Industry · Startups

Cold email agency for startups

GrowthPal runs outbound for startups that are ready for it, and says so plainly when a company is not ready yet. Best fit is past 3M in ARR, with an offer people already buy, a team ready to take sales calls, and roughly 30,000 reachable contacts. Below that bar the honest answer is usually not yet, and this page says what to do instead.

The buyer

Who you are really selling to

Role	What they care about	Where they pay attention	Best channel
Founder or CEO	Runway against the next milestone, whether outbound is the cheapest next channel or an expensive distraction, and whether anything built now survives the first sales hire.	Founder peer groups, private Slack and WhatsApp communities, investor updates from peers, podcasts, other founders' posts.	Cold email to a personal work address from a named human. They reply to something only readable about their own company, and ignore category pitches.
Co-founder who owns go-to-market	Whether the current pipeline is a repeatable sales motion or a run of warm introductions, and whether a sales person is about to be hired into a motion nobody has written down.	Go-to-market newsletters, sales communities, their own CRM, peer founders one stage ahead.	Email about the handover problem specifically, not about volume. They know what they have not documented.
First sales hire, founding account executive or head of sales	Whether they were hired to build a motion or inherit one, where their first list comes from, and whether the founder will keep closing over their head.	Sales leadership communities, their own quota, the founder's calendar.	Email framed as capacity underneath them. New in the seat, and most exposed if an agency-sourced meeting goes badly.
Chief of staff or head of operations	The board pipeline slide, whether the number is sourced pipeline or referred pipeline, and whether channel experiments have failure conditions written down.	Board reporting cycles, investor operating partner programs, reporting tooling.	Email about measurement, naming what would count as the experiment failing.
Operating partner at the investor	Not recommending a vendor that burns a portfolio company's only domain or its remaining runway. Gets asked before a contract is signed.	Portfolio Slack channels, vendor lists the fund maintains.	Rarely worth a cold email. Treat this person as a reference check that happens whether or not you plan for it.

Constraints

What makes outbound different here

The offer may still be moving

Outbound tests a message against a market. If the offer, the price and the target change every month, a negative result tells you nothing: you cannot separate a message failure from a market failure from an offer that moved mid-test. Offer stability is the precondition, not a nice-to-have.

What it decides: Whether anything gets built at all. The readiness check runs before a list exists. A company whose offer is still moving gets told to come back when it has held for a quarter, and gets the reason in writing rather than a shortened pilot.

Testing needs volume, and a startup's market is often too small for it

Separating a message that works from one that got lucky needs enough sends per variant to see a difference. Below roughly 30,000 reachable contacts the universe runs out before the learning does. You either burn the list on a weak hypothesis or re-contact the same people too soon.

What it decides: The method, and sometimes the answer. Under that count, named accounts worked manually by the founder beat a system, and the honest recommendation is no engagement. Above it, variant count and list segmentation are planned against the total universe before the first send.

Founder attention is the binding constraint, not budget

The scarce input is the founder's time: the offer brief, the objection list, notes from calls that went well, fast approval of angles, and a calendar with room for replies. No credible figure is published for how many hours this takes, and anyone quoting you one is guessing.

What it decides: The shape of the engagement. Input from the founder is front-loaded into the first weeks, the approval loop is sized to a founder's week rather than a marketing team's, and nothing waits on a committee that does not exist yet.

Replies land on whoever is free, and that is usually the founder

There is no SDR to triage. A reply sitting for two days reads as indifference to a buyer who moves fast themselves, and at this size the person who answers is often the person who was on a plane.

What it decides: Discovery call coverage is agreed before the first send, including who answers when the founder is traveling. The first ask is answerable without a meeting, so a slow reply costs less.

Founder-led sales hides the absence of a motion

A founder closes on conviction, context and improvisation, none of which transfer to anyone else. Pipeline built from a warm introduction network looks like a motion right up to the month someone else has to repeat it.

