WhoBuilt

Google Photos Product Lead and Bump Cofounder

Apr 10, 2019· 57 min· 9.8K views·Watch on YouTube

Speakers

David Lieb, Google Photos product director and Bump cofounder, discusses product discovery, startup growth, and building Google Photos with Gustaf Alström­er on Y Combinator’s podcast.

Bump’s launch and sudden growth

Lieb and two cofounders built Bump as a side project to make sharing phone numbers easier. Its phone-bumping novelty helped it spread by word of mouth, and being the billionth app downloaded from Apple’s App Store brought global attention. An Apple television ad later sent traffic surging, overwhelming Bump’s servers until YC founders helped fix them.

Growth, retention, and business fit

Bump reached 150 million downloads and about 10 million monthly active users, but the team initially emphasized usage totals over cohort retention and frequency. Lieb says people kept Bump installed but used it infrequently, and each use created too little value to support the venture-scale business they had pursued. Small monetization tests suggested roughly one dollar per user per year.

From Bump and Flock to Google Photos

Calls with Bump’s top users revealed that many mainly shared photos, which led the team to build Flock around automatic sharing between friends who took photos at the same time. Flock struggled because new users needed friends to join before it was useful, so the team explored a photo gallery app called PhotoRoll. Google acquired the team in 2013, and its product design insights combined with Google’s work on image recognition, search, and backups to help shape Google Photos.

Designing for users and leading at scale

Lieb describes imagining a tireless photo assistant that could back up pictures, identify people, make movies, and surface meaningful memories. At Google, he divides his time among product strategy, team development, product decisions, and feedback. He says large-scale user feedback includes thousands of daily reports, clustering tools, research, and direct conversations, while good product decisions still require interpreting what users say.

AI, focus, and durable needs

Lieb recommends using AI either to scale tasks people can do but cannot perform affordably at scale, or to tackle problems beyond human ability. He cautions teams against polishing performance when they may be solving the wrong problem. For consumer products, he looks for a durable human need that technology can serve more effectively, rather than a temporary novelty.

Takeaways

  • Track retention and usage frequency, not just downloads or total activity.
  • Ask users what they actually do with a product; Bump’s top users revealed photo sharing was its main use case.
  • Give new users value before asking them to recruit friends.
  • Test whether AI makes a real task cheaper, more accessible, or possible at all.
  • Before polishing a feature, check that it solves a problem users care about.

Mentioned in this talk

Search inside this talk

What was claimed 19

Show all 19 facts

Related talks