PinterestDiscovery5 min readA save button that doubles as research
Collecting for later is the input the recommender needs.
- The surface
- The masonry home feed, the save action on a pin, board selection, and the related pins shown after a save.
- What the user wants
- I am planning something and I want to gather ideas somewhere I can come back to.
The masonry grid
Images are laid out in columns of varying height with minimal text, and the grid loads continuously as the user scrolls.
Visual scanning over reading
The user's intent here is inspiration rather than an answer, so they are comparing images at speed. Uneven heights break the row rhythm that lets eyes skip content, which keeps more items actually looked at. Text would slow a process that works best fast.
Save as the primary verb
The dominant control on every pin is save, placed over the image and available without opening the item.
Intent capture at the point of interest
A like says the user enjoyed something; a save says they intend to use it. That is a far stronger and more specific signal, and it costs the same single tap. Making the stronger signal the easiest action is the whole trick.
Choosing a board
Saving prompts for a board, with the user's existing boards listed and an option to create one.
User-supplied taxonomy
The board name is a label the user wrote for their own purpose, which tells the system what the pin is for rather than what it depicts. Millions of these labels amount to a classification no internal taxonomy could match. The user does the work because it benefits them directly.
More like this, immediately
After a save, related pins are shown right below or in place, tuned to what was just saved.
Closing the loop on intent
The moment after a save is when the user's intent is clearest and their attention is already on that topic. Responding instantly makes the system feel like a collaborator rather than a feed. It also compounds, because each save sharpens the next set of suggestions.
The board as the payoff
Boards are browsable collections with a cover image and can be shared or made secret, and sections allow subdivision.
Endowment through curation
The value the user takes away is not any single pin but the collection they assembled, which exists nowhere else. That collection is the switching cost. Secret boards matter because a lot of planning is private, and forcing it public would suppress saving.
Put the strongest signal on the cheapest action
What not to copy
- The feed is infinite and heavily commercial, and sponsored pins are visually near-identical to organic ones, which blurs whether a save reflects taste or ad spend.
- Saved boards rot. The product is excellent at accumulation and weak at the moment the user actually comes back to execute a plan, which is where the promised value was.
- Much of the source content is scraped context-free, so a pin can lead to a dead link or an unattributed creator, and the platform carries the benefit while the origin carries the loss.
- Aspirational imagery at volume has a known effect on body image and spending, and an engagement-tuned feed has no mechanism that notices.
The takeaway
Make the action with the strongest intent signal also the cheapest one, and let users name their own categories.
Finished the teardown? Bank it and the day counts toward your run.
Where the principles come from
- Jobs to Be Done: intent over demographics, Clayton Christensen
- The Design of Everyday Things, Don Norman
- Infinite scroll usability, Nielsen Norman Group
Written from public behaviour of the product, not from inside it. Interfaces change often, so treat the flow described here as of the time of writing and check the live product before quoting it.