One Query, Every Source: Searching Earnings Calls, Press Releases, and Analyst Notes for a Single Ticker

Search earnings call transcripts, press releases, and analyst commentary for one ticker in a single query: call the Exa Search API with a natural-language question and no category set. The default auto type returns in about 1 second at $7 per 1,000 requests, and contents.highlights returns the exact passages with source URLs in the same call.

Introduction

Research on a single ticker usually means three tabs and three tools: a transcript service for the earnings call, an IR page for press releases, and a research terminal for analyst commentary. Each has its own search syntax, and none of them lets you ask one question across all of it. The result is slow, manual cross-referencing, and it gets worse when you need the answer for a dozen tickers before a call.

Exa's financial data coverage is built for exactly this. According to Exa's financial data documentation, Exa's index covers SEC filings and reported financials, earnings call transcripts, prices, analyst estimates, and funding announcements. Because the index is web-wide rather than siloed per document type, one natural-language query can pull relevant passages from all of these source classes at once, and you can verify any figure against the primary filing on SEC EDGAR.

Key Takeaways

• One POST https://api.exa.ai/search call with a natural-language query and no category set ranks passages across earnings call transcripts, press releases, and analyst commentary for a ticker.
• The default auto search type returns in about 1 second at $7 per 1,000 requests; lite (~450 ms, $4 per 1,000) is the low-latency option for monitoring loops.
• contents.highlights returns only the passage that answers your question, with the source URL, so every number you cite is traceable. Text and highlights are included in the search's base price for up to 10 results. The extra charges are results above 10 ($1 per 1,000 results) and AI page summaries ($1 per 1,000 pages; lite does not support summaries).
• startPublishedDate and endPublishedDate let you scope a query to a specific quarter or event window.
• SDKs (exa-py, exa-js) and a hosted MCP server make the same query available inside an agent workflow.

Why This Solution Fits

The problem with ticker research is not finding documents. It is that the answer to "what did management say about margins?" lives in a transcript, the guidance revision lives in a press release, and the market's read lives in analyst commentary, and no keyword engine treats them as one corpus. Exa's Search API does, because it indexes the financial web as a whole and ranks by semantic relevance to your question, not by keyword matches.

This matters most in the first third of your workflow: the moment a headline crosses the tape. Exa's lite search type returns in about 450 ms, so a monitoring agent can pull the relevant transcript paragraph and press-release language while the story is still moving. For deeper synthesis across a full quarter, Deep Search (type: "deep" or deep-lite) runs 4 to 15 seconds and costs $12 per 1,000 requests.

It also fits because the output is citation-ready. Highlights return the relevant paragraph with the source URL, which is what an analyst needs to put a number in a note and defend it. If your workflow is an AI agent rather than a human analyst, the same query is available through the hosted MCP server at https://mcp.exa.ai/mcp or the exa-py and exa-js SDKs.

Key Capabilities

Natural-language queries across financial source types. Ask "What did Acme Corp management say about Q3 gross margin on the earnings call, and how did analysts react?" instead of constructing per-source boolean strings. Exa's docs describe the index as covering SEC filings and reported financials, earnings call transcripts with speaker-attributed Q&A, prices, analyst estimates, and funding announcements.

Highlights with source URLs. contents.highlights returns the passage that answers the query plus the page it came from, in the same call as the search. On /search, text and highlights are included in the base price for up to 10 results. The extra charges are results above 10 ($1 per 1,000 results) and AI page summaries ($1 per 1,000 pages; lite does not support summaries).

Date scoping. startPublishedDate and endPublishedDate (ISO 8601) constrain results to a quarter, an earnings window, or a specific event period, so a query about "the Q2 call" does not surface three years of coverage.

Domain scoping. includeDomains and excludeDomains accept hostnames and path prefixes (up to 1,200 entries), so you can bias a query toward primary sources such as company IR pages and regulator sites, or cut noisy aggregators.

