Compare two companies by their earnings calls, properly

Pick pairs that share a reality

A comparison is only as good as the shared variable behind it. Direct competitors share end demand: when two chipmakers describe the same AI demand wave differently, one of them is reading it wrong, and finding out which is the whole game. Supplier and customer share a physical flow: the supplier's order commentary this quarter previews the customer's revenue next quarter.

The weakest pairs share only a sector label. Comparing a luxury retailer with a discount chain tells you about the pairing, not about either business. If the two managements are not describing the same underlying reality, side by side is just two monologues.

Compare claims, not adjectives

The unit of comparison is the specific claim: what each company said about pricing, about inventory, about demand by region, about capacity. "Company A sounded more confident" is mood reading. "Company A guided gross margin up 100 basis points while Company B guided flat, both citing the same input costs" is a comparison with content, and it forces the productive question: what does A know, or claim, that B does not?

This is why comparison answers should come as tables with quotes behind them. The table keeps the dimensions honest, one row per topic, and the quotes let you check that the paraphrase did not smooth away the difference that matters.

The comparison questions that work

• "What did A and B each say about demand, side by side, as a table?" The base structure.
• "Where do their explanations of the same market contradict each other?" The disagreement finder.
• "Who raised guidance and who held, and with what wording?" Commitment levels, compared.
• "Compare their margin commentary over the last three quarters." Trajectories, not snapshots.
• "Which analyst concerns appeared in both Q&As, and who answered more directly?" The market's shared worry, and who handled it.

Reading disagreement like a signal

When two companies in the same market describe it differently, there are only three possibilities: different exposure, different execution, or one of them is spinning. Different exposure shows up in the details, segments, regions, customer mix, and a follow-up question usually locates it. Different execution shows up over time: the claims of the better operator keep getting confirmed by the next quarter.

Spin shows up in the reconciliation: the company whose explanation requires the neighbours to be wrong. Three quarters of transcripts almost always settle who was describing reality, and that settlement, checked against what you believed at the time, is the best calibration exercise available to an investor.

From comparison to decision

A comparison is finished when it produces an asymmetry you can name: A is winning share on price, B is defending margin and losing volume; A's guidance assumes the recovery, B's does not. Named asymmetries are testable next quarter, which turns the comparison from an opinion into a tracked position.

In earnings.chat the whole exercise is one conversation: the side-by-side table, the quote checks, the three-quarter trajectory, and next season you extend the same thread and ask whose story held. The comparison does not just answer a question; it becomes an instrument you keep.