Your outbound makes the same 5 mistakes
What NOT to do when prospecting on LinkedIn and by email. With real data, no sugarcoating.
LinkedIn isn't what it used to be, and your campaign is still written for 2022
These are 5 mistakes almost everyone repeats, and they show up when you cross-reference Apollo, Gong, Lavender, HubSpot, Outreach, ConnectAndSell, Lemlist, UserGems and LinkedIn. What separates the best teams from the rest isn't budget.
The network's social contract has changed. The same messages that used to get accepted now get discarded unopened. The signal is in the numbers, not in the latest hot take:
- <20% · connection
Invitation acceptance has collapsed
The healthy benchmark used to sit at 25-30%. Today the average is below 20%. Sales Navigator · UserGems
- 5% · after a pitch
The "pitch slap" kills the first message
The average reply rate on a first DM is 18%; the moment you add a direct pitch, it drops to 5%. ConnectAndSell
- −33% · before 8 AM
Sending at 6:30 AM isn't being productive
Messages sent before 8 AM local time lose up to 33% of replies. LinkedIn engagement studies
And this is just the context.
Inside: the 5 mistakes with their fix, the table to benchmark your numbers and a prompt that reviews your messages.
01 · Broken tokens: "Hi [Name]"
The mistake: It happens more than you think, and when the lead sees an unrendered token they don't reply: they dismiss and delete. They don't forgive because they don't have to: their inbox is full of others.
The fix: Send yourself the first three of every campaign and read them on your phone before releasing the rest. It's the cheapest check in the whole campaign and the one that saves the most replies.
A badly built machine shows before the first word. These are the numbers for what happens when sending outruns hygiene.
- 9% · banned
Automation tools burn accounts
9% of LinkedIn accounts using aggressive automation extensions end up restricted or banned within 12 months. Your SDR's account isn't disposable, and the network knows it. Source · LinkedIn safety reports
- 70% · no warm-up
No warm-up = straight to spam
Over 70% of new domains sending outbound without warm-up end up in Promotions or spam. It's not cosmetic: the ESP decides within the first 200 deliveries. Source · HubSpot 2025 deliverability
- >5% · bounce
A high bounce rate pauses your account
Above a 5% bounce rate, the major ESPs (Google, Microsoft) start downgrading your sends. Clean the list before you send: pre-verification is the highest-ROI investment in your stack. Source · HubSpot / Google Postmaster
- −33% · before 8 AM
Sending at 6:30 AM isn't being productive
Messages sent before 8:00 AM in the lead's local time lose up to 33% of replies. Your reply rate drops for a dumb reason: they read you at the hour of the boss's emails. Source · LinkedIn engagement studies
- <20% · connection
Invitation acceptance has collapsed
The healthy benchmark used to sit at 25-30%. Today an average account is below 20%; the best barely touch 60%+ with polished profiles and real triggers. Source · Sales Navigator + UserGems
02 · "We're a company specialized in..."
The mistake: The corporate opener gets deleted before it's read. The recipient doesn't want to know what you do, they want to know which of their problems you solve that the others who wrote to them this week don't.
The fix: Open with the signal that made you write: what they posted, what they launched, who they hired. If you can't name that signal, you didn't have a reason to write yet.
Opening by talking about yourself has a price, and it's been measured. You pay it in the first line, which is where they decide whether there's a second.
