You sent four hundred messages last month and got eleven replies. Now you are staring at the message trying to work out which sentence to change.
Here is the uncomfortable part: the sentence is probably not the problem. Before you rewrite anything, look at your connection acceptance rate. If it is under 20%, most of those four hundred people never saw the message at all, and no amount of rewriting fixes that. You have a list problem wearing a copy problem’s clothes.
Once acceptance is healthy, the copy genuinely does decide the outcome. These are the principles that hold up, and the specific failures behind each one.
Your first line is a subject line, whether you designed it that way or not
On LinkedIn the opening sentence is often the only thing visible in the notification. It is doing subject-line work while most people write it like a greeting.
The test is simple. Read your opener and ask whether it could be pasted into a message to somebody else without editing.
Fails the test: “Hi Sarah, I hope this message finds you well. I’m the founder of a platform that helps sales teams scale their outbound.”
Passes: “Sarah, your post about killing your SDR team’s cold call quota got me thinking about what replaced it.”
The second one could only have been sent to one person. That is the whole standard.
Notice where the good version came from though. It came from the list, not from the writing. Someone who commented on a post about outbound hiring last week hands you a first line for free. A name pulled from a title filter does not. If your openers keep coming out generic no matter how long you stare at them, stop editing and go change where the list comes from.
Delete your first sentence
Most drafts carry two or three sentences of throat-clearing before they arrive. “I hope you’re well.” “I know you’re busy.” “I’ll keep this brief,” which is a particularly expensive way to not keep it brief.
Write the draft, delete the first sentence, read it again. Roughly half the time it is better and nothing has been lost. The urge to warm up is a speaking instinct, and this is not speech.
One idea, not four
When a message feels like your only shot, everything gets crammed in: the problem, the product, the proof, the case study, the offer. Five arguments in one paragraph read as none.
Pick the single idea most likely to matter to this specific person. The other four are not wasted, they are your follow-ups, and they land better once the first message has proved you are worth reading.
This is the practical argument for thinking in sequences rather than single messages. When you know a second and third touch exist, the first one is allowed to be small.
Ask for information, not time
The size of your ask sets the ceiling on your reply rate, and most first asks are far too large.
| Ask | What it costs the reader | What happens |
|---|---|---|
| ”Do you have 30 minutes this week?” | Calendar check, prioritisation, commitment | Ignored |
| ”Worth a quick call?” | Still a commitment, just vaguer | Ignored more politely |
| ”Is outbound quality something you’re working on this quarter?” | Yes or no | Answered |
| ”Are you still running this in-house, or has it moved to an agency?” | Pick one | Answered, and it qualifies |
The bottom two ask for information rather than time. They can be answered in four words while walking between meetings, which is exactly when people read LinkedIn.
They also do something a meeting request cannot. “No, we moved that to an agency” is a useful answer. An unanswered meeting request teaches you nothing.
Write it the way it will be read
Most outbound is optimised for the moment of sending. You read the draft at your desk, it sounds professional, you send it. But it will be read on a phone, between two other things, by someone who does not know you.
Read it out loud. If you would not say it to this person across a table, do not send it. “I wanted to reach out regarding” and “leverage our solution to drive efficiencies” do not survive that test, which is why they are everywhere in writing and absent from speech.
Then make it easy to decline. A final message that says “if this isn’t relevant, no problem, happy to leave it there” gets more honest answers than another pitch. It also protects the account: people who can say no gracefully do not report you, and reports are the strongest signal LinkedIn uses to find automated accounts.
What we would tell you not to do
Do not send one identical string to five hundred people. It is a detectable pattern, and more importantly it is obvious to the reader, particularly on LinkedIn where prospects compare notes far more than senders assume.
Do not reach for merge tags as the fix either. Inserting a first name and company is a mail-merge, and everyone recognises one. It is worth doing, it is just not personalization.
What actually works is writing three natural ways to say your greeting, your transition and your ask, and letting the combinations do the work.
How we set this up in Doreach
This is the part of our own process worth showing, because it is where most of the failures get caught.
Write the variations inline, not as separate campaigns. In the Doreach sequence editor you write one message containing your alternatives in braces: {Hi|Hello|Hey} {{first_name|there}}. That is spintax, and Doreach picks one option per send. Three greetings, three transitions and two asks is eighteen combinations from a single message, which is the difference between a pattern and a person.
Give every merge tag a fallback. Write {{first_name|there}}, not {{first_name}}. Doreach supports seven tags: first_name, full_name, last_name, headline, location, public_profile_url and day_today, and merge tags and spintax has the current list and what each returns. There is deliberately no company or title tag, because {{headline}} usually carries both and reads more naturally than a slotted-in company name.
Let the validator stop you. Doreach will not launch a campaign with an unclosed tag, an unsupported tag or a broken spintax block. We built it that way because those failures are invisible while you are writing and extremely visible in someone’s inbox. It also warns as a connection note approaches LinkedIn’s 200-character cap, because merge tags expand at send time and a note that fits in the editor can overflow once a real name and headline are substituted.
Preview against real prospects before launching. Step through ten or fifteen people from your actual audience, not sample data. You are checking that fallbacks read naturally, that every spintax variant makes sense, and that {{headline}} produces a sentence rather than a fragment. This takes two minutes and catches the errors that are expensive at four hundred sends.
How to tell whether the copy is working
Reply rate is the obvious measure and the wrong one to lead with. Read these in order:
- Acceptance rate tells you whether the list is right. Under about 20%, stop and fix the audience. More on the causes in why acceptance rates fall.
- Reply rate tells you whether the writing works. This is the number your copy actually moves.
- Positive reply rate tells you whether the offer works. A variant that doubles replies while doubling “please stop” is not a win.
Change one thing at a time. The opener or the ask, not both. And give each version at least a hundred people before you decide, because below that you are reading normal variation and calling it a result.
Start with the ask. It is the fastest thing to change, it needs no new research, and in our experience it is the element most often set wrong.
So: open your current campaign, rewrite the ask as a question someone could answer in four words, add a second variant of it, and send both to your next two hundred prospects. If you are not running anything yet, start a campaign and write the ask first. The rest of the message is easier once you know what you are actually asking for.
