you conduct and interpret research data, it’s important not to lose sight of your human perspective.
In the words of Nhi Ngo: “When data can’t give you a definitive answer, it is OK to be human and make a human decision. Prioritise user joy; treat them as you would any human in your life.”
Read the full UX research case study here: It’s OK to be Human in a Machine-Learned World.
UX research case study #4: Spotify and making personalisation scalable with agentic AI
Personalisation has long been a major part of the Spotify experience. But as AI-powered recommendation systems become more advanced, Spotify researchers are now exploring how music recommendations can adapt more dynamically to users’ changing tastes, moods and listening habits over time.
Traditional recommendation systems work well when users search for something relatively straightforward, like a specific artist, genre or song. But as Spotify researchers explain, these approaches become much more limited when users make nuanced or situational requests — such as wanting “music for a solo night drive through the city.”
In these situations, traditional recommendation systems often default to broadly popular or familiar content rather than capturing the specific mood or intent behind the request. As Spotify researchers note, a user skipping a high-energy pop track in favour of a moody electronic track may actually provide a much stronger signal about the kind of listening experience they’re looking for.
This raised an important challenge for Spotify researchers: how can recommendation systems continuously learn from user behaviour and feedback to deliver more nuanced, intent-driven recommendations over time?
To tackle this, Spotify researchers explored the use of LLM-based “agentic AI” systems capable of interpreting conversational requests, orchestrating recommendation tools and adapting based on ongoing preference feedback.
Now, instead of treating each interaction as isolated, the system continuously learns from behaviours like plays, skips, saves and refinements to better understand user intent and listening preferences over time.
The result? A more adaptive recommendation experience that moves beyond static recommendation algorithms and towards more conversational, context-aware personalisation.
The takeaway
Spotify’s research shows that good personalisation isn’t just about recommending more content. It’s about understanding the experience the user is actually looking for. As recommendation systems become more conversational and adaptive, UX researchers must increasingly focus on user intent, evolving preferences and long-term satisfaction.
Read the full research here: Personalizing Agentic AI to Users’ Musical Tastes with Scalable Preference Optimization.
What these UX research case studies teach us (and how to apply it to your own work)
While every company takes a different approach to UX research, these case studies highlight several common themes that UX designers and researchers can apply to their own projects.
UX research doesn’t stop after launch
Airbnb’s check-in tool evolved through continuous observation of real user behaviour, while Spotify’s recommendation systems continuously learn from ongoing user feedback and interaction patterns. The most valuable UX research often happens after a product or feature is already live.
How to apply this: Continue gathering feedback and observing behaviour after launch. Analytics, support requests and user interaction patterns can all reveal new opportunities for improvement over time.
User behaviour often reveals more than direct feedback
Several of these case studies show the importance of observing what users actually do, not just what they say. At Airbnb, researchers identified a design opportunity by noticing how hosts were sharing check-in information through photo messages. Likewise, Spotify researchers use behavioural signals like skips, saves and refinements to better understand listening preferences and user intent.
How to apply this: Look beyond interviews and surveys alone. Behavioural data, search patterns and real-world usage can often uncover unmet needs that users themselves may not explicitly articulate.
Modern UX research goes beyond usability alone
Traditional usability testing still matters, but newer AI-powered experiences are introducing entirely new research questions around trust, transparency, intent and emotional response. Google’s work on conversational AI experiences highlights how UX researchers are increasingly studying how users interpret and collaborate with AI-generated systems.
How to apply this: When researching AI-powered or highly personalised experiences, evaluate not just whether users can complete tasks, but whether they feel confident, informed and in control throughout the experience.
Quantitative and qualitative research work best together
Data can reveal patterns at scale, but human insight remains essential for understanding context, motivation and emotion. Across all of these examples, UX researchers combine behavioural data with human-centred research methods to create more meaningful user experiences.
How to apply this: Use quantitative research to identify patterns and qualitative research to understand the reasons behind them. Combining both approaches will give you a more complete understanding of your users and their needs.
|
Learn more about UX research
All of these UX research case studies emphasise the importance of user research in UX design.
If you’d like to learn more about UX research, check out the 9 best UX research tools, read about a day in the life of a UX research manager with Google’s Dr. Stephen Hassard, and master the art of analysing your UX research and pulling out useful insights in this guide.
Emily is a professional writer and content strategist with an MSc in Psychology. She has 8+ years of experience in the tech industry, with a focus on UX and design thinking. A regular contributor to top design publications, she also authored a chapter in The UX Careers Handbook. Emily also holds a BA in French and German and is passionate about languages and continuous learning.
