AI Photo Search

Summary

AI Photo Search is a capability that allows users to search for product(s) by taking a photo or using an image

Role

User Experience Lead

Company

Grainger

Opportunity

Analyzing Graingers product search data, we uncovered a product opportunity.

Customers may not always know the name of the part / product they are looking for. They may also call it by a different name, which the text search engine may not recognize.

Example:
Some popular text searches that returned null results were Hoky and Hotcut

A Hot Cut tool, is a blacksmithing tool used to cut and shape hot steel.
Hoky sweepers that can be used on floors and carpets, and in areas where vacuuming is difficult.

Discovery Research

Our solution hypothesis was to build a product that enables a customer to find a product / part by taking a picture.
In order to realize if we were designing the right thing; we created a quick mobile proof of concept that was powered by the ML model that could perform reverse image searches.

Findings - Discovery Research

Through research, we found that the machine learning model was not accurate at all.

Customers saw benefit in the feature, but lost trust after a few inaccurate search results.

We concluded that the inaccuracy was beacuse the machine learning model was trained curated images and not with real world images. Though customers searched with real world images.

See how we trained the model

Image:
The image shows an example of an inaccurate results. A picture of the scrubbing towels returned results for toner

Userflow

The userflow shows an outline of how user would interact with the camera featues of the grainger app.

Wireframes

The wireframes expand on the reverse image search flow that also includes scenarios where a users can search using an existing photo from their album.

Mockups

Building off of the wireframes, the hi-fidelity mockups focuses on integrating the design system and visual language in the feature.

Demo

The follwing video was created by me and served as a demo for the Photo Search feature. It was presented by senior leadership at graingers annual board meeting, at the company townhall and shared by associates over LinkedIn