This is a sprint, not a program
Budget three to four weeks for the first pass: a week to build the list and write the message, then two to three weeks running it and booking calls. If the list is under 100 accounts, a founder working this alongside product can run the whole thing solo.
This blueprint assumes a working ICP hypothesis from the ICP Discovery Framework (00.1), even an early one. Run five non-switcher interviews before building the list. A sharp list built on a vague ICP still produces a vague result.
The purpose of this sprint is not pipeline. It is signal. A stranger with no reason to be polite either replies or does not. Every message sent is both a prospecting attempt and a test of your ICP hypothesis, track it that way.
This is not a sales development representative (SDR) motion. Treat the list as a single, tightly targeted campaign, not the first cohort of a perpetual machine. Volume is not the lever here. Precision is.
A single sprint is not a permanent outbound function. Plan on running sprints back to back, each one narrower and sharper than the last, until the signals in Step Six tell you it is time to build something more permanent.
Why outbound is the only motion that works here
Warm intros and inbound produce dishonest signal at low volume. A prospect who takes a call because a mutual contact asked them to is being polite, not interested, and a founder reading that politeness as market validation is the single most common way early go-to-market (GTM) decisions get made on bad data.
The difference determines whether the next 30 conversations teach you anything.
Daniel Hsu, early operator at Retool. Feels like traction. Tells you almost nothing about whether the problem is real enough to pay for.
Pete Kazanjy, author of Founding Sales. A non-friendly party has no reason to soften the truth. That is exactly the data a pre-PMF founder needs.
Aggregate cold email reply rates have collapsed since 2019, and the decline follows a steep list-size penalty. A sharply targeted list of 300 outperforms a generic list of 3,000, and takes a fraction of the time to build.
Typical for a tightly qualified, hand-researched list.
Volume replaces precision, and the reply rate collapses with it.
The market-wide average cold email reply rate.
Current average, per Instantly's data across billions of sends.
Build the List
Start with the ICP canvas from the ICP Discovery Framework (00.1). Each of its four dimensions becomes a filter or a search term, sourced in this order.
- 1 / Your own network firstFirst- and second-degree LinkedIn connections inside your firmographic boundary. The highest-conversion segment on the list, and the easiest to get right. Do not skip it because it feels too easy.
- 2 / Trigger-event searchSearch LinkedIn and company news for the specific trigger event in your ICP canvas: a funding round, a leadership change, a compliance deadline, a headcount jump. A visible trigger beats ten companies where it is assumed.
- 3 / Behavioral-signal searchLook for the Dimension 4 signals from your canvas: posts, job listings, tool-stack mentions, conference attendance. These tell you who is actively feeling the problem, not just structurally eligible for it.
- 4 / Enrichment lastOnce you have a raw account list, enrich it for contact details and verified email. This is where Clay's waterfall enrichment earns its cost, querying multiple data providers in sequence until one returns a valid record.
Every account on the final list should pass a one-line test: can you write the first sentence of the outreach message without guessing? If you cannot name the specific trigger or signal that put this account on the list, it does not belong yet. Cut it or research it further before sending anything.
Write the Message
Every message in this sprint follows the same shape, in this order.
- 1 / The observationA specific, accurate statement about the account, tied to the trigger event or signal that put it on the list.
- I noticed [Company] closed a Series A in the last two months and posted a Head of Revenue Operations role last week.
- 2 / The patternOne sentence connecting that observation to a problem you have heard described by others in a similar position.
- In conversations with others at this stage, the reporting gap between finance and sales tends to widen right after that kind of hire.
- 3 / The low-pressure askA specific, small ask with a named time commitment. Answerable with a yes or a no.
- Would you have 15 minutes to talk about how you're handling this?
- 4 / Nothing elseNo feature list, no pricing, no deck attachment. If the prospect wants to know what you have built, that comes up in the conversation you are asking for, not in the message asking for it.
Positioning the founder as gathering information, not selling a product, removes the biggest reason a stranger ignores a cold message.
Reads as a sales conversation the prospect did not ask for. Ignored by default.
Answerable, low-pressure, and, at Scale 00, often the more honest framing anyway.
Sequence and Choose Channels
Four touches is a ceiling, not a floor. If an account clearly is not going to respond, do not force the full sequence, the point is signal quality, not sequence completion.
- Touch 1 / Day 1LinkedIn connection request or direct message, using the four-part structure from Step Two.
- Touch 2 / Day 4, if no responseA short email referencing the same observation, shorter than the LinkedIn message. Assume they saw the first touch and did not have time, not that they were uninterested.
- Touch 3 / Day 9, if no responseA different angle on the same trigger, not a repeat of touch one. If touch one referenced the funding round, touch three might reference the resulting hiring pattern instead.
- Touch 4 / Day 16, if no responseA genuine break-up message. State plainly that this is the last outreach on this topic and leave the door open. Converts at a meaningfully higher rate than any single touch before it, because it carries no pressure.