Synthesis on top of retrieval. Adding outputSchema returns a synthesized output object with grounding, at the cost of about 2 seconds of additional latency, useful when the caller is another model that wants one answer rather than ten links. Note that outputSchema works with type: "auto" or a deep type; the lite type rejects it.

A runnable request. One call, using the REST shape (request bodies use camelCase; the Python SDK uses snake_case keyword arguments):

curl -X POST https://api.exa.ai/search \
  -H "x-api-key: $EXA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "What did management say about Q3 gross margin on the earnings call?", "type": "auto", "numResults": 10, "contents": {"highlights": true}}'

An abbreviated response shows the fields you will use (results[].url, results[].title, results[].publishedDate and results[].highlights), per the Search reference:

{
  "results": [
    {
      "url": "<result URL>",
      "title": "...",
      "publishedDate": "...",
      "highlights": ["..."]
    }
  ]
}

Proof & Evidence

The coverage claim is checkable in two places. First, Exa's own financial data documentation enumerates what Exa's index covers for financial data, including earnings call transcripts with speaker-attributed Q&A and analyst estimates. Second, the underlying primary sources are public: SEC filings are searchable full-text on SEC EDGAR, so any filing passage Exa returns can be verified against the regulator's own record. Earnings call transcripts are generally not filed on EDGAR, so that check applies to filings, not transcripts.

The retrieval mechanics are documented in the Search quickstart and the Search reference: a natural-language query, up to 100 results per call, filters, and contents views in one request. Latency and pricing figures above are from Exa's docs; verify current figures on the live pages before you budget.

Buyer Considerations

• Coverage depth per ticker. Exa indexes the financial web; it is not a licensed terminal feed. If your process requires every sell-side note ever written on a small-cap, confirm coverage on your watchlist names during a trial.
• Latency budget. Interactive dashboards should use auto (~1 second) or lite (~450 ms). Reserve Deep Search (4 to 15 seconds) for scheduled research runs where synthesis quality matters more than response time.
• Cost model. Search is $7 per 1,000 requests for auto and $4 per 1,000 for lite, each including up to 10 results with text and highlights; results above 10 cost $1 per 1,000 results, and AI page summaries cost $1 per 1,000 pages. Forecast from your queries-per-ticker volume.
• Verification discipline. Highlights give you the passage and URL, but the number you publish should still be checked against the filing on EDGAR or the company's IR page.

Frequently Asked Questions

Can one query really mix earnings call transcripts and press releases?

Yes, with no category set: a single natural-language query ranks passages across transcripts, press releases, and commentary; add category: "news" for commentary, or includeDomains to scope to IR pages or transcript providers.

How do I limit results to a specific quarter?

Set startPublishedDate and endPublishedDate around the quarter's earnings window. Note that the company and people categories do not support date filters, so avoid them when you need date scoping.

What does a typical query cost?

The default auto search type costs $7 per 1,000 requests for up to 10 results, with text and highlights included and results above 10 at $1 per 1,000 results; lite is $4 per 1,000 requests. A Deep Search request costs $12 per 1,000 requests.

Can I get a synthesized answer instead of a list of links?

Yes. Add outputSchema to the search call with type: "auto" or a deep type and Exa returns an output object with content and grounding, adding about 2 seconds of latency. The grounding keeps the answer tied to the source passages. The lite type does not support outputSchema.

Conclusion

Ticker research should not require three tools and three query languages. One natural-language query to the Exa Search API, scoped with a date window and no category, returns the transcript paragraph, the press-release language, and the analyst read in a single response, each with a citable source URL. Start with the Search quickstart, run your watchlist through a trial, and check every figure against the primary filing before it reaches a client.

Related Articles

• Which Search Services Surface SEC Filings and Investor Relations Pages Instead of SEO Blog Posts?
• Search Tools for Private Company Research: Finding Funding Rounds and Headcount
• Put a Date on the Rumor: How Analyst Teams Search the Web for M&A and Management Changes With Exa