Public Signal Prospect Research: How to Investigate a Dated Business Change From the Public Record, Grade Each Source by Who Made It, and Turn an Event Into an Account Hypothesis Without Calling It Intent

Public signal prospect research is the source-backed investigation of a dated business change at a named company, using records anyone can open. It establishes that an event happened, grades the record by who made it and what it cost them to be wrong, and turns the event into a testable account hypothesis. It does not establish purchase intent: the record proves the change, the hypothesis is yours to verify, and the decision belongs to the account.

Disclosure, date and method

Bob Generale is President of Percepture, which is related to Lead Seeker, the publisher of this page; Percepture's intent-data service is linked once below and labelled as related. This guide was researched on October 1, 2026 from the primary sources listed at the end: the SEC's Form 8-K instructions and its investor glossary, the SEC's EDGAR full-text search (which we queried ourselves for the filing counts in the data section), the US Department of Labor's summary of the WARN Act, GOV.UK's guidance on reporting company changes to Companies House, the Federal Acquisition Regulation's contract-reporting clause, the federal courts' PACER fee page, New York City's building-permit dataset, the Internet Archive's Save Page Now help page, and Article 14 of the GDPR. The Record Ladder, the Hypothesis Card, the Independence Test and the research-log template are this guide's own devices and are labelled as such. No customer data was used, no vendor was tested on live accounts, and no response or conversion rate is claimed for any method described here.

We also read the pages we could retrieve from the results for this query and its close variants. Of the exact-match results retrieved on October 1, 2026, the greater part belonged to a different profession: fundraising prospect research, which studies donors. The B2B pages that remain are step lists (company website, LinkedIn, news, a data platform) that cite no primary source, do not rank sources against each other, and do not say what a researcher may and may not conclude from a record. That gap is what this page is for.

Public Signal Prospect Research: The Short Answer

• Research a change, not a company. A company profile is a state; a public signal is a dated event. The event is the reason to look now, and it is the only thing the record can prove.
• Grade the record by its author. Who made it, under what duty, and what it costs them to be wrong matters more than how many times it was repeated. The Record Ladder below has six classes for exactly this.
• Write the hypothesis down, with the evidence that would kill it. An event becomes useful when you state what the account may now need to do, who would own it, when it expires, and what would prove you wrong.
• Count independent records, not mentions. A press release, the trade-press story that rewrote it and the aggregator row that scraped the story are one record, not three.
• Stop short of intent. Public research can establish an event and support a need. Whether anyone is buying is a question only the account can answer, and a message that pretends otherwise reads as surveillance.

What counts as public-signal research, and what it is not

Three definitions are mixed together under "prospect research", and the mixing is behind a great deal of the bad outreach that follows.

Fundraising prospect research is the older discipline. It studies individuals as potential donors, estimates capacity and affinity, and has its own professional association, Apra, which runs a fundamentals course under that exact name. Its methods overlap with this page in one respect only: both rely on records that are public and both have a duty to record where each fact came from. Everything else differs. This page is about companies, business changes and the people who hold roles affected by them, not about individuals' wealth.

Contact enrichment fills fields on a person: email, phone, title. It is a lookup, not an investigation, and its quality is a matter of verification and freshness. The prospect dossier template covers how to record it.

Public-signal research is neither. It starts from a dated change at a named organisation, found in a record anyone can open, and asks three questions in order:

1. What happened, according to whom, and when? That is the event. The record answers it, and the record's class limits how firmly.
2. What might the organisation now need to do, by when, and who would own it? That is the account hypothesis. The researcher writes it; the record only supports it.
3. Is anyone there deciding to buy something? That is intent. No public record answers it. A filing says a chief operating officer was appointed; it does not say she is replacing the warehouse system. A permit says a building is going up; it does not say who will fit it out.

The discipline lives in keeping the three apart. Research can move a claim from the first question to the second with evidence. Only the account can move it to the third, and it does so in a conversation, not in a database.

The Record Ladder: six classes of public record, ranked by what it costs the author to be wrong

Public records differ in one way that matters more than format, length or recency: who made the record, and what happens to them if it is false. A statutory filing is signed under penalty; a logo on a vendor's homepage costs nobody anything if the customer churned last year. The Record Ladder sorts every public source into six classes on that axis. It grades the source, not your confidence in a particular fact; the dossier template's confidence rungs do the second job, and the two are meant to be used together.

Three rules follow from the table.