What it decides: What the engagement produces besides meetings. Every campaign leaves a written record of which angles were tested and retired against which segment, because that record is the sales playbook handover the first sales hire inherits.

Signals

Buying triggers, and the angle each one earns

Trigger	How we find it	The angle it earns	Proof the email needs
A fresh round or private placement	SEC EDGAR full-text search for Form D notices by US private issuers. The SEC's own guide confirms the notice is due within 15 days after the first sale, is filed free, and becomes publicly available on EDGAR. Outside the US: Companies House filings, Crunchbase, Dealroom, the company press page.	The round was raised against a plan, and the hiring plan usually runs a quarter or two ahead of the pipeline plan.	The dated filing or announcement, plus one concrete thing they said publicly they would do with the money.
A first sales seat posted	The careers page and the public applicant tracking board most startups expose per company. Greenhouse documentation states candidates are redirected to a per-company board to apply, and Lever, Ashby and Workable expose comparable public boards. LinkedIn Jobs read manually for posting dates.	The company is about to stop selling founder-first, and the new hire will be asked for pipeline before they have built a list.	The exact role title, whether it is the first seat of its kind, and the segment or quota named in the posting text.
That first sales seat open a long time, or reposted	The same public boards, comparing posting date against today, plus visible churn in the seat on LinkedIn.	The motion stalled rather than never started, and hiring has become the bottleneck instead of the fix.	How long the role has been live, stated plainly and without smugness.
An accelerator cohort ending	Public accelerator directories and demo day pages. The Y Combinator startup directory is public and carries batch labels, as do Techstars, Antler and most regional accelerator portfolio pages.	The cohort ends, the warm introduction network thins within a quarter, and the company needs a channel it owns outright.	The batch label, the date, and what they said they were building.
A price published for the first time	The pricing page appearing where a contact form used to be, compared against earlier versions in the Internet Archive. A launch post. Product Hunt.	Publishing a price is the moment an offer stops moving every month, which is the precondition for testing messages against it.	The page, the date it changed, and the tier that now exists.
A second market opened	Companies House incorporations in the UK, state business search records in the US, equivalent national registries elsewhere, plus a localized site or a job posting in the new market.	The list rules and the consent rules differ there, and the home-market sequence usually travels unchanged.	The dated filing or the localized page, with the market named correctly.
A bridge round, seed extension or venture debt	Form D amendments on EDGAR, which are generally required annually when an offering continues past twelve months or when certain information changes, plus round records labeled as extensions on Crunchbase and Dealroom.	The company is buying time to prove a repeatable sales motion, which is when channel discipline matters most and improvisation costs most.	The filing or record, handled without condescension. This trigger lands badly more often than any other, so it is used rarely and never as the opening line.

Peek inside

Example angles we would test

Examples of angles to race in this market, not results. Each one is a hypothesis with its proof assembled before a single send.

You raised, and the hiring plan is running ahead of the pipeline plan.

Proof it needs: The dated Form D notice or round record, plus the count of open sales seats on their own public board.
The ask: Who is building pipeline while the seats fill.

You are hiring your first sales person into a motion nobody has written down.

Proof it needs: The first-sales-seat posting, with its date and the segment named in the text.
The ask: What list that person starts from on day one.

That sales seat has been open long enough that the plan behind it has slipped.

Proof it needs: Posting date against today, plus repost history on the same public board.
The ask: What is covering pipeline in the meantime.

Your cohort ended and the warm introduction network is thinning.

Proof it needs: The batch label and demo day date from the public accelerator directory.
The ask: Which channel the company owns outright today.

You published a price for the first time, so the offer finally stopped moving.

Proof it needs: The current pricing page and its archived predecessor in the Internet Archive.
The ask: Whether anyone is testing messages against the new packaging.

The founder is still the only person who can close, and the calendar is the ceiling.

Proof it needs: A public statement by the founder, or the absence of any sales role at a company with visible customers.
The ask: What happens to pipeline in a travel-heavy month.

Outbound is going out from the one domain that also carries product and billing mail.

