GTM Engineering: Outbound

By the time a buyer talks to you, they have usually already picked someone. 6sense studied more than 3,500 B2B buyers across three regions over two years and found that buyers make first contact with the vendor they end up choosing 81% of the time, and that they do not speak to any seller until they are roughly 70% of the way through their own process. Gartner puts the share of the total buying cycle that a buying group spends with all potential suppliers combined at 17%.

So the job of outbound is to make your name familiar to the six or seven people who will be in the room when they decide who to call, which is a data and sequencing problem more than a writing one. 54% of the fastest-growing B2B SaaS companies now employ a GTM engineer to handle it, at an average US base salary of $182,000 (Clay, with The Signal).

Email, LinkedIn and targeted ads are the three channels, and most founders run all three at once. They work as three steps, each building on the one before it: email reaches one person at a time and tells you who responds, LinkedIn puts more of your people in front of more of theirs, and targeted ads add paid touches to those same names once you know they are worth converting. In that order, each step tells you how to aim the next.

Step zero: find the correct ICP to target

Before spending your monthly credit budget on people who do not have the problem you are solving, or don’t know about the problem you are solving, you need to find a correct ICP. We have an article on that here: https://simplicitygroup.xyz/blog/icp

You can a/b test responsiveness from different ICPs and nail down a final target using email, as we explain below.

Step one: email finds out who responds

Email is the narrowest of the three, one message to one person, and the cheapest way to ask a large number of people the same question, which makes its first job diagnostic rather than commercial. Use this to a/b test how responsive different types of prospects are to figure out who you should be focused on (the one that has highest response rate and CTR).

Delete leads before you send

Enrichment is the part everyone talks about. You stack data providers in a waterfall (filling in data via many different providers until a cell in your database has 100% correct and recent data), so that you can identify a more niche market. Clay's benchmarks put one provider at 55% to 70% coverage of a B2B list in terms of data acquired for each prospect (emails, job roles, etc.) and a three-provider chain at 80% to 92%; we have covered how to build those separately.

What matters more is how many of those contacts you then delete.

A list of 5,000 where 1,200 are wrong will damage your sending domain, and that damage carries into every campaign after it. Since February 2024 Google and Yahoo have required SPF, DKIM and DMARC on any domain sending more than 5,000 messages a day to Gmail, and they enforce a spam complaint ceiling of 0.3%. Google's guidance is to stay under 0.1%, because 0.3% is where enforcement starts rather than where you are safe, which leaves you a margin of three complaints per thousand emails.

So the filtering has to run before the send. That means removing contacts whose title matched your search string but whose function does not, which is most of the Heads of Growth at four-person agencies selling something adjacent to you; companies outside your size band that enrichment happened to find a good email for; and anyone who replied no in the last six months.

Expect to lose between a fifth and half of what you just paid to enrich. That feels wasteful, but the domain you protect by not emailing the wrong 1,200 people is the same one that has to reach the right 3,800.

The sending itself runs through something like Instantly, which rotates mailboxes and handles warm-up so the volume never lands on one domain, or Smartlead if you want unlimited mailboxes and full API access and can live with a rougher interface. Both of them make volume easy, which is the trap, because easy volume is how teams end up the wrong side of that 0.3% ceiling. The wider stack is a separate piece.

Read the replies by segment

The mistake is reading the aggregate number. A 2% reply rate across 5,000 contacts averages together segments that behave nothing like each other, so it tells you nothing you can act on. Split the same send into cohorts of a few hundred, one variable apart, and the differences show up: Series A fintechs replying at 6%, Series C fintechs at 0.4%, agencies at 1%, all of them receiving the same message, etc.. Those numbers are illustrative, but the spread is real and it is usually about that wide.

What the spread tells you about is the segment. Copy is a second-order variable and there is no point tuning it until the segment is settled.

Negative replies are data as well. "We already do this in-house" means the pain is recognised and the budget is allocated somewhere else, which qualifies the segment, whilst something like "what is this" means the segment does not have the problem at all. In our own campaigns the pull is to file both as failures and go and rewrite the subject line, which is how outbound programmes end up spending four months tuning the wrong variable.

6 to 8 weeks and a few thousand contacts gets you a response map by segment, and every step after this one is built on it. It is the stage worth being slow on.

What the filtering and the segmenting buy you is the ability to open on something true about the specific person reading. Instantly's 2026 report puts the average cold email reply rate at 3.43%, and Backlinko's study of 12 million outreach emails found that 8.5% of them got any reply at all, while Autobound's 2026 data puts campaigns that open on a trigger, something the company did recently, between 15% and 25%. That gap is the entire return on the data work. An email opening with "you are hiring a compliance lead in Dubai" has earned the next line, and the same email without that opener is spam that happens to be well targeted.

Step two: LinkedIn widens who does the asking

Email goes out from one mailbox to one contact. The second step widens both ends of that, putting more of your people in front of more of theirs.

Gartner puts a typical buying group at 6 to 10 people per big company, each arriving with 4 or 5 pieces of research they have done independently and then sharing with the others on their team. 1 founder/c-suite member cannot answer all of that credibly, which is why founder-led content tends to flatten into generalities after a couple of months without his team.

Everyone in the company posts

The alternative is putting the whole company on the platform, with each person writing about the work they do. An engineer writes up the technical problem they solved last week; someone in sales writes about the objection they have now heard four times this month. Very little of it mentions the product, and it lands better with its specific reader than anything the founder could write, because it reads as one practitioner talking to another.

It goes wrong when it is mandated, and when people are handed brand-approved copy to post, both of which produce a company page with extra steps and a team that stops after three weeks. It works when people write in their own voice about work they are proud of, so the internal job is editing and encouragement.

