Outbound Playbook for Startups

This is a rewrite. The first version of this playbook went up in 2023, and most of the tactics in it still work. The strategy underneath them doesn’t. Not because the fundamentals changed, but because the bottleneck moved.

In 2023, the hard part of outbound was getting into the inbox. Deliverability was the wall. Clear it, send enough volume, personalize a little, and you booked meetings. The teams that won were the teams that could send clean at scale.

That edge is gone. Deliverability is still hard, but it’s solved-hard - a known checklist any team can execute. Meanwhile every inbox on earth now gets ten AI-written emails a day that all open with “I noticed you’re the VP of Sales at [Company].” Volume went to zero marginal cost, so volume stopped being an advantage. It became the problem.

The hard part of outbound in 2026 is not the inbox and it’s not the copy. It’s knowing who is actually struggling, right now, and describing that struggle back to them before anyone else does. Everything else in this guide is downstream of that one idea.

So the order has changed. The 2023 version led with deliverability because that was the wall. This version treats deliverability as table stakes - get it right in week one and stop thinking about it - and spends the rest of its length on the part that actually decides whether outbound works: who you target, and how you learn.

Deliverability is table stakes now, not the game

You still have to get this right. You just don’t get to win on it. Here’s the whole checklist:

Send from a separate domain, never your primary. Buy a few lookalike domains (yourco.co, getyourco.com, tryyourco.com). Your main domain’s reputation is not something to gamble on a cold campaign.
Set up SPF, DKIM, and DMARC on every sending domain. Since 2024 these are not optional - Google, Yahoo, and Microsoft reject or spam-folder bulk senders that skip them. Add one-click unsubscribe to every email. Keep spam complaints under 0.3% or you get throttled at the gateway.
Warm every inbox before you send. Two to three weeks, ramping slowly, using a warmup service. Keep daily volume per inbox low (30-50 cold sends) and spread it across multiple inboxes rather than blasting from one.
Keep bounce rates under 3%. Verify every list before you send it. A dirty list will torch a warm domain in a day.
Plain text. No images, no link-heavy signatures, no HTML formatting that screams “marketing email.”

Two things from the old playbook are now actively wrong.

The first is tuning on open rates. Stop. Apple’s Mail Privacy Protection broke open tracking in 2021, and by now a large share of recorded opens are machine prefetches, not humans. If you’re picking subject lines by open rate, you’re picking on noise. Measure replies.

The second is spintax. Swapping “Hi {John|there}” and rotating synonyms to fake variety doesn’t fool filters anymore and never fooled humans. The variation that matters isn’t at the character level. It’s at the segment level - different signals, different situations, genuinely different emails. That’s the rest of this guide.

Get the checklist right, then forget about it. Deliverability is a tax, not a strategy.

Find the struggle, not the industry

This is the whole game now, so slow down here.

Almost every outbound list starts the same way: open Apollo or ZoomInfo or Sales Navigator, pick an industry from the dropdown, set a headcount range and a few titles, export. That dropdown is the problem.

The industry filter is descended from SIC and NAICS codes - a taxonomy built last century for government statisticians who needed to count businesses. It was never built to tell you who to sell to. When you filter for “healthcare,” the database scans self-reported company descriptions and pattern-matches the word. You get a chiropractor, a health-tech startup, an MRI-parts manufacturer, and a marketing agency that works with hospitals. All technically healthcare. None of them a list.

The signal you actually wanted was never in the description field. You didn’t want “companies that describe themselves as healthcare.” You wanted “companies that sell to hospital procurement,” or “companies whose software touches patient data,” or “companies that just hired their first compliance lead.” Those signals don’t live in a dropdown. They live in the company’s website, its job postings, its integrations, its case studies, the people on its team.

