What actually works in outbound.

Outbound is easier to run than it has ever been and worse than it has ever been. Both things are true because sending got cheap. When volume costs nothing, volume is what you get, and the reply rate falls for everyone including the people doing it well. It is the same arithmetic behind the lead machine myth.

What follows is what holds up. Some of it is measurable, some of it is just what happens every time, and a few pieces are things almost nobody adjusts.

Who sends it changes the result

Outreach from women gets more replies than identical outreach from men. This shows up in team data, in split tests, and in every sales org I have worked inside. The gap is large enough that it should affect how you staff and who signs the message, and small enough that nobody talks about it.

What drives it is ordinary: less pressure in the phrasing, more willingness to ask a genuine question, less presumption on the reader’s time. That also means it is a writing habit as much as a fixed advantage, and habits can be taught.

So if your outbound is signed entirely by men and underperforming, test that before you rewrite the copy again.

People know when they are on a list

This is the part teams underestimate. Buyers are not offended by being contacted. They are offended by being processed — by the recognition that a system selected them, a template addressed them, and nobody involved would notice if they never replied.

Nobody minds being contacted. They mind being processed.

The tell is rarely the merge field. It is the shape of the message: an opening line that could apply to any company, a compliment that took no effort, a problem statement written from your category rather than their week. It is the same failure mode as content that could have come from anyone.

One genuinely specific sentence fixes most of this. Something you actually read, watched, or noticed — a job posting, a talk, a change in how they describe themselves. Not flattery. Evidence that a person spent 2 minutes understanding how they might be able to authentically help. Observe, do not diagnose: guessing at their problem is a template with better manners, and buyers can tell the difference between being noticed and being profiled.

The basics that still work

• One idea per email. If there are 2, send 2 emails, or send 1 and drop the second.
• Short enough to read on a phone without scrolling twice. Under about 120 words.
• Their language, not yours. Describe the problem the way it gets described internally, not the way your category describes it.
• A small ask. A question they can answer in 1 line beats a meeting request.
• No attachment, no deck, no calendar link in the first message.
• Plain text. Tracked images and heavy signatures both hurt deliverability and look automated.
• Say what you sell, plainly.
• Do not write to sound impressive. The reader is deciding whether to answer, not grading the prose — vocabulary that signals expertise to your peers reads as distance to a buyer.

None of this is new. It is just consistently skipped, because it is harder to write 90 good words than 300 mediocre ones.

Things almost nobody adjusts

The follow-up that adds nothing is worse than no follow-up. Bumping a message to the top of an inbox tells the reader you have nothing further to say and are contacting them anyway. Either add a fact, or stop.

Give them an easy exit. A line offering that no reply is needed if this is not their area raises reply quality, because the people who do respond are self-selected. Teams avoid it because it feels like handing back leverage. It does the opposite, and it sits comfortably alongside the CAN-SPAM requirements you are already subject to.

Reply to the reply properly. Most outbound programs are built entirely around the first send and fall apart at the response, where a real person now has to answer a real question quickly and in a voice that matches the email they just got. That handoff is where most pipeline is lost.

Watch the sending domain. Lookalike domains and burner subdomains protect deliverability at the cost of the brand — a buyer who checks the sender and sees a domain that is nearly your company learns something about you. Mailbox providers publish what they expect on authentication and bulk sending; Google's sender guidelines are the baseline worth meeting before you start renaming domains.

Read your own outbound as a brand asset. For most companies, cold email reaches more of the market than the website does. It is often the only writing nobody in brand or marketing has approved — which matters most if you have already done the work of deciding how the company sounds.

What to measure

Reply rate is a weak target: it rewards provocation. Track qualified replies instead, measured per hour of human effort rather than per message sent. That is the number that shows whether selectivity or volume is working, and it is what a signal-led program optimizes for.

Negative replies are data, not failure. “Wrong person” corrects your targeting: the title you picked does not own the problem, which is far cheaper to learn from a one-line reply than from a quarter of flat numbers. “We already bought this” is better still — it confirms the budget line exists, names who won it, and tells you where in the cycle you arrived. “Not now” gives you a date to work backwards from. Silence gives you none of that, which is why a program producing only silence is usually mistargeted rather than badly written.

Part of our work on demand & lead gen.