Visual daily information needs

Overview

A major search client needed a behavior-led account of how people actually use images to meet daily information needs — not another feature-led survey of visual search. THEFT Studio ran a two-phase study: a seven-day remote digital diary with 30 participants across the US and India captured behavior in context, followed by 16 sixty-minute depth interviews grounded in the diary entries. The work surfaced six mental-model archetypes, three distinct journey types, and an eight-step behavioral map, reframing visual search as a behavior-specific mode rather than a default, with image collection treated as load-bearing to shopping decisions.

The problem

Visual search tooling at consumer scale is a feature-shaped product decision wearing a behavioral question. Every existing study had asked some version of "would you use a visual search feature" and gotten predictable yes-answers from people who had never reached for one in real life. The product team needed to know what people actually did when an information need first surfaced as an image — what triggered it, what tool they reached for, what they did with the result, and how that behavior differed across markets.

That question was unreachable through feature surveys. Behavior in the moment is hard to recall accurately a week later. The shape of the study had to capture it where it lived: in the user's day, on their phone, in their context.

Method

Two phases. Phase one: a seven-day remote digital diary with 30 participants across the US and India. Each participant logged visual information needs in the moment they happened — trigger, tool, query, result, decision, follow-up. The diary surfaced patterns of behavior that would have been invisible to a single interview.

Phase two: sixteen sixty-minute depth interviews, each grounded in the participant's own diary entries. The IDIs pushed on mental models, decision moments, and the role of image collection. Because we were referencing the participant's specific lived behavior — "on day three you saved this image, what happened next?" — the interview signal was sharper than recall-based research could deliver.

Outcome

The product team got a behavior-led framework for where visual search belongs and how it should show up. Product decisions moved from feature-led framing ("should we ship a visual-search button?") to archetype-led framing ("which archetypes does this surface serve, and at what step in their map?"). Image collection was treated as a load-bearing behavior in subsequent product direction, not as an adjacent nicety.

The research framework also had legs beyond the immediate engagement — the eight-step map became an internal reference point that other research and design work was sequenced against.