The guide
Does Your Outreach Sound Automated?
First Person Outbound, explained by SimplyB2B
Most founders never notice it happening. The message feels personal when they write it, then lands in an inbox sounding like every other cold pitch the recipient deleted that morning. Automation tells are rarely about the tool you use. They are about the patterns that accumulate when the same voice sends at scale.
Key takeaways
- Generic openers ('I came across your profile') are the single most common automation tell, because no real person talks that way.
- Structural symmetry, every message the same sentence count and rhythm, signals batch-and-blast even without a single wrong word.
- Compliments that could apply to any company in the recipient's industry are the fastest way to signal a mail-merge.
- A reply that sounds subtly different from your LinkedIn posts or DM history creates a trust gap the recipient feels but cannot always name.
- The fix is not rewording the template. It is making sure the message originates from your actual voice, not a shared prompt.
Does my message sound automated?
Read the opening line aloud and ask whether a real person, who had actually looked at the recipient's profile, would phrase it exactly that way. If the sentence would survive a find-and-replace of the recipient's name or company, it sounds automated. Specificity that cannot be copied to the next person is the only reliable test.
The clearest signal is the opener. Phrases like 'I came across your profile and was impressed' survive unchanged across thousands of sends. A real person who was genuinely impressed would say what impressed them, in terms that only fit that company or that post.
A secondary check is rhythm. Read three of your recent messages in a row. If the sentence count, the structure, and the closing line are nearly identical, the recipient's spam filter and the recipient's brain are running the same pattern-match.
Strip the tokens
Remove name, company, title. Does it still read as a pitch?
Say it aloud
Would a real person open a conversation this way?
Check the compliment
Could you send this to their direct competitor unchanged?
Match the voice
Does the DM sound like your posts and comments?
Read the follow-ups in sequence
Is the interval and structure identical across each one?
How to check if your outreach sounds like a template
Copy the message and remove every personalisation token: name, company, job title, industry. If what remains still makes grammatical sense as a pitch, the personalisation was cosmetic. A message genuinely written for one person collapses when you strip the context, because the context was load-bearing, not decorative.
Token-stripped reading is the fastest practical test before you send. It surfaces the difference between a message that references the recipient and a message that was written for the recipient. Only the second one passes.
A second check is the compliment test. Compliments that apply to an entire industry ('companies like yours are navigating a lot of change right now') were written about no one in particular. If the recipient could forward your message to a direct competitor and it would read as equally relevant, it is a template.
What are the signs a cold message is automated?
Six tells experienced buyers spot immediately: an opener no human would say out loud, a compliment too broad to be genuine, a value proposition that names your category instead of their situation, a call-to-action that assumes a meeting before trust exists, structural symmetry across follow-ups, and a tone that does not match the sender's public posts or comment history.
The last tell is the most underestimated. Recipients increasingly cross-reference your LinkedIn activity before they reply. If your comments and posts carry one voice and your DM carries another, the mismatch registers as borrowed-face automation even if every word was technically written by a human.
Follow-up cadence is a compounding tell. A perfectly timed sequence arriving at identical intervals, without any reference to elapsed time or context, reads as scheduled regardless of copy quality. Real follow-ups acknowledge that time has passed and that the recipient's silence was a signal worth respecting.
Why does automated outreach fail even when the copy is good?
Copy quality is not the primary failure point. Trust is built across a profile, not a message. A recipient who sees a well-written DM but a sparse post history, a low connection count in their network, or a recently created account is resolving the mismatch by discounting the message. The message cannot compensate for the account context it arrives in.
This is what makes borrowed-face tools the wrong frame for the problem. An AI agent writing on your behalf from a separate persona, or from a shared voice model trained on other users, does not inherit your profile age, your post history, or your mutual connections. Those trust artifacts cannot be faked or transferred. They accumulate on your own account over time.
The copy is the last mile. The first mile is whether the recipient trusts the account the message came from. Both need to be true for a reply to happen.
How does voice matching actually fix the automation problem?
The mechanism matters here. A voice model seeded from your own sent messages and posts, per account and never shared with other users, reproduces the vocabulary, sentence structure, and hedging patterns that are specific to how you write. It is not a style questionnaire. It is an inference drawn from your actual output, corrected forward every time you edit a draft.
SimplyB2B seeds each user's voice from that user's own LinkedIn sent messages and posts. The model is per account. When you edit a draft, that edit trains your voiceprint and only your voiceprint. The autonomy the system extends to itself expands only as your edits shrink, not on a fixed schedule. The result is outreach that carries the same fingerprint as your public profile, because it was built from the same source.
A B2B advisory practice running its entire outreach on SimplyB2B achieved around a 31 percent reply rate on cold outreach in their own voice, against a cold-outreach norm typically cited between three and ten percent. That result was produced over a single month, with the message arriving from the founder's own account, carrying the founder's own voice.
Can I fix automated-sounding outreach without rebuilding from scratch?
Yes, but the fix is not cosmetic editing of existing templates. Reworded templates still carry template structure. The practical path is to write one message as if you were sending it to a single named person with no copy-paste involved, then identify which parts of that message required knowing something true about them. Those parts are your real personalisation layer.
Everything that survived without requiring real knowledge of the recipient is structural filler and should be cut or rewritten. Once you know what genuine personalisation looks like in your own voice, you can evaluate every future draft against that standard before it sends.
If the volume you need means writing that way at scale is not realistic, the honest answer is that you need a system trained on your voice, not a system trained on best-practice templates. The difference is where the voice originates.
Common questions
Does LinkedIn penalise automated messages?
LinkedIn does not publish a public list of detection criteria, but the platform applies restrictions to accounts showing non-human behavioural patterns such as uniform activity timing and volume spikes. Accounts that run within a warm-up ramp and a fixed daily ceiling, with per-account timing variance, present a behavioural profile closer to organic use.
Is there a difference between an AI-written message and an automated message?
Yes. A message is automated if it is sent without human review at send time or if the voice it carries was not derived from the sender's own writing. A message can be AI-assisted and still sound genuine if the model was trained on the sender's own sent messages and posts, and if drafts passed through the sender's approval before sending.
Why do personalisation tokens like {{first_name}} not fix the automation problem?
Tokens address surface recognition, not structural trust. A message with the recipient's name in the opener but a pitch body that could belong to anyone is still a template. Recipients have read enough of these to recognise the pattern before they reach the second sentence. Real personalisation changes the substance, not just the salutation.
What trust signals does a recipient check that a message alone cannot control?
Profile age, post history, comment activity, and mutual connections are all visible before a recipient opens your DM. These signals accumulate on your own account over time and cannot be replicated by a third-party persona or a shared AI agent. The message arrives in that context and is evaluated against it.
Related: Why outbound at scale fails · First-person outbound · AI SDR vs founder-led outreach · How to do LinkedIn outreach · The graduation model
Hear what your voice actually sounds like at scale
Start with the free Voiceprint read, no account connection required.
Free to set up · pay only when you connect · 30 day money back guarantee