Prerequisites
Outbound amplifies whatever clarity you already have. If you can't yet describe which customers get the most value and why, outbound will return a confusing signal at some expense.
Before starting, you want a handful of closed deals with something in common, pricing that survives a cold conversation, and someone who can take a sales call within 48 hours of a positive reply.
- Evidence of willingness to pay
- A pattern across won deals
- Capacity to take meetings
- A named owner for the motion
Define the ICP
Write the ICP as operational criteria, not adjectives. A researcher should be able to build an account list from it without asking a question, and a copywriter should be able to write a first line from it.
Include disqualifiers explicitly. Knowing who to exclude improves outbound faster than almost anything else.
Segment the market
Split the ICP into segments that behave differently: by industry, motion, stack, size or trigger. Segmentation is what makes results interpretable — without it you only ever learn an average.
Start with two or three segments. More than that and you won't reach meaningful volume in any of them.
Map the addressable market
Size each segment. If a segment contains 200 accounts, it cannot carry a pipeline target that assumes thousands of conversations — better to find that out now than in month four.
Build the data layer
Source accounts per segment, deduplicate against your CRM, then map the buying group: who feels the pain, who signs, who implements.
Accuracy beats size. A small verified universe outperforms a large decayed one on every metric that matters.
Enrichment and signals
Use a waterfall across providers rather than trusting one source, and add the signals that indicate urgency: hiring, funding, stack changes, leadership changes, launches.
Store the signal on the record. It becomes both the personalisation input and the variable you later analyse.
Infrastructure
Register dedicated sending domains, separate from your primary domain. Configure SPF, DKIM and DMARC per domain and verify alignment rather than assuming it.
Provision mailboxes, warm them on a schedule, and cap volume per mailbox. Scale by adding identities, not by pushing one harder.
- Dedicated domains
- SPF / DKIM / DMARC verified
- Warmed mailboxes with volume caps
- Sequencer connected to CRM
- Suppression lists loaded
Deliverability
Verify lists before every launch, keep messages plain, watch bounce and complaint rates, and pause automatically when thresholds are breached.
Deliverability is an operating discipline, not a setup step. It degrades quietly, and by the time results drop you've usually lost several weeks.
Messaging
Write about the buyer's situation, not your feature set. The strongest first messages name a specific problem, reference why now, and ask for a small, concrete next step.
Your best material already exists in founder sales calls. Extract it rather than inventing new language.
Cold email
Short, single-idea messages with one ask. Follow-ups should add a new angle or piece of context rather than repeat the first email in a smaller font.
Run LinkedIn as a parallel channel where your buyers are genuinely active, sequenced with email so the prospect experiences one conversation rather than two campaigns.
Sequence architecture
Decide touch count, spacing, channel mix and exit rules based on your sales cycle. Long cycles justify longer, slower sequences; transactional products don't.
Always define how an account exits cleanly so you can approach it again later without looking careless.
SDR workflows
Every reply gets a human response quickly. Neutral replies deserve the most attention — that's where most pipeline is quietly lost.
Track interested-but-not-now with a date and a reason. It compounds into a pipeline of its own.
Qualification
Agree the bar before launch: fit, confirmed problem, authority, timing. Consistency matters far more than which framework you name it after.
Metrics that matter
Vanity metrics make outbound look healthier than it is. Judge the motion on conversations and opportunity quality.
- Qualified conversations per segment
- Reply sentiment breakdown
- Objection categories and frequency
- Meeting-to-opportunity conversion
- Cycle length by segment
Running experiments
Change one variable at a time, ordered by how much a surprising answer would change your model: segment, then offer, then problem framing, then structure.
Record what was inconclusive as well as what won. Half of a good experiment log is negative results.
Scaling
Increase volume only behind segments and messages that produced qualified conversations. Scaling an unproven motion is the standard way to burn a domain estate and a quarter simultaneously.
Common mistakes
Nearly every failed outbound programme we review contains at least three of these.
- Sending from the primary domain
- Buying a list instead of defining a market
- Targeting everyone at once
- Judging performance on opens
- No owner reading replies weekly
- Scaling volume to fix a message problem
Launch checklist
Before the first send:
- ICP written as operational criteria with disqualifiers
- Two or three segments defined and sized
- Verified, enriched account and contact data
- Domains, DNS and warmed mailboxes ready
- Suppression lists loaded from CRM
- Messaging hypotheses written per segment
- Qualification bar agreed
- Weekly review scheduled with a named owner
Not sure this is your constraint?
Build your GTM engine