FactIQ
Browse documentation

Find the signal in earnings calls

Use management's own words to understand what matters now, what changed, how peers differ, and whether the story holds up against the numbers.

On this page

Start here

Earnings-call research is most useful when it resolves one of three questions: what changed, what is different, or what can be tested.

What changed?

Follow one claim across calls instead of summarizing each call in isolation.

Who is different?

Compare peers on the same theme and equivalent dates.

Does it hold up?

Test management's story against guidance, filings, peers, and outside data.
Check company and date coverage →Use the live inventory before choosing a period

Choose sharp questions for the calls

Do not just ask “summarize these transcripts.” Begin with the conclusion you want the evidence to resolve.

What matters now?

Latest-call note

Find the few developments, challenges, and watch items that could change the next two quarters.

What changed?

Multi-quarter change

Track one claim consistently through time and distinguish a real change from different wording.

Who is different?

Cross-company theme

Find where peers agree, diverge, disclose more, or avoid the same question.

Does it hold up?

Claim vs. data

Test a management explanation against filings, market data, or independent economic series.

Did they deliver?

Guidance scorecard

Compare the spoken target, formal guidance, and the actual result when it arrives.

What stayed unanswered?

Q&A pressure

Turn analyst pressure, partial answers, and refused numbers into a watch list.

What matters now?

A useful latest-call note prioritizes change and consequence. It does not retell the prepared remarks in order.
Latest-call research
What matters most from Micron's latest covered earnings call? Confirm the exact call date and fiscal period first. Give me the 3–5 developments most likely to matter over the next two quarters, followed by the strongest challenge to that view. For every finding, include the exact earnings-call quote, speaker and role, prepared remarks or Q&A, call date, fiscal period, and citation. End with the numbers or unanswered questions I should watch next quarter, and note any coverage gaps.

For a stronger note, ask which claims are measurable, which answers were partial or refused, and what evidence would disprove the headline interpretation.

What changed?

Hold the question constant across periods. That makes a shift in management's position easier to distinguish from a change in wording or analyst focus.
  1. 01

    Pick one claim

    Choose a concrete theme such as trade-down, pricing, hiring, capex, inventory, or demand.
  2. 02

    Set the window

    Check live coverage and use exact covered fiscal periods and call dates.
  3. 03

    Build the timeline

    Show the conclusion first, then the strongest quote from each relevant period and any conflicting evidence.
Change-over-time research
Has Walmart's explanation of consumer trade-down changed across its covered earnings calls from 2024 through 2025? Give me the answer first. Then show a quarter-by-quarter timeline using exact management quotes, with speaker, role, call date, fiscal period, section, and citation. Separate what management volunteered from what analysts pressed them on. Call out genuine changes, contradictions, and missing periods. Do not treat a search miss as proof that management was silent.

Who is different?

Peer research should reveal disagreement, disclosure asymmetry, and different exposure—not reward the company with the most search results.
Peer-comparison research
Where do Walmart, Target, and Costco agree—and disagree—about the health of the US consumer in their latest four covered calls? Check coverage first and compare equivalent call dates. Use the same themes for each company: trade-down, discretionary demand, traffic, private label, and credit pressure. Start with the two or three most important differences. Support each with exact quotes, speaker and role, company, call date, fiscal period, section, and citation. Explain unequal coverage and do not use the number of search results as a measure of emphasis.

For a larger cross-sector project

Name the universe yourself—for example Microsoft (information technology), Walmart (retail trade), and CBRE (real estate/rental/leasing) from 2019–2025—then work one company and a manageable date window at a time. Preserve the evidence before moving to the next batch, and align comparisons by call date because the same fiscal-quarter label can cover different calendar windows.

Does the story hold up?

The most valuable analysis often begins after the quote is found: translate a management claim into something observable, then look for evidence the company does not control.
Claim-versus-data research
Test Walmart management's consumer trade-down story against independent data from 2024 through 2025. First summarize the claim using exact earnings-call quotes. Then compare it with Census retail sales, BLS inflation and real earnings, and Federal Reserve consumer-credit data. State whether each outside series supports, contradicts, or cannot test the claim. Keep dates, units, and sources visible. End with the strongest alternative explanation and the next data point that would change the conclusion.

Consumer demand

Retail sales, inflation, real earnings, and consumer credit

Input or freight costs

Producer prices, energy prices, freight, shipping, and port activity

Hiring or wage pressure

Payrolls, job openings, and employment-cost data

Industry demand

Orders, shipments, trade flows, and peer operating metrics

For numerical thresholds

Use threshold search to discover candidates, not to prove that every qualifying statement was examined.

Best-effort threshold search
Find candidate statements from WMT, TGT, and COST covered FY2024 calls that mention closing more than 50 stores. Treat this as a best-effort, non-exhaustive search. Check coverage, search each company and exact quarter separately, and verify the number and unit in the full returned quote. Return only verified candidates with exact quote, speaker and role, company, call date, fiscal period, section, citation, and the number as stated. List the periods searched, missing coverage, and any batch that reached the result limit.

Make the answer easy to trust

Every important finding should connect the conclusion to the words that support it and the boundary around the search.
  • Lead with the finding, not a list of calls searched.
  • Use only exact returned earnings-call wording as a quote.
  • For management claims, attach speaker and role, company, prepared remarks or Q&A, call date, fiscal period, and citation.
  • For pressure points, keep the returned quote labeled as a composite analyst ask and management response; attach the analyst firm when available, specific ask, response quality, any refused number, call date, fiscal period, and citation without inferring speaker names.
  • Keep analyst suggestions and management refusals separate from management claims.
  • Keep spoken remarks, formal guidance, filed actuals, and outside data labeled before comparing them.
  • End with missing coverage, result-limit hits, and what would change the conclusion.

See Evidence and citations for a reusable evidence-table format.

Know the limits that change the conclusion

The mechanics matter only when they affect what you can safely say. These are the three boundaries to remember.

Coverage

Check the live company and date inventory. A covered range does not guarantee every intermediate quarter is present.

Result limits

A search returns at most 50 rows. For quote-level work, narrow by company and exact fiscal period and preserve the quotations and citations from each batch.

Thresholds

Complete numerical-threshold screening is not available yet. Describe the result as a best-effort, bounded, non-exhaustive screen.
How FactIQ finds call evidence

FactIQ searches structured call intelligence: atomic, quote-anchored claims with speakers, dates, fiscal periods, and sections, plus Q&A pressure points with a composite analyst-ask/management-response excerpt. It does not provide bulk full-transcript downloads.

Retrieval is lexical, not embedding-based semantic search. Exact terms rank first, then partial-term matches; typo-tolerant trigram fallback is used only when full-text search returns nothing. If a result is thin, try the product, segment, or phrase the company itself uses and inspect every returned row.