AI & Technical question

Wayfair Sells to B2C as well as B2B. However there is no easy way to know which are B2B customers. Imagine that you have access to click stream data, order history data or any other data what attribute will you will look at to know whether a customer is B2B?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Data and feature reasoning: identifying the right signals to classify B2B versus B2C customers from behavioral data.

Clarify the goal: build a classifier to segment existing customers, or flag B2B at signup, since available data differs.
Propose the strongest signal: order size and frequency, and purchases of business relevant categories like bulk furniture.
Consider clickstream signals: repeated browsing of bulk options, business email domain if captured, and shipping to a commercial address.
Consider account signals: tax exempt certificate usage or requests for invoicing or net 30 terms, strong explicit indicators.
Propose combining signals into a score rather than one rule, since no single attribute is definitive.
Define validation by checking the model against any known labeled B2B accounts to test precision.

Prioritizes order size and category as the strongest behavioral proxy, rather than guessing a single arbitrary field.
Notes explicit signals like tax exempt or net 30 requests, much stronger than inferred behavior alone.
Proposes a combined score model and validation against existing labeled accounts, showing applied rigor.

Naming only one attribute without combining multiple signals into a robust classifier.
No validation approach to check if the classifier's guesses are actually correct.

How would you handle a customer who buys both personally and for their business?
What would you do if you had no labeled B2B accounts to validate against?