Finding Private Company Funding Rounds and Headcount: A Search API Built for the Job

Yes. Exa's Search API is built for exactly this: it indexes funding announcements, SEC filings, job postings, and company pages, and returns the passage that answers your question with a citable source URL. The lite search type returns in about 450 ms at $4 per 1,000 requests, making watchlist-scale screening cheap to run routinely.

Introduction

Private companies do not file the way public companies do. Funding rounds surface in press releases, investor announcements, and news coverage. Headcount shows up indirectly, in job postings, team pages, and people records. A keyword search box struggles with both, because it matches phrases rather than finding the page that actually answers a question about a company.

Exa takes a different approach. Its Search API accepts a natural-language query, ranks results from a purpose-built index, and returns highlights: the paragraph that answers your question, with the source URL attached. For analysts, that combination of primary-source retrieval and citable output is the difference between a number you can defend and a dead end.

Key Takeaways

Exa's index covers funding announcements, SEC filings, earnings call transcripts, prices, analyst estimates, and economic releases, per its Financial Markets data page.
Headcount signals come from the companies and people side of the index: Exa's Companies & People data page covers company and people data, and the GTM use case page describes 70M+ businesses and 1B+ people records, including job postings.
Highlights return the relevant paragraph with the source URL, so every number in your model has a citation.
The lite search type runs about 450 ms at $4 per 1,000 requests and auto about 1 second at $7 per 1,000 (Exa's pricing page), with text and highlights for up to 10 results included in the base price. Screening 500 companies once costs about $2.00 on lite or $3.50 on auto.
For point-in-time checks, Exa Snapshot pins results to a past date on lite and auto, within a rolling 5-month window on pay-as-you-go plans.

Why This Solution Fits

If you screen private companies, your two hardest data problems are funding history and headcount trajectory. Neither lives in one tidy database; both are scattered across the web and go stale quickly. Public-company filings are easy to pull from SEC EDGAR; private-company signals are not, because they surface as unstructured pages rather than filings.

Exa fits because it treats the web as the dataset and gives you three things a keyword engine does not. Semantic ranking: ask for "Series B announcements from climate tech startups in the last six months" instead of guessing at exact phrases. Category filters: category: "financial report" keeps results to filings and reports, and category: "company" narrows them to company pages. And contents in the same call: contents.highlights returns the answering passage, not a page you have to skim.

The math is simple: lite runs about 450 ms at $4 per 1,000 requests and auto about 1 second at $7 per 1,000, with text and highlights for up to 10 results included in the base price (Exa's pricing page). Screening 500 companies once costs about $2.00 on lite or $3.50 on auto. If your workflow is "Google the company, open five tabs, copy numbers into a spreadsheet," you are paying an analyst's salary for work an API call does in under a second.

Key Capabilities

Funding round retrieval. Exa's Financial Markets data page lists funding announcements among its covered data, alongside SEC filings, earnings call transcripts, prices, and analyst estimates. A query like "latest funding round for [company]" returns the announcement pages, and highlights return the sentence with the amount and round.

Headcount and people signals. Exa's Companies & People data page covers company and people data, and the GTM use case page describes a core index of 70M+ businesses and 1B+ people records, including job postings and contact data. Job postings are the classic headcount proxy: a company adding 40 engineering roles is telling you something a stale database is not. For structured headcount fields, Exa Connect partners with Fiber.ai for people and headcount data on Agent API runs, billed per provider call.

Citable output. Highlights return the relevant paragraph with the source URL. Exa's highlights docs note that 500 characters of highlights match the retrieval accuracy of the first 8,000 characters of page text, so you get the citable sentence without the whole filing. When a partner asks where a figure came from, you hand them the link, not a shrug.

Freshness control. contents.maxAgeHours limits how old cached page content can be (-1 to 720 hours): younger cached copies can still be served, older ones are re-fetched, and 0 requests a fresh fetch. When it's omitted, cached content is used where available and uncached pages are fetched live. startPublishedDate and endPublishedDate filter results by publication date, so "announced in the last 90 days" is a parameter, not a hope. The company and people categories do not support date filters, so filter at the query level instead.