- 5% · after a pitch
The "pitch slap" kills the first message
The average reply rate on a first DM is 18%. The moment you put a direct pitch in the invitation note, it drops to 5%. The hook has to be a trigger, not a proposal. Source · ConnectAndSell
- 4.45% · 2025 average
The average reply rate dropped again
From 5.1% (2024) to 4.45% (Lemlist 2025). Below 1.32%, your campaign isn't badly written: it's badly targeted or it's landing in spam. Source · Lemlist outbound report 2025
- 90% · silence
9 out of 10 B2B buyers never reply to cold outreach
HubSpot 2025: 90% of B2B professionals don't reply to cold outreach without a clear signal. The asymmetry is brutal, and cranking up volume doesn't move it, it just multiplies the noise. Source · HubSpot State of Sales 2025
- −25% · open
Subject lines over 50 characters
Short subject lines (≤50 chars) hold a 40-47% open rate. Going over 50 characters drags opens down to 18-25%. Mobile cuts them off sooner, so write for that screen. Source · HubSpot deliverability data
03 · "I saw your profile and found it very interesting"
The mistake: Empty personalization. Zero information and zero specifics: the sentence fits your lead just as well as anyone else on the list, which is why it doesn't count as real personalization.
The fix: Quote a specific detail from their profile, one of their posts or their career path. And if you don't have anything to quote, better not say you checked out their LinkedIn profile.
Real personalization requires knowing who you're writing to and why today. Without that, the friendly line just gives away that you didn't look at anything.
- 2× · vs. cold
Personalizing the note doubles acceptance
Invitations with one line of real personalization get accepted 12% of the time, versus 6% for the default note. The barrier to raising your connection rate is 15 seconds per lead. Source · LinkedIn / Lemlist
- 60% · no ICP
6 out of 10 B2B companies have no defined ICP
Companies that say they have an ICP usually have a segment ("mid-sized companies in Spain"), not a profile. Those with a real ICP (industry + size + trigger + pain + decision-maker) play in a different league. Source · LinkedIn / Forrester
- 80% · unqualified
80% of your leads shouldn't be leads
Accounts that don't fit the ICP burn resources: SDR time, domains, AE patience. Tightening the list cuts pipeline in volume but multiplies closes. Source · Aberdeen / Salesforce
- −75% · spray & pray
Spray & pray makes you pay twice
Massive unqualified lists drag reply rates down by up to 75%. And they burn your domain. When you go back to targeting properly, you no longer land in the primary inbox. Source · ZoomInfo / Lemlist
- <10% · InMail reply
InMail gets opened, but not answered
InMail open rates are healthy (57.5%) but replies are under 10%. The credit price doesn't pay off: a well-written invitation delivers more per dollar. Source · LinkedIn data 2025
04 · 3 CTAs in a single message
The mistake: "Book a demo", "download the case study" and "reply if you're interested", all in the same message. Every extra ask lowers conversion, because it forces a decision on someone who hasn't even decided whether you're interesting.
The fix: One ask only, and a cheap one to answer, like a question that takes a yes or a no. The demo gets booked in the third message, not the first.
One ask per message, with the rest spread out over time. It's the easiest mistake to fix and the one that rescues the most pipeline.
- −84% · conv.
Without lead scoring, conversion plummets
Teams without scoring lose up to 84% of conversion versus those that prioritize by intent. It's the difference between "calling the first name on the list" and "calling the one who opened 3 emails this week". Source · Aberdeen Group
- 14% → 39% · multi-touch
3-4 touches triple the reply rate
A single touch gets a 14% reply rate on warm accounts. The same list with 3-4 well-cadenced touches reaches 39%+. People aren't rude, they're just swamped on Tuesday. Source · Outreach / Apollo
- 8.5× · meeting rate
Longer sequences convert more
Apollo: multichannel sequences of 16+ touches multiply meeting rate by 8.5× versus 1-3 touches. Patience isn't a virtue, it's math. Source · Apollo benchmarks
- 16.7% · vs. 2.4%
Multistep multiplies CTR by 7
A single touch converts at a 2.4% CTR. Properly cadenced multistep sequences scale to 16.7%. The friction of building a sequence pays for itself by the second touch. Source · Cliently
- 5-10% · elite
Elite cold calling vs. broken cold calling
The average cold call converts at 1.5-3% (lead → meeting). Teams that train their opener, listen and book on the cold call reach 5-10%. More volume without training doesn't break that ceiling. Source · Gong call analytics
- 50% / 27% · email / phone
You're using the wrong channel
50% of B2B buyers prefer email for first contact, 27% prefer phone and the rest is split. If you only call, you're ignoring the majority. If you only write, same thing. Source · HubSpot State of Sales 2025
- 7× · 1st-degree
1st-degree connections convert 7× more
A message to a first-degree connection gets 7× more replies than a cold one. Growing your network doesn't scale with automation, but it does scale with relationships. Automation loses where the network wins. Source · LinkedIn engagement
- Touch 1 · day 0
Main message with a trigger
One idea. One verifiable signal. One closed question. Whoever replies here was already warm, so don't let them go.