For most of the last decade you couldn’t filter on those things, because reading them across a hundred thousand companies was too expensive. So everyone filtered on the description and lost the rest. That constraint is gone. You can now scrape the public web at scale and, more importantly, you can read what you scraped - a model can look at a homepage and tell you whether a company sells to clinicians or to procurement, whether it offers an API, whether its case studies are about hospitals or insurers. The thirty-second judgment a human analyst would make is now something you can run overnight across your entire market.

So here’s the reframe. Stop searching for industries. Start searching for signals.

A signal is any observable fact about a company that implies a specific struggle. The art is the second half of that sentence - the struggle. A job post for a “Head of Compliance” at a Series B fintech is not interesting because they’re hiring. It’s interesting because someone over there is staring down a regulatory deadline with no system and no team. That is a Monday-morning problem you can name. The signal is the door. The struggle is the room.

A few signal types that consistently reveal struggle:

Hiring. A company posting for a role tells you what’s breaking. Posting for three SDRs means they’re scaling outbound and almost certainly fighting deliverability and ramp. Posting for a “RevOps” hire means their data is a mess.
Tech stack. The tools a company runs imply the problems they have. A business running two systems that both take payment has a reconciliation problem. A company on a tool you know has a specific gap has that gap.
Leadership change. A new VP who joined sixty days ago has a mandate, a budget, and no loyalty to the incumbent vendor. That window closes fast.
Funding and growth. A raise means hiring pressure and new targets, usually with a number attached that someone is now personally accountable for.
Public friction. Reviews, support forums, Reddit threads, status pages. People describe their struggles in public constantly. Most of it is unread.
What they sell and to whom. Their case studies and homepage tell you who their customer is. That’s often a sharper filter than any firmographic.

Once you target this way, two things follow.

The first: score your signals. Not every signal is equally hot. A job post that matches your exact solution, a competitor’s customer publicly complaining, a brand-new leader with a mandate - those are high-intent, call it an 8 to 10. A firmographic match with one soft signal is a 5 to 7. Just firmographics, no signal, is a 1 to 4. The score decides how you approach them, which is the next section.

The second, and this is the bigger shift: the list stops being a list. For years a target list was an artifact. You built it, exported it, uploaded it, worked it, watched it go stale, built another one. When the list is a query against the live web, it isn’t a frozen photograph anymore. The query is what you maintain. When a new VP joins, when a company ships an integration, when a hiring pattern shifts, the list updates itself and the campaign moves with it. You stop maintaining a spreadsheet and start maintaining a worldview about who’s worth talking to. The hard, valuable skill becomes writing that query well - specific enough to mean something, broad enough to surface a list a human can actually work.

The list is the message

Here’s the part most teams get backwards. They build a mediocre list, then try to rescue it with clever copy and heavy personalization. It never works, and the reason is simple: if you need a paragraph of personalization to make an email land, your targeting was wrong.

When the list is built on a real signal, the copy almost writes itself, because the email is just you naming the struggle the signal implies. That’s the relationship to internalize: how you target determines what you say. Invest 80% of your effort in who you reach and what qualifies them. The copy is the easy 20%.

Personalization is the most misunderstood word in outbound. It does not mean knowing someone’s bio. “I saw you went to Stanford” is trivia, and worse, it now reads as the output of an automation, because it is. Real personalization is understanding their condition. “Most teams running [tool A] and [tool B] together end up reconciling two payment flows by hand” is personalization, and you can send it to four hundred companies that share the signal without changing a word. Target the condition, not the person.

The structure we use for a first email is three lines: Situation, Insight, Inquisition.

Situation. Name their reality, based on the signal. The first line is about them, never about you. If your email opens with “I” or “We” or your company name, delete it and start over.
Insight. One specific thing that proves you’ve actually seen this problem before - something they’d only hear from someone who’s worked on it, not a feature pitch.
Inquisition. Ask if you got it right. “Is this you?” beats “Can we grab 30 minutes?” every time.

That last point is the one people resist. Your goal in a cold email is not the meeting. It’s the reply. You earn a reply by asking for the truth, not for time. A prospect will tell you whether you read their situation correctly long before they’ll give you a calendar slot, and once they’re in a conversation, the meeting is easy. Ask for truth, not time.