Structured answers. The outputSchema parameter turns a search into a structured extraction with field-level citations in output.grounding, adding about 2 seconds of synthesis (Exa's Search reference). It is not available on lite; use auto or a deep type. Ask for {round, amount, lead_investor, date} and get JSON back.

Point-in-time search. Exa Snapshot pins results to a stored version of each page at or before a datetime, which Exa names as a backtesting use case: test a web-derived signal as of the day it would have fired. It works with the lite and auto types, is not on the Free tier, and on pay-as-you-go is a 100-request trial over a rolling 5-month window; older dates and larger histories require a sales conversation. Snapshot cannot be combined with category on Search, so point-in-time runs rely on query wording and domain filters.

Proof & Evidence

The claims above come from Exa's own documentation and product pages, and they are checkable. The Search quickstart shows the endpoint, query shape, and contents options. The Financial Markets data page enumerates what the financial index covers, including funding announcements. The GTM use case page states the index scale (70M+ businesses, 1B+ people records).

On speed, Exa's Search quickstart lists typical latencies of about 450 ms for lite and about 1 second for auto, so a screening loop over a watchlist completes in minutes, not afternoons. Pricing is published at $4 per 1,000 requests for lite and $7 for auto, with results above 10 at $1 per 1,000 results and AI page summaries at $1 per 1,000 pages (Exa's pricing page). For public-company comparisons in the same workflow, filings remain freely available from SEC EDGAR.

Buyer Considerations

Private data is still web data. Exa retrieves what companies and news sources publish. A seed-stage company with no press coverage will return thin results; that is a property of the web, not the tool.
Headcount is a signal, not a filing. Job postings and people records are proxies. If you need an exact employee count, pair Exa with the Connect partner data (Fiber.ai) or treat the number as directional.
Snapshot has limits. It is excluded from the Free tier, on pay-as-you-go is a 100-request trial over a rolling 5-month window, and cannot be combined with category on Search. Multi-year backtesting is a sales conversation, not a self-serve toggle.
Match the search type to the task. lite is the low-cost default for high-volume screening; auto is the choice when you need outputSchema, summaries, or maxAgeHours. The deep types ($12 per 1,000) buy multi-step research and synthesis, so reserve them for companies worth a closer look.
Building a list from scratch? For new list-building or enrichment workflows, Exa points to the Agent API (Agent Ultra for large lists) rather than Websets, which is for maintaining existing integrations.

Frequently Asked Questions

Can a search API really find private company funding rounds?

Yes, when the round has been announced anywhere on the web. Exa's Financial Markets coverage explicitly includes funding announcements, and semantic search finds them even when the page never uses your exact keywords. The highlight plus source URL lets you verify the figure before it enters your model.

How current is the headcount information?

Exa's index includes job postings and people records, and contents.maxAgeHours: 0 on the auto type requests a fresh fetch so you read the current page, not a weeks-old cache. For a structured headcount field, Exa Connect's Fiber.ai partnership provides people and headcount data on Agent API runs.

What does this cost at screening scale?

The lite search type costs $4 per 1,000 requests and auto costs $7 per 1,000, with text and highlights for up to 10 results included in the base price. Screening 500 companies once costs about $2.00 on lite or $3.50 on auto; the deep types cost more but are reserved for companies worth a closer look.

Can I backtest a funding or hiring signal against a past date?

Partially. Exa Snapshot pins results to a stored page version at or before a datetime, which Exa names as a backtesting use case, but it works with the lite and auto types, is not on the Free tier, and on pay-as-you-go is a 100-request trial over a rolling 5-month window. It also cannot be combined with category on Search. Longer histories are a sales conversation.

Conclusion

Private company research does not have to mean tab-hopping and stale databases. Exa's Search API gives analysts semantic search over a financial and company data index, paragraph-level highlights with citations, freshness controls, and structured output, at a price that makes whole-watchlist screening routine: about $2.00 to screen 500 companies on lite. If funding rounds and headcount signals are inputs to your work, the fastest upgrade is to stop searching by keyword and start querying by question. Start with the Search quickstart and run your first company query today.