Signal Check

What this is

A five-step check that turns one real, dated event into the reason you reach out this week.

• Writes one sentence about the signal, backed by a public page you can point to
• Builds the company list from the signal source instead of an industry filter
• Kills any row you cannot verify yourself in three minutes

The problem

• → 217 emails sent. Every one opens with a merge tag, the auto-filled "[First Name]" that does not make it personal.
• → Four opens. Zero replies.
• → The list came from an industry filter, not a dated event.
• → Line one proves nothing, so nobody reads line two.

Cost: divide what the send cost you by the replies it earned.

How it works

Not software. Nothing to install. Five steps you run once per segment and reuse for every list built against it.

• Step 1 names one public, dated, causal event in a single sentence.
• Step 2 starts at the source where the event lives (job board, funding page, licence board), never a bought list.
• Step 3 is the only step that cannot be handed to a tool. You open the page yourself.
• Step 4 keeps one signal on its own list so a reply traces back to the event that earned it.
• Step 5 is a two minute gate that catches a stale date or a first line that never mentions the signal.

What you need first

Get the files

Everything for this build sits in one folder on GitHub. GitHub is just a website that holds files.

Direct folder link: https://github.com/gary-chakraborty/gap-build-vault/tree/main/builds/signal-prospecting-system

To get them onto your own computer without any software: open the link, click the green Code button near the top right, then click Download ZIP. That downloads the whole vault. Unzip it, then open the folder builds/signal-prospecting-system. You can also click any single file on GitHub and read it right there in the browser.

Build it

Phase 1: Pick one signal and write the sentence

1. Open signal-picker.md. Read the six rows in the table.
2. Pick the one category you can already find a source for today. Stop at one.
3. Open five-step-system.md and scroll to Step 1. Copy the sentence shape into a blank note.
4. Fill it in for your own market: company type, the dated public event, the plain reason it creates a need, and where you will look.
5. Run the trap check in that same section: if the sentence is true of every company in your market, it is a description, not a signal. Rewrite it.

You know this phase worked when: you can read one sentence out loud that names an event, a reason, and a public place to find it, and you can name the exact page type that proves it.

Phase 2: Set up the sheet and run the list prompt

1. Go to sheets.google.com and click Blank spreadsheet.
2. In row 1, type five headers across: Company, Signal, Date, Source, Best contact title.
3. Rename the sheet after your signal, not after the month. One sheet per signal.
4. Open a new tab, go to claude.ai, and turn web search on in the chat box before you send anything.
5. Paste this prompt, filling in the bracketed fields from your Phase 1 sentence:

I'm building a list of [NICHE / INDUSTRY] companies that match one specific signal.

Signal definition: [PASTE YOUR ONE-SENTENCE SIGNAL DEFINITION FROM PHASE 1]

Find [TARGET COUNT x 3] companies where this signal happened in the last
[90 days / 12 months]. For each one, give me:

1. Company name
2. What the signal is, in one sentence, specific to this company (not a copy of the
   general definition, the actual event, e.g. "hired a VP of Sales on [date]", not
   "recently hired someone")
3. The exact date the event happened
4. A direct link to the public page where you found it (job post, press release,
   filing, news article, the actual page, not a search results page)
5. The best contact title to reach at this company for this signal

Rules:
- If you cannot find a direct source link for a company, do not include that company.
  A row with no link is worse than a shorter list.
- Do not include a company because it "fits the industry" if you can't point to the
  specific dated event. General industry fit is not the signal.
- If two companies have the exact same signal wording, check both again. A signal that
  reads identically across companies is usually the industry, not the event. Rewrite the
  wording to be specific to each company, or drop the weaker one.
- Give me the list as a table: Company | Signal | Date | Source link | Best contact title

6. Paste the returned table into your sheet.

You know this phase worked when: your sheet holds three times your target number of rows, and every row has something in the Source column.

Phase 3: Verify the signals yourself

1. Start at row 2. Click the source link.
2. Confirm three things on the page itself: the event happened, it happened to that company, and the date sits inside your window.
3. If all three hold, paste the real page URL into the Source cell, replacing whatever was there.
4. If you cannot confirm it in about three minutes, delete the whole row. Do not soften the wording to keep it.
5. Repeat to the bottom of the sheet.
6. Count what survived. Fewer than a third surviving means your niche or your date window is too loose. Tighten one of them and rerun Phase 2.

You know this phase worked when: every remaining row has a URL you personally clicked, and you can say the date of each event without looking it up again.

Phase 4: Route one signal to one campaign

1. In your sending tool, create a new campaign, not a new variant inside an existing one.
2. Name it after the signal and the month, so a reply traces back to the event.
3. Upload only the verified rows from this one sheet. Nothing else goes in this campaign.
4. Write line one so it states the event, with the company's own detail in it. Write line two as what that event means for them.
5. Open send-gate-checklist.md and walk the five boxes. Every one has to be a real yes.
6. Send.

You know this phase worked when: your sending tool shows one campaign holding one signal type, and every lead in it has a source link sitting in your sheet.

Run it the first time

Made-up example, invented to show the shape. Do not treat any figure below as a result.

Input to the prompt:

Niche: B2B software companies. Signal: closed a seed or Series A round in the last 90 days. Target count: 8, so ask for 24.

What comes back:

Company              | Signal                          | Date       | Source
Fielder Labs         | Closed $6.4M seed               | 2026-05-14 | company press page
Northrail Systems    | Closed Series A, amount undisclosed | 2026-04-02 | trade news article
Quillbank            | "Recently raised"               | unknown    | no link given

Row 1 survives: the press page loads, the date matches, the round is real. Row 2 survives if the article names the company and the date. Row 3 dies on the spot. No date, no link, no row.

When it breaks

Tell me how it went

I am not asking for your email. There is no list, no sequence, nothing to unsubscribe from.

Two minutes, three things: how you found this, whether it was useful, and what you want me to build next. https://form.jotform.com/262194392563059

The next build comes from those answers.