How Pinterest’s Guided Search Redefines Visual Discovery

During his student days, Pinterest chief designer Jason Wilson struggled with complex topics until a friend introduced him to mind maps, a visual method that breaks down subjects into small, connected blocks using simple tools like sticky notes.

Wilson applied this principle to Pinterest’s Guided Search, launched on mobile in late April, allowing users to enter a basic keyword such as “shoes” and receive a cascade of related items, progressively narrowing down options to match user intent.

Although visual search was not originally a core mission for Pinterest, Wilson convinced CEO Ben Silbermann to assemble a dedicated team, arguing that visual discovery would help users find and share photos, recipes, locations, and videos more easily.

Backed by a rapidly growing user base, over $500 million in funding, and a valuation of $3.8 billion as of last October, Pinterest added this extra search layer to attract more returning users and present them with precisely what they care about.

For example, when searching for Mother’s Day gifts, Google returns precise results for “where to buy roses,” whereas Pinterest’s guided search narrows the query step‑by‑step, helping users discover the perfect present.

Wilson, who previously worked at Facebook, Apple, and Lytro, spent months in Pinterest’s underground research lab developing the feature, emerging only after the project’s completion.

He likens guided search to a different type of gasoline: while Google relies on PageRank—evaluating site authority, crawling the web, and tracking user behavior—Pinterest’s approach focuses on what users are collectively searching for and what’s trending at a given moment.

Google’s dominance as a search engine is undeniable, but Wilson notes its limits, citing personal expertise that can surpass algorithmic knowledge in niche areas.

Guided search does not require detailed knowledge of individual user actions; instead, it assumes Pinterest’s community can collectively identify what matters most, without exhaustive algorithmic crawling.

Observing user habits, Wilson found that people often delete earlier keywords to refine results (e.g., removing “roller skates” when searching for “shoes”). Guided search automates this by letting users complete a sentence with a single keyword, reducing manual effort.

Currently, Pinterest supports two‑dimensional search—combining user input with selected topics. Wilson aims to expand to three‑dimensional or even multi‑dimensional search, integrating image recognition and visual search, as hinted by the recent acquisition of VisualGraph.

Early user feedback indicates guided search’s success, and Wilson, despite his unconventional ideas, remains focused on continuously improving the service to attract more users.