A claim inherits the class of its weakest necessary link. If your account hypothesis depends on a vendor's logo wall, it is a class 3 hypothesis however many class 4 articles repeat the logo.

Classes 1 and 2 may be quoted to the account; classes 3 to 6 may only be asked about. "Your 8-K of September 12 reports a new chief operating officer" is a fact the account filed. "A case study on an integrator's site says you run their platform" is a claim someone else made, and the honest form is a question.

Climbing the ladder is the work. Research is the act of taking a class 5 detection or a class 6 rumour and finding, or failing to find, the class 1 or 2 record underneath it. When no such record exists, the finding is "unconfirmed", not "probably true".

Which business changes must be recorded, and how quickly

Class 1 is the only class with a clock written into law, and knowing the clock tells you how fresh an obligated record can be. The table lists the duties this guide could verify from the issuing authority's own pages; it is a calendar of what has to be filed, not a list of everything that is.

Two points about the calendar change how you read a record.

First, the deadline dates the record, not the change. An 8-K filed on a Thursday reports an event that happened up to four business days earlier, and the agreement it describes may have been negotiated for months. A WARN notice is, by design, at least 60 days ahead of the layoff it describes. The permit precedes the building. Each record sits at a known position relative to its event, and the position is part of what the record tells you.

Second, the obligated record is deep but narrow. It covers listed companies, large employers, regulated activities and public money. A private, 300-person software company that replaces its finance system leaves no class 1 record of it anywhere. For that company, class 2 is the top of the ladder, and the research has to be built from its own pages and the people it hires.

How large the obligated record is: an original count

To put a size on the US corporate record, we ran the SEC's EDGAR full-text search ourselves on October 2, 2026 (UTC) for 8-K documents filed between October 1, 2025 and October 1, 2026 that contain each item's exact caption number. The search tool reports counts of documents, not events or companies: an amendment, an exhibit or a repeated reference counts again, and the two largest items were counted in thirteen contiguous date windows and summed because the tool caps any single result at 10,000. The method is reproducible from the public search page with the same phrase, form and date filters.

The same search for role titles inside 8-K documents over the same year returned 6,150 documents containing "Chief Operating Officer", 673 containing "Chief Technology Officer", 373 containing "Chief Marketing Officer", 319 containing "Chief Revenue Officer" and 284 containing "Chief Information Officer". The number says nothing about how many of those appointments involve a purchase, which is the point: the obligated record is large enough to be a daily research feed for anyone selling to listed companies, and silent on everyone else.

From event to account hypothesis: the Hypothesis Card

The failure mode in signal work is not finding too few events; it is treating the event as the conclusion. The Hypothesis Card is a one-page discipline that forces the step in between. It is this guide's device, and every field exists because its absence has produced a bad email.

Three rules govern the card. It never contains the word "intent". Intent is the account's state of mind, and the card is a researcher's claim about a need. It never names a product on your price list. "May need to re-platform order management" is a hypothesis; "needs our WMS" is a pitch. It is retired, not forgotten. A card whose disconfirming evidence turned up is filed with that record attached, because the next researcher will find the same event in six months and should not repeat the work.

A filled card, in prose

A class 1 record: a state WARN page lists a notice from a manufacturer closing its older plant, effective in 70 days. Observed: the closure, the headcount, the effective date, the site. Implied: production is moving somewhere. Hypothesis: if the volume is consolidating into another site, that site may need to absorb new lines and new people within the quarter, which would involve operations, plant engineering and HR. Rival explanations: the volume may be leaving the company entirely to a contract manufacturer; the closure may be the end of a product line rather than a move. Disconfirming evidence: a press release or earnings remark describing an exit from the product category. Verify-next: read the company's newsroom and any recent class 1 filing for the words "consolidate", "transfer" or "discontinue". Roles implied: VP Operations, plant manager at the receiving site, HR director. Independence count: one, until the newsroom or a permit at another site is found. Expiry: the notice's effective date plus a quarter.

Nothing on that card tells a seller of anything to send an email. It tells them what to read next, and what they would be allowed to say if the reading confirms it.

Corroboration: when do two records count as two?

The reason to count records is that one record is a claim and two independent records are a pattern. The reason counts go wrong is that the public record copies itself. A company issues a release (class 2). A wire distributes it. A trade title rewrites it (class 4 in form, class 2 in substance). An aggregator scrapes the trade title (class 5). A database vendor ingests the aggregator. A researcher who finds the last four thinks they have found four records. They have found one, four times.