Two patterns cover most situations:

Pain-qualified. When the signal is hot (8 to 10), name the pain directly and ask if it’s real. Signal, specific pain, “is this you?” That’s the whole email.
Permissionless value. When you have rich data on a prospect, do the work first. Run the audit, find the gap, package the finding, and send it before asking for anything. “I found this for you” earns more replies than any pitch.

And keep it short. Three lines for the opener, under 75 words. If you need more than that, you don’t understand the situation well enough yet. Follow-ups should add a new angle, not guilt-trip - never send “just bumping this to the top of your inbox.”

Run experiments, not campaigns

The single biggest difference between teams that win at outbound and teams that don’t is not copy or tooling. It’s that the winners treat every send as a hypothesis and the losers treat it as a campaign.

A campaign is a thing you launch and hope works. An experiment is a thing you launch to learn something specific. The same emails go out either way. The difference is whether you defined, before sending, what you expected to happen and what you’d conclude if you were wrong.

The operating rules:

Every send is a hypothesis. Signal X, sent message Y, should produce reply rate Z. Write that down before you launch. Every reply is data. Every non-reply is data.
One variable at a time. Test the signal, or the subject, or the angle - not all three at once. If you change everything and the numbers move, you’ve learned nothing about why.
Mind your sample. Roughly 200 sends per variant before you trust the result. Below that you’re reading noise.
Measure the right thing. Reply rate, positive-reply rate, and meeting rate. Not opens. A campaign with a 2% positive-reply rate is beating one with a 60% open rate and nothing underneath it.
Scale segments, not copy. When something works, pour volume into that segment and leave the copy alone. When something fails, change the angle, not the words. A weak insight doesn’t get fixed by a better verb.

Then the cadence: ship weekly. A team that runs four experiments in a month will beat a team that spends the month perfecting one, because at the end of the month it has four times the data on what actually converts. The first campaign is almost never the winner. What matters is how fast you get to the next one.

This is also where the compounding happens. Every experiment adds a row to a map: which signals convert, which messages land, which segments are worth more. That map is the real asset. Campaigns end. The map of what works in your market is something you keep, and it gets sharper every week you run the loop. The win condition was never a single great campaign. It’s a repeatable engine and a clear learning loop.

What this means for your team

The old model was a room full of SDRs sending as much volume as humanly possible. That model is dead, and not because SDRs are bad. It’s dead because volume went to zero marginal cost. When anyone can send ten thousand AI-written emails a week, sending ten thousand emails is worth nothing. The constraint moved from how much you can send to how well you can target, and targeting is not a headcount problem.

So the highest-leverage person on a 2026 outbound team is not the one who sends the most. It’s the one who can write the query and design the experiment - part analyst, part copywriter, part operator. One person who deeply understands your buyer’s struggle and can encode it into a signal-based query will out-produce five reps spraying a title-filtered list. That’s not a prediction, it’s already happening.

This is also why fully autonomous “AI SDRs” mostly disappoint. They automate the wrong layer. They make the spray faster and cheaper, which makes the saturated-inbox problem worse, not better. The useful place for AI is inside specific steps - reading the web to find signals, enriching them, drafting against real data - with a human owning targeting and judgment. The system wins. The robot replacing the human does not.

If you’re an early-stage founder reading this, the practical takeaway is: don’t hire the spray. A junior SDR costs around six figures all-in and fails often at the early stage, because you’re asking one inexperienced person to figure out your market by sheer volume. Build the system instead. Find the signals that reveal who’s struggling, write the query, send sharp emails to the right people, and run the whole thing as a loop you learn from. Whether you do that in-house or with a partner, the shape is the same.

The bottleneck was never your copy. It was that your list was a frozen photograph of a market that had already moved by the time you finished taking the picture. Make the list a query, point it at struggle instead of industry, and run it as a system of experiments. The list, finally, gets to be the campaign.