Proof it needs: Public DNS and DMARC records on the root domain, and any visible sign of volume sending from it.
The ask: Whether marketing mail is isolated from product mail.

The board asked where the pipeline number comes from, and the answer is introductions.

Proof it needs: The public record of the raise and the plan attached to it.
The ask: What share of pipeline is sourced rather than referred.

A previous outbound attempt produced nothing and nobody can say which part failed.

Proof it needs: Something the founder or sales lead published about their own sending.
The ask: Whether the failure traces to the list, the message or the infrastructure.

The list

How the list gets built in this market

Lists here come from the regulator, the hiring surface and the cohort directories, because a startup's readiness is more visible in a public filing than in any database field. A filter of software plus a headcount band is the list every competitor already bought.

Funding, entity and regulator data

SEC EDGAR full-text search for Form D notices and amendments by US private issuers. Free, dated, filed by the issuer itself.
The Form D revenue range item, where the issuer discloses it. It is the rare public, regulator-filed signal of whether a company is anywhere near a readiness bar. Many issuers decline to disclose, which is informative in its own way.
Companies House in the UK for incorporation dates, officers, registered office and confirmation statements. Equivalent national registries and US state business search records elsewhere.
Crunchbase, Dealroom and PitchBook for round records, amounts, dates and lead investors, including rounds labeled as extensions.

Cohort and program directories

The Y Combinator startup directory, public and filterable, with batch labels attached.
Techstars, Antler and regional accelerator portfolio and demo day pages.
Government-backed program lists and innovation grant registers where they are published.
Investor portfolio pages, which carry current holdings and often the stage and year.

Hiring surface

Company careers pages, which at this size often state the segment and the quota.
Public applicant tracking boards per company: Greenhouse, Lever, Ashby and Workable, all readable without an account.
LinkedIn Jobs for posting dates and repost history, read manually.
Posting age treated as its own signal. A first sales seat live for months says something a new posting does not.

Company surface and people data

The pricing page, compared against earlier versions in the Internet Archive, for the moment a price was first published.
The changelog, the docs and launch posts, plus Product Hunt, G2 and Capterra placement.
Public investor update posts, which founders increasingly publish openly.
LinkedIn for role and tenure verification only. LinkedIn's User Agreement prohibits scrapers, bots and automation extensions and warns that accounts using them may be restricted, so it is a place to read and verify, never a place to run a harvest.

How a row gets verified

Readiness check on the row, not just the company. Does the public record suggest a real offer being bought, or a company still in discovery. A pre-revenue issuer is a dead row for most startup-targeted offers and gets dropped rather than softened.
Entity check: does the company still exist, still sell the same thing, and still match the segment. Startups pivot faster than any database updates, and a filing from last year describes a company that may no longer be there.
Role check: is the person still in that role as of a dated source. At this size titles change in weeks, and a stale title is the most common reason a good angle lands wrong.
Trigger age check: the trigger has to be younger than the sales cycle. A round from two years ago is a fact about the company, not a reason to write this week.
Suppression and conflict check: existing customers, prior opt-outs, do-not-contact entries, and investor conflict suppression, so a portfolio company of your own investor is never contacted cold without anyone internally knowing.

What makes a bad list here

A filter of industry plus a headcount band with no trigger attached. Every competitor bought that list from the same vendor on the same day.
Pre-revenue or pre-launch companies, when the offer needs a budget holder. The row cannot buy, and the send costs reputation for nothing.
Role catch-alls: hello@, founders@, team@. At startup size these often do reach the founder, which sounds useful and is not: it reaches them in the inbox they triage fastest and trust least.
Rows whose only trigger is a funding round older than your sales cycle, which forces the angle back to a category pitch.
An accelerator batch used whole. A batch is a date, not a qualification, and the companies in it have nothing else in common.
Mixed-geography lists sent under one set of rules, which ignores that UK sole traders and ordinary partnerships are treated as individuals under PECR while limited companies are not.
Investor-conflicted rows contacted cold, which costs more goodwill than any meeting they produce is worth.