The Independence Test, this guide's second device, has three conditions, and a pair of records must pass all three to count as two:

1. Different origin. The records were made by different authors, and preferably in different classes. Two trade stories by the same freelancer are one origin.
2. No derivation. Neither record cites, quotes, links to or paraphrases the other, and neither derives from a third record that the other also derives from.
3. Separate causation. Each record would have been made even if the other had not. A permit and a job posting pass; a release and its coverage do not.

Applying the test means tracing upstream: for every class 4 or class 5 record, find the earliest dated version of the claim and treat the whole cluster as that one record. The exercise has a second benefit, which is that it exposes how old the claim is. A database row refreshed last week may trace to a release from two years ago.

Independent records are strongest when they come from different sides of the event. The permit (class 1) and the posting for a plant engineer (class 2) and the integrator's announcement (class 3) describe the same change from the regulator's side, the employer's side and the supplier's side. A researcher who holds three such records holds a hypothesis worth a call. A researcher who holds three rewrites of one release holds a rumour with good typography.

Dating the change: the four dates every record carries

In our experience, public research goes wrong on dates at least as much as on facts, because a record has several and the useful one is not always the one shown largest. The sibling guides on reading a job posting as a buying signal and verifying an executive hire before outreach treat the clocks of those two signals in depth; here is the general rule.

• Event date: when the change happened, as stated inside the record. The appointment took effect; the first sale closed; the notice was served.
• Record date: when the record was filed or published, which the calendar above shows can be days (8-K), weeks (Form D) or, for a WARN notice, two months ahead of the event.
• Retrieval date: when you read it. Write it down. Class 2 pages change, and your retrieval date is the only evidence of what the page said when you relied on it. The Internet Archive's Save Page Now help page describes the simplest way to fix a copy in time: put a URL into the form, press the button, and you have a permanent URL for that version of the page.
• Expiry date: when the hypothesis built on the record goes stale. It belongs to the card, not the record, and it comes from the event's own mechanism: the posting is filled, the building opens, the integration goes live, the appointee's first quarter ends.

A record with only one of the four dates attached to it in your log is a record you will misread later.

From account to people, without collecting what you do not need

The card names functions. Turning a function into a current person is a separate step, and it is where public research can drift into something the person on the other end would not recognise as research.

Two rules keep it in bounds. Verify the role from classes 1 and 2. A leadership page, a filing, a conference agenda or the person's own public professional profile can confirm who holds a role today; a vendor database can suggest it. The prospect dossier template sets an as-of date on that field for exactly this reason. Record the role, not the biography. The research log needs the function, the source that confirms the holder and the date you confirmed it. It does not need the person's history, family, or anything you learned about them that is unrelated to the role.

Where European data-protection law applies, Article 14 of the GDPR addresses the situation directly: where personal data have not been obtained from the data subject, the controller must provide that person with specified information, including the controller's identity and the purposes of the processing, within a reasonable period and at the latest within one month, or, if the data are to be used for communication with that person, at the latest at the time of the first communication, subject to the exceptions in Article 14(5). That is a restatement of the article's text, not legal advice; the practical consequence is that the first message to a person you researched from public sources has to carry, or link to, that information. The guide to B2B data-provider compliance under GDPR and CCPA covers the surrounding obligations.

What the public record cannot tell you

A method is only as honest as its stated limits. These are the ones that have cost us the most time.

• Private companies are largely silent at class 1. For a private account outside the duties in the calendar above, there may be no filing to find. Research starts at class 2 and the independence count is harder to reach.
• The record reports what was filed, not what is happening. An 8-K says an agreement was signed; implementation may have started a year before or may never start. A permit may lapse. A posting may be evergreen.
• Entity resolution is on you. Subsidiaries, holding companies, namesakes and recently renamed firms produce records that look like your account's and are not. Match on registration numbers, addresses and officers, not on names.
• Class 2 pages are revised without notice. A careers page that listed a role last month may show no trace of it today. Without an archived copy, you cannot show it was there.
• Derived data carries its own clock. A detection date on a class 5 row is when the crawler looked, not when the change happened. The sibling guide on what technographic data can and cannot tell you about an account explains the gap for one family of data.
• Survivorship and selection shape what is visible. Vendors publish the case studies that worked; journalists cover the changes that are newsworthy. The public record over-represents success and scale.
• Nothing in any class states a decision to buy. This is not a limitation to work around; it is the boundary of the method.

