How to track tariffs and supply chains through earnings calls

Exposure is specific, and companies say so

Two companies in the same industry can have opposite tariff positions: one imports the components it assembles, the other manufactures where it sells. Both are "exposed to tariffs" in a headline and only one of them pays. The transcripts separate them, because management is asked to be specific about sourcing, and vague answers on that question invite follow-ups they would rather avoid.

The specific things to look for are the share of cost of goods sourced from an affected region, whether pricing can absorb it, and whether existing contracts lock in terms for a period. A company that can pass a tariff through to customers within a quarter has a materially different problem from one whose prices are fixed for a year.

Where the tariff conversation actually sits

Across our current-quarter coverage, 430 calls carry tariff or trade-policy language in their extracted evidence. That is roughly a quarter of the number that mention AI, which is itself a useful calibration: the topic is loud in the press and comparatively narrow in the transcripts.

The concentration is where you would expect on reflection and not where the headlines put it. Consumer Discretionary leads with 129 calls, Industrials follows at 122. Health Care is at 47, Materials at 36, Consumer Staples at 27. Information Technology, the sector with arguably the most globally distributed supply chain of all, sits at 28.

The reason is who owns the customs exposure rather than who depends on the goods. A retailer imports finished product under its own name and pays the duty directly. A software company with hardware in its supply chain is two steps removed, and the cost arrives as a supplier price increase that gets discussed as input cost rather than as tariffs.

That distinction matters for research: searching for tariff language finds the companies that pay it, not the companies affected by it. To find the second group you have to search for the pass-through, which is a different set of words entirely.

The vocabulary problem, and why it matters

Executives rarely say "tariff" as often as you would expect. They say customs duties, import costs, trade headwinds, cross-border friction, or they name the policy directly. A keyword search for one word finds one slice of the discussion and silently misses the rest, which is how a search can return three results and leave you thinking the topic barely came up.

This is the single biggest practical failure in transcript research. The fix is to search the way the topic is spoken about rather than the way it is labelled, and to check more than one phrasing before concluding that something is absent. An absence is a finding only if you looked for it properly.

Questions that map the exposure

• "Which companies named tariffs as a cost headwind this quarter, and what did they quantify?" The base screen.
• "Who said they can pass tariff costs through to customers, and who said they cannot?" Splits the same exposure into two very different outcomes.
• "Which companies are moving production, and where to?" Relocation is expensive and slow, so saying it out loud is a commitment.
• "How did the tariff language in this sector change from last quarter to this one?" Escalation and de-escalation both show up in wording first.
• "Which suppliers to this company mentioned the same tariffs?" Supply chains run through several transcripts, and the story is more complete read across them.

Reading a supply chain across companies

A tariff rarely stops at one company. The importer talks about input costs, its customer talks about price increases, and the customer's customer talks about demand. Read one transcript and you get one link. Read the chain and you can see where the cost is actually landing, which is often not where the headline says.

This is work that scales badly by hand and well by search: the same question, asked across a set of related companies in one window, with the answers side by side. What you are looking for is the point in the chain where the language changes from "we absorbed it" to "we passed it on", because that is where margin is moving.

How earnings.chat helps

The archive holds more than 253,000 earnings calls, and questions can be scoped by country, sector or period. That matters here because tariff exposure is geographic before it is anything else: asking what companies in one country said about import costs is a different question from asking the same of their competitors elsewhere.

Every answer carries the quotes and the speaker behind it, so a claim about exposure can be traced to the call where it was made. When a search comes back empty, the answer says so rather than filling the silence, which for a topic with this much vocabulary drift is the difference between a finding and a false negative.