← Back to work

Agoda (Booking Holdings) · UX research

What users do when they think no one's watching.


Role: UX research — design & execution  ·  Years: 2015–2019  ·  Methods: Mixed-methods, cross-cultural  ·  Markets: 14 countries

Context

Prior research had confirmed that hotel and room photos were one of the biggest factors in a customer's decision to book on Agoda. The Content team needed to raise photo quality across the platform, and to do that they needed a precise vendor specification — the exact criteria for quality, composition, angle and coverage. My brief: define, from validated user insight rather than assumption, what makes a hotel photo convert.

The problem

The standard approach — interviews across markets showing people photos and probing their preferences — broke down in early fieldwork. Participants had strong reactions but couldn't articulate them. Words weren't an adequate tool for visual preference, and the discussion aid was producing superficial answers that wouldn't survive contact with a vendor spec. Watching that happen live, I made the call to pause the study and rethink the method before going further.

What I did

Instead of relying on words, I designed an activity around the Premo tool — a non-verbal emotion-measurement instrument (a cartoon character expressing 14 emotions, built on Plutchik's emotion wheel) that works across cultures precisely because it isn't language-dependent. Participants were shown a curated set of hotel photos where each primary image had controlled variations — camera angle, items in frame, colour filter, presence of people — and assigned an emotion and an intensity to each. Emotion plus intensity on controlled variants made implicit preference suddenly explicit: people could specify what made a photo compelling without needing the language to describe it.

Outcome

Fieldwork resumed across 14 countries, and the method produced unusually precise, actionable findings — low and eye-level angles, natural light, compositional clarity, and context-dependent rules for human presence. Synthesised across markets, it became an evidence-based hotel-photo vendor specification tied directly to booking-intent signals.

What it proves

Built as part of standing up Agoda's research function from zero, this shows research leadership as owning the quality of the insight — having the judgment to stop a failing study, invent a method to crack a problem standard tools couldn't, and restart rather than push through with compromised data.

  • Research methodology design
  • Cross-cultural research
  • Emotion measurement
  • Mixed methods

The deck

Want the detail behind this? Let's talk →