Phone calls and multi-channel automation belong in The Outbound Machine (01.3), once the message and list logic here are proven. Adding a channel before the first two are validated adds complexity without adding signal.
Set Up the Lean Stack
Do not buy an autonomous AI SDR platform at this stage. Tools like Artisan and 11x carry steep monthly minimums that assume a dedicated operator you do not have yet, and the practitioner consensus is generic output.
Waterfall enrichment queries multiple data providers in sequence until one returns a verified record. Often replaces a €15k/year ZoomInfo or Cognism contract at this stage.
Dedicated deliverability for a small volume of highly personalized email. Apollo is a reasonable single-tool alternative at €50–80/seat.
Outreach and Salesloft, priced at €130 to €180 a seat with four-to-eight-week setup times, are built for teams running volume through a dedicated SDR function. Running a 300-account founder sprint through them is significant overhead for no corresponding benefit. The same logic applies to Salesforce as a customer relationship management (CRM) at this stage, a Notion table or a free HubSpot CRM tracks a 300-account sprint perfectly well.
Run the Sprint and Read the Signal
Roughly 50 to 75 accounts a week for a list of 300 run over a month gives a running reply rate to compare week over week, rather than one lagging number at the end. Given the list-size penalty in Section 1, a well-built list run with the structure from Step Two should produce a reply rate meaningfully above the 3.43% aggregate average. If it is tracking the average or below, the list is not narrow enough, the message is not specific enough, or both. Fix targeting before adding volume.
Log every reply, not just the ones that turn into a call. A "not right now, check back in Q3" is a timing signal for your disqualification log. A "we already use [competitor]" is a competitive-alternatives signal for positioning. A flat no with no explanation is weaker data but still worth a line, patterns across enough flat nos tell you something about the account type, even without a stated reason.
The sprint is run.
A real inflection point, whether or not it produced a wave of meetings. You now have a tested message, a sharper ICP, real conversations, and an early read on whether the motion travels beyond your personal network.
You know, from real replies, which part of the observation-pattern-ask structure lands and which part gets ignored. Worth more than any amount of internal debate about messaging.
Every reply, positive or negative, is a data point for the ICP canvas. Accounts that replied well and accounts that went silent both belong in that revision.
A small number of real conversations, and the raw material for the First 10 Customers blueprint (00.4): why each one is talking to you, and what almost stopped them from replying at all.
If a second person ran a smaller version of this sprint and got a comparable reply rate, that is an early signal the motion travels. That matters when the First Sales Hire Blueprint (00.5) becomes relevant.
When to Leave the Sprint Model Behind
- Reply rate and message have stabilizedAcross two or more sprints, you are no longer learning something new about targeting or messaging, you are running the same play again.
- First hires are in the sales motionA documented outbound sprint is exactly what a first Account Executive (AE) needs to inherit. Once that hire is in seat, the manual, single-founder version is no longer the right shape.
- Volume has outgrown one founderIf the list your ICP data justifies has grown past what 50 to 75 personalized sends a week can cover, it is time for the Clay-powered workflow architecture and health metrics dashboard in the Outbound Machine, not a bigger version of this sprint.
Until all three are true, resist the temptation to prematurely systematize. A founder-run sprint, sent by hand with real research behind each message, remains the highest-signal motion available at this stage.
Addenda
Reference material for specific parts of the sprint. Each addendum addresses a distinct challenge founders consistently hit when running this process for the first time.
Trigger-event search patterns for funding rounds, compliance deadlines, and headcount growth, including a worked Boolean LinkedIn Posts search ("SDR team" AND "building", filtered to Past month) that surfaces founders announcing a live hiring trigger in real time. Behavioral-signal search using LinkedIn's content feed and comment activity. Network-first sourcing before searching cold, and how to rank the raw list once assembled.
Read the full addendum →Full templates for a funding-round trigger, a compliance-deadline trigger, a job-posting trigger, and a behavioral-signal trigger, each following the observation-pattern-ask structure. Includes the touch-four break-up message that consistently outperforms the touches before it.
Read the full addendum →A full breakdown of the €200 to €700 a month stack: Clay for enrichment, Smartlead or Instantly for sending, and where a free HubSpot CRM or a Notion table beats Salesforce at this stage. Includes the specific anti-patterns to avoid and why they cost more than they save.
Read the full addendum →Two operator-tested prompts: one to draft the observation and pattern lines from a verified trigger-event detail, and one to batch-check a set of messages for vague observations or premature product mentions before a weekly send. Every draft still gets a human review before it goes out.
Read the full addendum →A simple table kept alongside the disqualification signal log from the ICP Discovery Framework. What a flat or declining reply rate week over week actually means, and how a repeating reply theme becomes pattern data for the next ICP canvas revision.
Read the full addendum →