Building an outbound engine from scratch: a 30-day plan

Why the order beats the tactics

I'm Hlib Storchak. I build outbound engines for B2B founders and sales teams, and this is the order I build them in.

Outbound fails from bad sequencing far more often than from bad execution. The classic version: the team buys a sending platform in week one, spends two weeks perfecting copy, then builds a list for it, then discovers the domains need warming and the launch slips a month. Nothing there is wrong on its own. The order wrecks it. Copy written before you know the segment is a guess, and infrastructure bought before either is a subscription you pay for while you think.

Before day one

What you need is decisions, not purchases. Who owns this: one named person with real time each day, not a responsibility shared across three people with other jobs. What a meeting is worth to you, roughly, because without it you cannot judge anything at day 30. What you will spend and for how long, decided before the first flat week triggers a panic. And where replies land, with a name against the inbox. Teams routinely sort that last one on day 22, after losing a week of replies.

Days 1 to 5: the offer

Write down the one painful problem you solve, who has it, and the proof that you solve it. If it takes more than two sentences, the market will not decode it from a subject line.

Pick a single segment. Three means three lists, three sets of copy and a third of the data on each, which is the reliable way to learn nothing about all of them. Pick where your proof is strongest, even if it is not the biggest market. Then test it on ten real people in conversation. If nobody leans in when you explain it out loud, no sequence will rescue it. This is where the low-risk first step gets designed, because an offer that is hard to ignore has the risk reversal built in rather than bolted on.

Days 6 to 12: the list

Start from accounts, not contacts. Define the company criteria first: size, market, stage, technology, and the trigger that says the problem is live now. Then find the buying titles inside those companies. Filtering a database for job titles across an industry produces accurate records of people who will never buy.

Aim for a few hundred exact-fit accounts. A precise list is cheaper at every stage and gives a cleaner read on the offer. You can scale a list that works, but you cannot unlearn a false conclusion from one that never fit. Verify every address, check a random sample by hand, and write the criteria down: in month two the criteria are the asset, not the spreadsheet.

Days 8 to 14: sending infrastructure

This starts while the list is still being built, because warmup is the longest pole. Register dedicated sending domains, set up SPF, DKIM and DMARC, create mailboxes with real human names, and start warming. Never send cold volume from your main company domain: if it gets flagged, your invoices, support replies and recruiting emails go to spam with it.

Plan for three to five domains with two or three mailboxes each, so each mailbox stays at a modest daily number and one damaged domain does not stop the campaign. The ramp schedule and pre-send checks are in my domain warmup playbook. On tools, I default to Salesforge for sending and Warmforge for warmup: personal preference from client work, not an objective verdict. Instantly and Smartlead do the same job well, and familiarity beats any feature difference in month one.

Days 15 to 20: copy and channels

Short, first person, one idea per email. Reference a real reason you are writing to this company, make one claim you can back, ask for one small thing. No five-part value proposition, no attachments, minimal links. Build a three to four step sequence, because a meaningful share of replies arrive after the first message, and write two genuinely different angles rather than ten variants.

Two channels maximum, and one should be email: it is the only one where you can run enough volume in four weeks to learn something real. LinkedIn works best as a support layer, a connection request around the time the email lands, rather than a second full sequence. Calling belongs in month two, once the message is proven.

Days 21 to 25: the controlled launch

Start small and deliberately slow. Send to a slice of the list, a few dozen per mailbox per day at most, and watch before opening the taps. A controlled launch is diagnostic: it separates a deliverability problem from a list problem from a copy problem while each is still cheap to fix.

Run the pre-flight checks first. Authentication passing on an independent tool. A test email to a personal Gmail and Outlook to see where it lands. Verification run on the file. Reply stops sequence switched on. Link tracking off or minimal, because it hurts placement more than the data helps.

Days 21 to 30: working the replies

The first replies are the point of the month, and where new engines leak worst. Answer in minutes, not days. Expect most early replies to be objections: timing needs a dated reconnect, budget early is usually unproven value, and we already use someone is the warmest reply in the set. The moves are in my guide on getting from a first reply to a booked meeting.

Log every reply by category from day one. At day 30 the shape of your objections is more useful than the raw reply rate, because it tells you which part of the engine to work on next.

Days 26 to 30: read the data

Look at reply rate and positive reply rate by segment, positive replies converted to meetings, meetings held, and bounce rate as an infrastructure check. Ignore open rate: it has been unreliable since privacy protection became the default on major mail clients.

Then make one decision per stage. Which segment answers and which gets cut. Which angle earned its place. Whether the offer needs sharpening or the list rebuilding. A month is enough for those calls. It is not enough to declare the channel dead, which is the most common overreaction I see.

The 30-day map

Note the overlaps in the days column: parallel tracks with one hard dependency, which is that nothing launches until warmup is done.

What it costs to stand this up

Two budgets matter, cash and hours, and the second surprises people more. Everything below is an assumption, shown so you can replace it. None are researched numbers. Start with capacity, because it drives the cash line: daily capacity = domains x mailboxes per domain x emails per mailbox per day. Assume 3 to 5 domains, 2 to 3 mailboxes each, 30 to 50 emails per mailbox per day after warmup. That is roughly 200 to 700 a day, enough for a few hundred accounts through a four step sequence.

Total = cash lines added up + (hours x your loaded hourly rate). The cash cost is usually smaller than people expect and the time cost larger. The hours concentrate in the offer and the list, exactly the stages people rush. If they are not available from the named owner, extend the timeline or have someone run it for you. Compressing it does not work.

Where people lose the month

Polishing copy before the offer is validated, so a week goes into wording the first ten conversations make irrelevant. Buying a huge list because volume feels like progress, which buries the signal and takes the domains down with the bounces. Launching at full volume on fresh domains, which costs a replacement set and another warmup cycle rather than a bad week.

Then two quiet ones. Nobody owning the inbox, so replies arrive on day 23 and get answered on day 26. And changing everything at once when week one looks flat. Week one is meant to look flat. Change one variable at a time or you learn nothing.

What day 31 looks like

If the month went well you are not holding a campaign, you are holding a system: a validated offer, written account criteria you can rebuild a list from, warmed infrastructure that gets more valuable the longer it runs cleanly, two tested sequences, and a categorised set of objections telling you what to fix next.

That compounds in a way a one-off campaign does not. Month two is not a rebuild, it is scaling the segment that answered. After 2000+ meetings booked for B2B clients, the engines still producing a year later are the ones built in this order, by someone who owned it daily and resisted the urge to skip the slow step.

FAQ