Rules

Compliance and trust constraints

Nothing here is legal advice. Rules differ by jurisdiction and change, so read each source directly and confirm your own position with counsel before sending. Three points matter more to a startup than to a larger sender.

CAN-SPAM needs a valid physical postal address, which a remote company often lacks

The FTC guide states the law covers commercial email including business to business, and requires accurate header and routing information, a non-deceptive subject line, disclosure that the message is an advertisement, a valid physical postal address, and a working opt-out honored within ten business days and live for at least thirty days after sending. A remote-first startup with no office has to answer the address requirement before the first send, usually with a registered agent address. Hiring an agency does not move liability off the promoted company.

Read the source

A list of small UK companies carries more consent-required rows than a list of large ones

The ICO separates corporate subscribers, meaning limited companies, LLPs, Scottish partnerships and public bodies, from individuals. Corporate subscribers can be emailed without prior consent. Sole traders and most ordinary partnerships count as individuals, so consent or the existing-customer route applies. Very early companies and solo founders are disproportionately sole traders, so a list skewed to the smallest UK businesses needs a higher share of rows checked before sending, not fewer.

Read the source

A startup usually has one domain, so a reputation hit is not contained

Google requires SPF, DKIM and a DMARC record for senders approaching 5,000 messages a day to personal Gmail accounts, with From-domain alignment, one-click unsubscribe under RFC 8058 for marketing mail, a spam rate below 0.1 percent that never reaches 0.3 percent, and 48 hours as the working window for honoring an unsubscribe. A larger company can absorb a mistake on a secondary domain. A startup running product mail, billing mail and the founder's own correspondence on one Google Workspace domain cannot, which is why sending happens on separate domains from the first day.

Read the source

The actual document

The readiness brief we fill in before any list exists

GrowthPal completes the document below before a single row is built, and publishes it whole rather than as a screenshot, because the shape of it is the argument. Several fields can end the conversation, which is the point of filling them in first.

Field	What goes in it
Offer stability	When the offer, the price and the target last changed. If any of the three moved inside the last quarter, the rest of this brief is premature and we say so here.
Proof people already buy	Deals closed at a price you can repeat, and whether anyone other than the founder closed one of them. Design partners and discounted pilots are noted separately, because they prove interest rather than a price.
Reachable market count	Not the industry, the people who can sign, counted from a real source. Against roughly 30,000 contacts, with the shortfall stated plainly if there is one.
Discovery call coverage	Who answers a reply within a day, who answers when that person is traveling, and how many calls a week the calendar can actually hold.
Stage and trigger	The dated public event that earns the email, with the URL it was read from: the Form D notice, the first-sales-seat posting, the batch label, the archived pricing page.
Trigger age	How old the signal is against your sales cycle. Anything older than the cycle is a fact about the company, not a reason to write this week.
Sending domain plan	Which domains carry outbound, confirmed as separate from the one carrying product mail, billing mail and the founder's own correspondence. Warmed before volume, no open or click tracking.
Investor conflict suppression	Portfolio companies of your own investors, existing customers, prior opt-outs and anyone already in a live conversation, removed before the first send rather than apologized for after.
Founder time budget	What is needed and when: the offer brief, the objection list, notes from calls that went well, and the approval turnaround we can rely on. Stated as tasks and deadlines, never as an invented hours-per-week figure.
What would tell us this is wrong	Written down before sending. A hypothesis with no failure condition teaches you nothing, because no result would ever count against it, and at this stage that is the most expensive mistake available.

We publish the brief rather than a dashboard screenshot because a screenshot proves a number and a brief proves a method. The first three fields are the ones that most often end the conversation, and they stay at the top for that reason.

Process

How the engagement runs

Three stages and nine steps, the same ones in every market: build the list from public sources, race message angles against it, scale the one that wins. The nine steps from kickoff to booked meetings are written out in full.