- Touch 2 · day 3
Add proof (case study, data point, example)
Recontextualize, don't repeat. It adds replies the first one doesn't bring in. This is where you catch the people who needed to see you're serious before replying.
- Touch 3 · day 7
Switch the pain angle
If the first pain didn't land, offer another specific one. Give two options so the lead doesn't have to think it through. It rescues the ones who didn't see themselves in the first.
- Touch 4 · day 10
37% of replies land here
Lemlist 2025: touch #4 brings in the most absolute replies in a well-built sequence. Most people give up at #2.
- Touches 5-7 · days 14-21
Multichannel: add LinkedIn, voice notes and calls
Mixing email, LinkedIn, voice notes and calls gets more meetings per send than a single channel. Switching channels kills fatigue.
- Breakup · day 21+
Honest close + open door
"Last one from me." No pressure. It converts better than touch 2. Honesty frees up the lead, and sometimes that's exactly what opens the conversation.
05 · The default AI tone
The mistake: The em dash, the "hope you're doing well", the "quick question" and the "just wanted to circle back". People who get them every day spot them a mile away, and end up sending even the ones a human wrote to spam.
The fix: Give the AI your own messages as examples before asking it for anything, and ban those four crutches by name. Read it out loud: if it doesn't sound like you talking, it doesn't go out.
Almost everyone has added AI to their prospecting. Almost nobody has trained it, and the recipient can tell.
- 60% · use AI
Almost everyone uses AI. Almost nobody trains it.
HubSpot 2025: 60% of B2B sales teams use AI for outbound. The vast majority with the default prompt. The result: machine-sounding messages, all with the same em dash. Source · HubSpot State of Sales 2025
- −4× · replies
"AI-mass" gets 4× fewer replies
Bulk-generated messages with no real variable get 4× fewer replies than a human message with a trigger. Token personalization ([Name], [company]) isn't personalization anymore. Source · Lavender benchmarks
Your numbers vs. everyone else's
The five mistakes above show up in this table. If one of your numbers doesn't reach the Average column, you know which of the five to look at first.
| Metric | Bad | Average | Good | Elite | Typical mistakes if you're at the bottom |
|---|---|---|---|---|---|
| Cold email open | <20% | 27-44% | 40%+ | 47%+ | Long or spammy subject, no warm-up, low sender reputation, list poorly matched to the domain |
| Cold email reply | <2% | 3-5% | 5-10% | 10%+ | Deliverability ↓, broken ICP, copy/CTA in the wrong order |
| Bounce rate | >5% | <2% | <1% | <0.5% | Pause campaigns, clean the list, rebuild sender reputation |
| Spam complaint | >0.3% | <0.1% | <0.05% | ~0 | Suspicious subject, stale list, aggressive frequency, no unsubscribe |
| LinkedIn connection | <20% | 25-30% | 45-60% | 60%+ | Pitch in the note, poor targeting, weak profile, no trigger |
| LinkedIn DM reply (1st) | <5% | 10-15% | 20-30% | 30%+ | Generic templates, immediate pitch, no follow-up |
| LinkedIn InMail response | <10% | 18-25% | 25-30% | 30-40% | Too long, no shared context, asking for a meeting before opening the conversation |
| Cold-call connect | <2% | 5.4% | 10%+ | 13%+ | Bad timing, bad list, no triggers, weak voicemails |
