Turning earnings calls into leasing signals
An MBAi capstone for an industrial real estate company whose teams read earnings calls, 10-Ks, and trade press by hand to guess what customers will need next. We're building a system that turns those sources into signals each team can act on. In progress.
- Organization
- Kellogg MBAi capstone, industrial real estate client
- My role
- Business side of a five-person team
- When
- Fall 2026
The problem
The client’s teams track what their customers are saying in earnings calls, annual filings, and trade press to anticipate demand for warehouse space. Today that reading happens by hand, across teams that don’t share notes, and what one group learns rarely reaches the others in time to matter.
What we’re building
I’m on the business side of a five-person team, with two teammates owning the technical build. The work runs in three parts:
- Define what counts as a signal. A C-suite briefing and a leasing rep’s meeting prep need different things, so we start by interviewing the internal audiences and pinning down which mentions actually predict space needs versus general business noise.
- Extract and classify. Ingest public sources, pull out the mentions that matter, and tag them by customer, market, and type of signal.
- Go from signal to implication. Each output says what the signal might mean for the client, with a confidence level attached rather than stated as certain, and it’s formatted for the audience reading it.
The hard part
Extraction is the easy half. The real product question is which claims are worth surfacing, at what confidence, to whom. Too strict and the system says nothing useful, too loose and people stop reading it.
I’ll update this page with results after the final client presentation at the end of the quarter.