What changes for this market

Week one is the readiness check, and it can end the engagement. Offer stability, team coverage for replies, and a counted reachable market, done against your own numbers rather than an estimate. If the answer is not yet, you get that in writing with the reason, and nothing is built.
List building starts at the regulator and the job board, not at a database filter. Form D notices on EDGAR, Companies House filings, public applicant tracking boards, accelerator batch directories, the pricing page against its archived version. Every row carries its trigger, the public URL it was read from, and an investor conflict check.
The send calendar is paced for a founder's week. Separate sending domains warmed before any send, no open or click tracking, and an approval loop short enough that angles do not sit waiting for a reply from someone on a plane. Discovery call coverage, including the backup, is agreed before the first send.
The first ask is answerable without a meeting, and every campaign leaves a written record of which angles were tested and retired against which segment. That record is the part you keep when the first sales hire starts, whether or not the engagement continues.

Proof

What GrowthPal has published

GrowthPal takes on outbound engagements in this market, and the method above is the one it runs here. The published record below is the whole company record, across every client.

What was published	What the figure counts	Source
449 interested replies from 67,934 contacts reached in one month, about 1 per 150, with 955 replies in total	Contacts reached, in one month.	Read it
A study of 412 campaigns across 47 companies over 18 months: campaigns scoring above 0.7 on the composite message-market fit score averaged an 11.4 percent reply rate, against 1.1 percent for campaigns below 0.3. Volume, warm-up and time of day explained less than 8 percent of the variance.	Total reply rate, measured against messages in those campaigns. This is a different measure from interested replies per contact and the two must never be compared on the same page.	Read it
4M+ cold emails sent for 40+ B2B companies	Emails sent, cumulative across all clients. Not one campaign.	Read it

How these are counted

An interested reply means a human reply expressing interest in continuing the conversation. Out-of-office messages, unsubscribes, referrals to someone else and automated responses are not counted. A figure measured against contacts reached and a figure measured against emails sent are different measures: the column above says which each one is, and we do not convert between them or add them together.

The full counting method, and what an interested reply is.

On an industry-specific result. GrowthPal has not published a client outcome for a startup at the moment it crossed the readiness bar, so this page offers the method instead: the public sources a list is built from, the checks that decide whether to start at all, and the angles we would test first. The best-month figures shown on this site come from campaigns Prabesh Khanal led at a previous agency, used with written permission, not from a GrowthPal client engagement.

Honestly

Who this is not for

Before product-market fit, which is the most common no. Outbound finds people who have a problem you already solve. It cannot tell you what to build. If the offer, the price or the target still moves every month, every negative result is unattributable: you will read a message failure as a market failure and retire a market that was never tested. Discovery conversations, design partners and losing deals for one consistent reason teach you more at that stage, and cost less. Come back when the offer has held for a quarter.
Below roughly 3M in ARR. The bar is not about the size of the invoice. Below it, the two things that make outbound work, an offer people already buy and enough reachable contacts to test against, are usually missing at the same time, and an engagement spends real money measuring the wrong variable. If you call us below the bar, the useful output is which of the two you are short on, because they have different fixes. That answer comes in the first call, not after a contract.
A reachable market smaller than roughly 30,000 contacts. With a few thousand accounts, a system costs more than it returns. The list is exhausted before enough variants have been tested to learn anything, and re-contacting the same people too soon makes the next attempt worse. The right method at that size is manual: twenty to fifty named accounts you would be proud to win, real research per account, founder-sent from a personal mailbox, no automation layer. Outbound still works. A program does not.
Nobody with a calendar to take the calls. Replies arrive within days of the first sends, and at this size they land on whoever is free. If the founder is the only person who can answer and is traveling for three weeks, interest goes cold and the spend is wasted. Discovery call coverage gets agreed before launch, including the backup. A founder who cannot name the backup is telling you the timing is wrong.
If you want outbound to validate the product. This is the expensive version of the first no, and it is worth stating separately because it is usually unspoken. Using cold email as market research produces a result with too many moving parts to interpret, on a channel where a bad first attempt leaves a domain reputation you then have to repair. Validation is cheaper by conversation than by campaign. Published startup failure rates will not settle the question for you either: most of the figures in circulation have no primary source behind them.