The research log: a template that survives an audit

The log is the artefact. A finding with no log is an opinion, and a log with these columns can be handed to a colleague, a manager or a counsel and read without the researcher in the room. One row per record; one card per hypothesis; the card cites rows.

Pre-call checklist for a public-signal hypothesis

1. The event is stated in one sentence that quotes or tightly paraphrases a class 1 or class 2 record, with its date.
2. The record was opened at its source, and a copy was archived with a retrieval date.
3. Observed and implied are written on separate lines.
4. The hypothesis names a capability or decision and a window, not a product.
5. At least two rival explanations are written, and at least one was checked.
6. The disconfirming evidence was named before the search and was looked for.
7. The independence count is two or more under all three conditions of the Independence Test, or the card is marked as a question.
8. Roles are listed as functions; each named holder was verified from a class 1 or 2 source with an as-of date.
9. Every date on the card is one of the four kinds, labelled.
10. The expiry date is set from the event's own clock, and the card has a retirement rule.
11. Nothing on the card or in the log would surprise the person it concerns if they read it.
12. The first message states or links to who you are and why you are writing, and quotes only class 1 or 2 records.

A synthetic example: one account, six records, one hypothesis

Calder Ridge Foods is invented for this example: a private frozen-foods manufacturer with about 900 employees, two plants and a distribution centre, which sells into grocery. The seller is a vendor of labour-management software for warehouses. Six records were found in a week of research.

The card reads: If records 1 to 4 are as stated, Calder Ridge may be consolidating distribution into a new automated site by spring, and may need to decide how it staffs, measures and schedules that site before go-live; the decision would sit with operations and the incoming director of automation, with IT and finance involved. Rival explanations: the automation integrator may have been chosen already and may bundle labour management; the spring date may slip, as construction dates do. Disconfirming evidence: a newsroom item or a permit amendment naming a turnkey integrator with a labour module. Verify-next: read the permit record for the named general contractor and any systems integrator; check whether the director posting has been filled and by whom. Independence count: three origins (A, B, C) across two classes, so the card is a hypothesis, not a question. Expiry: the stated spring opening, or the posting's fill date if earlier.

What the seller may say, if the card survives verify-next: that the company has announced a new automated centre for the spring and is hiring to run it, both in its own words, and that the seller has a question about how the site will be staffed in its first quarter. What the seller may not say: that Calder Ridge is "in the market" for anything.

Where Lead Seeker fits

Lead Seeker is built for the part of this method that does not scale by hand: watching the public record for the dated changes that match an ICP. Its Trigger Signals draw on public sources only — company sites, job boards, regulatory filings, press releases, podcast and conference transcripts, news and earnings transcripts — and each signal links back to its original source, which is the condition for grading it on the Record Ladder at all. When a search is run, each account arrives as a Prospect Dossier with a verified contact, the public signal that surfaced the person, a short personality read and a suggested first line, with every claim linked to its public source where available and the freshness date stamped on the record. You can see how Lead Seeker works end-to-end, read how the prospect dossier works, or claim 5 free verified leads to grade a live record against the ladder yourself.

The card, the independence count and the retirement rule remain the researcher's: Lead Seeker surfaces, scores and links the record, and the hypothesis built on it is still yours to write and to retire. Teams that want the research run as a managed programme rather than a tool can brief Percepture's B2B intent data service; Percepture is related to Lead Seeker and the author of this page is its President, so read that recommendation as a disclosure, not an independent referral. For the wider library on evidence, timing and signal reading, start with the intent data insights hub and the signal based prospecting workflow, which picks up where this page ends: at the verified person and the first message.

Frequently Asked Questions

What is public signal prospect research?

It is the investigation of a dated business change at a named organisation using records that anyone can open: filings, notices, permits, the organisation's own pages, counterparty announcements, independent reporting and derived datasets. The output is a graded account hypothesis with its evidence, its rival explanations, its verification step and its expiry date. It establishes events and supports needs; it does not establish purchase intent.

Is public signal prospect research the same as prospect research in fundraising?

No. Fundraising prospect research studies individuals as potential donors and has its own profession and association. Public-signal research in B2B sales studies organisations and the business changes recorded about them, and only then verifies who holds the roles those changes affect. The two share a reliance on public records and a duty to cite them, and little else.

How is public-signal research different from intent data?

Intent data reports or infers research behaviour, at an account level, from sources you cannot open yourself. Public-signal research works from records you can open, quote and archive, and it grades them by who made them. The companion guide on buying signals versus intent data sets out which questions each kind of source can answer; this page covers how to investigate the public kind.

