Signal-Based Prospecting for Startups, Without a Huge Stack

Pete Kazanjy wrote to founders on 28 October 2026: "Signal-based outbound has three traits: Event-driven (triggered by an action, not a cadence). Account-specific (orgs showing real interest). Evidence-based (proof, not hope). If you're reaching out because you have evidence, it is NOT cold outbound."

Pramod Kumar, co-founder of Kingdom of Kumar. On X

·

A fortnight later, a widely saved post from thatGTMdude began, "I'm begging you:" and then laid out a weekly method: watch comments and profile views, hiring posts, funding announcements, put them in one sheet, cap it at 20 leads a day, open with the signal you saw. He said he would post reply rates later. I will not treat that promise as a result.

On Reddit, founders ask the same thing in plainer language: "How are you identifying buyer intent before outreach?" and how they are handling "timing" without buying a crowded intent platform.

This is a way to do that work with a sheet and a habit, not a six-tool stack. A catalogue of signal types is in Types of buying signals. A 30-day build is in Your first outbound motion in 30 days. The wider practice is on GTM engineering and GTM engineer.

The three traits, in working language

Event-driven means something happened. A person commented. A company announced a hire. A funder published a round. A user changed jobs. You are not writing because it is Tuesday and the sequence said so.

Account-specific means this organisation, not a persona bought in a list. You can point at the company and say why them, this week.

Evidence-based means you can show the artefact: a URL, a screenshot, a post, a job ad. If you cannot, you do not have a signal. You have a hunch. Hunches can start a research step. They should not start a first line that pretends you saw something.

Kazanjy is right that this is not really "cold" in the old sense. It is still outreach to someone who did not ask for your email. Be respectful, be accurate, and make it easy to say no.

"I'm begging you" — a method you can run this week

I would steal the shape of that post, and I would keep the cap.

  1. 1Pick three competitors or three voices your buyers already follow, not twenty.
  2. 2Each weekday, note people who commented on their latest public posts, or who viewed your profile, if you can see that honestly.
  3. 3Add companies that are hiring for a role your product replaces, or that just announced funding in your niche, if that is genuinely a reason to write.
  4. 4Put each row in one sheet: name, company, signal, date, link, fit (yes/no), email once verified, status.
  5. 5Cap the daily list. Twenty is a reasonable starting cap if you are writing the first lines yourself. Ten is better if you also have a product to ship.
  6. 6Open with the signal. "Saw your comment on X about Y" is only allowed if you did. Save the pitch for later, or for a second paragraph at most.
  7. 7One follow-up if there is no reply, with something useful. Then stop.

This is founder-led outbound with better timing. It does not need Clay on day one. It needs honesty and a calendar block.

"How are you identifying buyer intent before outreach?"

Intent, here, is not a vendor category. It is evidence that the problem is on someone's mind, or that the company is in a moment when a tool like yours gets bought.

Sources a small team can actually watch:

  • Public comments and posts by people who match your buyer
  • Profile views, used carefully and quickly
  • Hiring posts for the problem you solve, or for a role you replace
  • Funding and product-launch announcements in a defined niche
  • Job changes of people who already used a product like yours
  • Inbound you already have: form fills, waitlist, replies to your own content

What I would not start with: a broad intent-data contract, a scrape of everyone who ever visited a comparison page, or a model that invents "I saw you expanded to Europe." If the evidence is not in the sheet, it does not go in the email.

If you are pre-product, this system is premature. Take the App Launch Score and talk to users first.

A one-sheet template

Columns I would actually use:

  • Date seen
  • Company
  • Person and title
  • Signal type (comment, view, hire, funding, job change, inbound)
  • Link to the evidence
  • Fit (yes / no / unclear) against five written rules
  • Verified email
  • First-line draft
  • Approved (yes/no)
  • Sent date
  • Reply (none / not now / conversation)
  • Notes

If a column is not used in the weekly review, delete it. Sheets die of unused columns.

What to measure, and what to ignore

Measure conversations and meetings by signal type. You want to learn whether hiring posts are worth more than comments, or the other way around, *for your offer*. Ignore vanities: rows added, emails sent, "personalisation score."

Give a signal two honest weeks before you kill it, unless the first lines are embarrassing or the list is obviously wrong. Then cut.

Mistakes that turn "signals" back into spam

The most common: collecting a signal and then writing a generic first line anyway. If the email could have been sent to anyone in the same job title, you wasted the event.

The second: stretching the evidence. A company hired a recruiter; you write as if they hired a VP of Sales to buy your tool. A person liked a post; you write as if they asked for a demo. Over-claiming is worse than no signal.

The third: watching so many sources that you never write. Three sources, one cap, one block in the calendar. If the block does not exist, the sheet will not save you.

The fourth: sending from bad infrastructure. A careful first line on a domain that bounces is still a failed motion. Get cold email infrastructure in place before you raise the cap.

When to graduate off the sheet

When you cannot keep the cap without dropping evidence, or when the same signal is clearly repeating and you are the bottleneck on collection. That is the moment for a workflow: a watcher, enrichment, a draft queue, an approval pass. We build and run that kind of system at Kingdom of Kumar on Grok and Claude for $500 a month, with a three-month minimum. You still approve what goes out. We will not "scale" a signal you have not already seen work by hand.

FAQ

Is this the same as intent data?

Sometimes the same events show up in paid tools. The difference is that you can see and cite the evidence. If a tool cannot show you why this account is on the list today, it is not a signal system. It is a list with a story.

Can I let a model watch the sources for me?

Yes, as a collector. No, as a narrator. If it proposes a first line, you still check the link.

How is this different from a bought list plus a mail merge?

The list is not the starting point. The event is. The list is what you build after the event, for accounts that also fit.

Do I need a GTM engineer to do this?

Not for the sheet version. You need one, or a done-for-you team, when the collection and drafting have become a system you can no longer run in a daily cap.

Contact

hello@work.kingdomofkumar.com