Then the outreach

Outreach comes after the posting rather than alongside it. 6 to 8 weeks of the team posting means that by the time your engineer sends a connection request to their engineer, the name has already been through that person's feed a few times attached to something useful, and the same message reads completely differently coming from someone whose posts they have half-read.

The targeting widens here too. Email went to the one contact you picked; now your engineer is talking to their engineer while your operations lead is talking to theirs, so you are working the buying group rather than a single name on it.

HeyReach will run that outreach across several team accounts at once, with the caveat that automating LinkedIn breaches the platform's terms and restricted accounts are common, so the daily limits stay conservative. The messages worth the most should be going out by hand anyway.

And one message by hand

The step that converts best is the one no tool does for you. Once the automated sequences have run on both channels, someone writes individually to the accounts that engaged without replying, from their own name and their own mailbox.

It has to come after the automation rather than instead of it, because the sequences have already done the useful work of showing you who is paying attention. Anyone who opened repeatedly, accepted the connection, engaged with a post, or started a reply and went quiet goes onto that list, and the list should be short enough for a founder to clear in an hour a week.

The sender matters as much as the message: in a small company that is a founder, in a larger one whoever the recipient would recognise from the posting, so your engineer writes to their engineer.

The message itself references the specific thing that prompted it, runs to a few lines, and leaves out the pitch they have already had four times. Keep the ask smaller than the one in the sequence, closer to a question than a meeting request, because the point is to start the conversation.

P.S. hot tip, send multiple messages as if you’re texting a friend. Response rate is much higher.

Step three: targeted ads convert the leads you already have

Paid ads are sold as a lead source, and at B2B volumes they are an expensive one. You pay per impression whether or not the targeting is right, so buying strangers is a slow way to learn what a 400-contact email cohort would have told you in a fortnight.

The version that works is narrower. You upload the people already in the sequence, pay to appear in front of those names and nobody else, and the ad becomes the third or sixth touch on someone who has already had an email and seen a couple of posts. The budget buys extra frequency against a list you already have, so it converts people rather than acquiring them.

That is why it comes third. Steps one and two produce the list of people worth spending on, and until a segment has replied on email and engaged on LinkedIn there is nothing to reinforce. A segment earns ad budget once it has cleared a reply-rate threshold, and everything else stays on the free channels until it does.

Upload the list, then exclude everything else

Upload the outbound list as a matched audience and layer in a retargeting pool of everyone who has opened the site or engaged with the team's posts. Then exclude everything else, including the lookalike audiences the platform will push at you, because a lookalike is prospecting with extra steps and prospecting is what step one is for.

By this point the ad is the fourth thing that person has seen from you in a fortnight, which is what moves a name from vaguely familiar to worth a reply. Judge it on what the accounts in that audience do next, in replies and meetings booked, rather than on the leads the ad account claims credit for.

It also settles the format for you. LinkedIn's Thought Leader Ads, which promote an individual's post from their own profile instead of a brand asset, benchmark at around 4.65% click-through and $0.51 per click against 0.68% and $2.42 for standard formats. Treat those numbers as directional, since nearly all of them are published by vendors who benefit from the format looking good, and your own trailing baseline is the only honest comparison. The direction holds consistently enough across sources to plan around, and what it points at is the post your team has already written.

You cannot run that ad if nobody at your company posts, which is the whole reason ads sit third.

You probably cannot A/B test properly

The standard advice is to A/B test the ads, and most B2B teams do not have the volume for it.

At a 3% baseline conversion rate, detecting a 20% improvement at 95% confidence needs somewhere around 4,800 conversions per variant. Meta's own floor is 50 conversions per variation over a seven-day minimum, and that is for spotting large differences. A campaign producing 40 leads a month, which is a good month for plenty of B2B teams, gives you twelve conversions per arm across a fortnight, and at twelve you are reading random variation.

The audience is not a variable here either, since it is fixed to the list you uploaded, so the message is the only thing left to test. Keep the differences large enough to show up at your volume, meaning whole angles and whole offers rather than headline variants and button colours, and run significance on clicks and engagements, which arrive in the hundreds, treating differences at the lead level as directional until the sample catches up.

Most of the testing has happened in the first two steps anyway. A message that reaches paid has been through a few thousand emails, a few dozen posts and a round of hand-written follow-ups, so you already know which framing pulled replies out of that segment; the free channels have the sample size and the paid one does not.

Doing it out of order

Running the three at once disconnects the feedback loops that make each of them work. Budget goes out against segments email has not qualified, outreach lands on people who have never seen your name because the team has not started posting, and the sending domain picks up complaints from contacts that should have been filtered out, which quietly degrades the one channel you were relying on to tell you whether any of this was working.

In sequence, each step pays for the next: filtering makes the email data trustworthy, the email data tells you which segments deserve human attention, the posting makes the outreach land and produces the creative you will put budget behind, and between them they tell you which segment is worth paying for and what to say to it.

None of it is quick. A properly sequenced outbound build takes about a quarter before the paid step switches on, and the first number to move is the share of your list worth emailing at all, which takes a while to feel like progress.

This is the outbound half of what we build inside GTM engagements at Simplicity Group, and we run the same stack on ourselves; the enrichment, the filtering, the drafting and the sending are automated, which is the only reason a team our size can hold a list this large without a full-time SDR. The same work sits behind 60+ B2B clients delivered and $8M+ in ARR secured. If you want yours pressure-tested, that is what our GTM advisory covers.

Most of the work happens before anything gets sent, and nobody has ever been thanked for deleting 1,200 contacts.