How many sources do you need before contacting an account?

Two independent records under the Independence Test, from different origins, with neither derived from the other, is the threshold at which a card becomes a hypothesis rather than a question. Records that quote or rewrite each other count once. A single class 1 record can justify a research step; it should not, on its own, justify a claim about what the account needs.

Can a public record prove that a company is buying?

No. A filing, notice, permit or posting proves that the recorded event occurred as stated by its author. It can support a hypothesis that the organisation may need to decide something, and it can tell you which functions would own that decision. Whether anyone is evaluating a purchase is known only inside the account, and the honest way to learn it is to ask.

How long does a public signal stay useful?

Until the mechanism behind it runs out, which is set per event rather than by a general rule: a posting until it is filled, a notice until its effective date, a permit until the building opens, an appointment until the appointee's early priorities are set. Write the expiry on the card when you write the hypothesis, and retire the card when a fresh record contradicts it or the date passes without one.

Is it legal to research people from public sources before outreach?

The answer depends on jurisdiction and purpose, and this page is not legal advice. Where the GDPR applies, Article 14 requires a controller who obtained personal data from a source other than the person to give that person specified information, at the latest within one month or at the first communication, subject to the exceptions in Article 14(5). The practical consequence for researchers is to record roles rather than biographies, to verify from the organisation's own records, and to make the first message say who you are and why you are writing.

Which public records are free to read?

EDGAR, state WARN pages, the UK Companies House register, FPDS and municipal open-data permit sets are free at the source. PACER charges $0.10 per page and bills quarterly only users who accrue more than $30 in the previous quarter. Archived copies of pages can be made for free through the Internet Archive's Save Page Now. Derived datasets from commercial vendors are not free; the record underneath them, when it exists, sits at the source rather than behind their paywall.

Sources

• U.S. Securities and Exchange Commission, Investor.gov glossary, Form 8-K: "the report that companies must file with the SEC to announce major events that shareholders should know about"; "Companies generally have four business days to file".
• U.S. Securities and Exchange Commission, Form 8-K, General Instructions and Items: "a report is to be filed or furnished within four business days after occurrence of the event"; item captions quoted in the tables.
• U.S. Securities and Exchange Commission, EDGAR Full Text Search, and Search and Access: "Find keywords and phrases in more than 20 years of EDGAR filings". Counts retrieved October 2, 2026 (UTC), for 8-K documents dated October 1, 2025 to October 1, 2026.
• U.S. Department of Labor, Plant Closings and Layoffs: employers with 100 or more employees must provide at least 60 calendar days' advance written notice of a plant closing and mass layoff affecting 50 or more employees at a single site of employment.
• GOV.UK, Running a limited company: company changes you must report: "You must tell Companies House within 14 days" of changes to directors, their details or company secretaries; "within a month if you issue more shares".
• Acquisition.gov, FAR 4.604 Responsibilities: the contract action report "must then be completed in FPDS within three business days after contract award".
• Administrative Office of the U.S. Courts, PACER Pricing: How Fees Work: "$0.10 per page"; "Users are billed on a quarterly basis if they accrue more than $30 in the previous quarter".
• NYC Open Data, DOB Permit Issuance: "The Department of Buildings (DOB) issues permits for construction and demolition activities in the City of New York."
• Internet Archive, Save Pages in the Wayback Machine: Save Page Now.
• GDPR, Article 14 (gdpr-info.eu text; official text on EUR-Lex): information to be provided where personal data have not been obtained from the data subject.
• Apra, association home page: lists "Apra Fundamentals: Prospect Research" among its programmes, cited only to distinguish fundraising prospect research from the subject of this page.
• Lead Seeker, Trigger Signals, Prospect Dossier and How it works pages, for the product statements restated above.

About the Author

Bob Generale is President of Percepture. He works across SEO, AI search, digital PR, sales intelligence and AI-powered revenue systems, with a focus on connecting visibility, buyer intent and sales action.

Disclosure: Lead Seeker is related to Percepture and Pyra. The Percepture link on this page is labelled as related, and no data vendor or sales intelligence platform named here was tested or engaged in the course of writing it.

Next Steps

Take the last five accounts your team contacted on the strength of a public signal and rebuild each one as a Hypothesis Card from the log you have. Where the independence count is one, or the card cannot be written without a product name, you have found the pattern behind the silence. Then claim 5 free verified leads to see what a source-linked record for one of those accounts looks like, or see how Lead Seeker works end-to-end before you decide what the feed should watch.