Questions

Answered straight

We just raised. Is that not exactly the moment to start outbound?

A round funds a plan. It does not create an offer people already buy. The useful question is what changed about the offer, not what changed about the bank balance. If customers were buying before the round, at a price you can repeat, the round makes the timing better and usually funds the first sales seat too. If the round is funding the search for that offer, outbound will measure the search and report it back as a message failure.

We are below 3M in ARR. Are you just going to tell us no?

Probably, and you will hear it on the first call rather than after a contract. The bar exists because below it the offer is usually still moving and the reachable market is usually too small, at the same time. What you get from that call is a written note on which of the two is missing, because offer stability and market size have completely different fixes. The bar is checkable, so you can tell when you have cleared it.

Our reachable market is about 4,000 accounts. Does that disqualify us?

From a program, usually yes. From outbound, no. At that size the right approach is named accounts worked manually: real research per account, founder-sent, no automation layer, and a list you can afford to be slow with. A system burns through 4,000 accounts before enough variants have run to tell a message that works from one that got lucky, and then the same people cannot be approached again for months.

How much founder time does this actually cost?

More than founders expect at the start, less later. The front-loaded work is the offer brief, the objection list, notes or recordings from calls that went well, and fast approval of angles. After that it settles into a short weekly review and a calendar with room for replies. No credible hours-per-week figure is published for this anywhere, including by us, so treat any vendor who quotes you one as guessing.

Our founder still closes every deal. Is that a problem?

Not yet, and it is the healthy state at this size. It becomes the problem the month the founder's calendar caps the pipeline, or the month the first sales hire starts and finds no sales playbook handover waiting. The practical tell is whether anyone other than the founder has closed a deal of the normal shape. If nobody has, outbound produces meetings only the founder can take, which raises the ceiling without moving the constraint.

What should we do instead if we are not ready yet?

Three things, in order. Pick twenty to fifty named accounts you would be genuinely proud to win and work them manually, founder-sent, with no tooling. Write down the offer, the price and the one objection you keep losing on, and hold all three steady for a quarter. Then count your reachable market honestly: not the industry, the people who can actually sign. Come back when the offer has stopped moving and the count clears the bar.

Will outbound tell us whether our product has a market?

No, and this is the costliest misreading of the channel. Outbound reaches people who already have the problem you solve, so a low reply rate from a moving offer is not evidence about the market. It is evidence that too many variables moved at once. Be wary of the startup failure statistics used to argue otherwise: the widely quoted rates have no primary source, and the definitions behind them change from one telling to the next.

How quickly does a startup get a read on the message?

First sends land two to four weeks after kickoff, once separate sending domains are warmed, and week eight is the usual point at which one angle pulls ahead. The startup-specific caveat is that week eight is only a read on the segment and the offer you launched with. If the offer changed in week five, which happens at this stage more than anywhere else, the eight-week result describes a message that is no longer the one you sell.

Who you are really selling to

What makes outbound different here

The offer may still be moving

Testing needs volume, and a startup's market is often too small for it

Founder attention is the binding constraint, not budget

Replies land on whoever is free, and that is usually the founder

Founder-led sales hides the absence of a motion

Buying triggers, and the angle each one earns

Example angles we would test

How the list gets built in this market

Funding, entity and regulator data

Cohort and program directories

Hiring surface

Company surface and people data

How a row gets verified

What makes a bad list here

Compliance and trust constraints

CAN-SPAM needs a valid physical postal address, which a remote company often lacks

A list of small UK companies carries more consent-required rows than a list of large ones

A startup usually has one domain, so a reputation hit is not contained

The readiness brief we fill in before any list exists

How the engagement runs

What changes for this market

What GrowthPal has published

How these are counted

Who this is not for

Answered straight

Related reading