Google · AI product · Mid · 45 to 60 minutes

Should Google Photos organize clothes in photos?

Google is a technology company whose products help people search for information, store media, and use software across phones and the web. Answer this product management case and get it graded line by line against a weighted rubric.

product sense
AI scoping
prioritization
metrics
risk management

The situation

You are a PM on Google Photos. The team has made photo search and automatic organization useful for people looking for pets, trips, faces, and screenshots, but there is a new question from leadership: should Photos help users find clothing they have worn in past pictures? The idea sounds simple, but the product case is not. People might want to refind an outfit for travel, shopping, events, or sharing, yet clothing is harder to recognize than broad photo categories and mistakes could feel creepy or low value. Engineering says modern vision models can probably detect apparel in images, but grouping many photos of the same garment, across lighting, poses, and partial views, is much less certain. Design is worried about cluttering the product with a niche feature, and privacy reviewers want a clear user benefit before the team scans personal libraries in a new way. Your director asks you to recommend whether Google Photos should invest now, and if so, what the first version should be.

What to produce

Write a 1 to 2 page product strategy memo.

Your answer must

Define the primary user problem this feature would solve, and name the target users and top use cases.
State a clear recommendation on whether to build now, later, or not at all, and explain the trade offs.
Propose an MVP scope, including what the user can do in the product and what should be left out of the first release.
Explain a plausible AI approach at a PM level, including how apparel might be detected and how photos of the same garment might be grouped, plus what existing capabilities or libraries could be reused.
Specify success metrics, quality thresholds, and a lightweight launch or experiment plan.
Identify key risks, including privacy, trust, and model error, and give mitigations.

Constraints

Do not assume access to retailer catalogs, purchase history, or external fashion partners at launch.
The feature must work from users' existing photo libraries without asking them to retake photos in a special format.
You cannot assume perfect garment recognition, especially when items are partly hidden or shown in different lighting and poses.

Format

Use headings for user problem, recommendation, MVP, AI feasibility, metrics, and risks.

How Google actually played it

Analyzes Google Photos' wardrobe feature that scans photos to catalog clothing, including the clustering approach that groups images of the same garment.

Spot a user problem behind a feature
Understand the technical approach enough to scope it
Use existing libraries to create new value

Built from How Would You Build Google Photos’ New Wardrobe Feature? by Technomanagers.