| Cold-call → meeting | <1% | 1.5-3% | 5-10% | 10-15%+ | Bad opener, inverted talk ratio, no concrete reason for the call |
| Meetings / emails sent | <1/1000 | 1/344 | 1/100 | 1/40 | The compound metric: targeting + copy + cadence. If the first fails, no copy can save it |
| Reply → meeting | <30% | 30-40% | 40-60% | 60%+ | Vague / multiple CTAs, "thoughts?", asking for too much time in the first step |
| Touches before giving up | 1-2 | 3-4 | 5-8 | 8-12 multichannel | Brevet Group: 80% of closes take 5+ touches. Stopping earlier is giving pipeline away |
Cold email open
- Bad
- <20%
- Average
- 27-44%
- Good
- 40%+
- Elite
- 47%+
- Typical mistakes if you're at the bottom
- Long or spammy subject, no warm-up, low sender reputation, list poorly matched to the domain
Cold email reply
- Bad
- <2%
- Average
- 3-5%
- Good
- 5-10%
- Elite
- 10%+
- Typical mistakes if you're at the bottom
- Deliverability ↓, broken ICP, copy/CTA in the wrong order
Bounce rate
- Bad
- >5%
- Average
- <2%
- Good
- <1%
- Elite
- <0.5%
- Typical mistakes if you're at the bottom
- Pause campaigns, clean the list, rebuild sender reputation
Spam complaint
- Bad
- >0.3%
- Average
- <0.1%
- Good
- <0.05%
- Elite
- ~0
- Typical mistakes if you're at the bottom
- Suspicious subject, stale list, aggressive frequency, no unsubscribe
LinkedIn connection
- Bad
- <20%
- Average
- 25-30%
- Good
- 45-60%
- Elite
- 60%+
- Typical mistakes if you're at the bottom
- Pitch in the note, poor targeting, weak profile, no trigger
LinkedIn DM reply (1st)
- Bad
- <5%
- Average
- 10-15%
- Good
- 20-30%
- Elite
- 30%+
- Typical mistakes if you're at the bottom
- Generic templates, immediate pitch, no follow-up
LinkedIn InMail response
- Bad
- <10%
- Average
- 18-25%
- Good
- 25-30%
- Elite
- 30-40%
- Typical mistakes if you're at the bottom
- Too long, no shared context, asking for a meeting before opening the conversation
Cold-call connect
- Bad
- <2%
- Average
- 5.4%
- Good
- 10%+
- Elite
- 13%+
- Typical mistakes if you're at the bottom
- Bad timing, bad list, no triggers, weak voicemails
Cold-call → meeting
- Bad
- <1%
- Average
- 1.5-3%
- Good
- 5-10%
- Elite
- 10-15%+
- Typical mistakes if you're at the bottom
- Bad opener, inverted talk ratio, no concrete reason for the call
Meetings / emails sent
- Bad
- <1/1000
- Average
- 1/344
- Good
- 1/100
- Elite
- 1/40
- Typical mistakes if you're at the bottom
- The compound metric: targeting + copy + cadence. If the first fails, no copy can save it
Reply → meeting
- Bad
- <30%
- Average
- 30-40%
- Good
- 40-60%
- Elite
- 60%+
- Typical mistakes if you're at the bottom
- Vague / multiple CTAs, "thoughts?", asking for too much time in the first step
Touches before giving up
- Bad
- 1-2
- Average
- 3-4
- Good
- 5-8
- Elite
- 8-12 multichannel
- Typical mistakes if you're at the bottom
- Brevet Group: 80% of closes take 5+ touches. Stopping earlier is giving pipeline away
The prompt that flags them for you
Knowing the 5 mistakes isn't the same as spotting them in your messages. Paste this into Claude along with the ones you've been sending: it gives them back flagged one by one, with each one's mistake and the rewrite right next to it. No paid tools.
I'm sharing prospecting messages I've actually sent. Audit them against these 5 mistakes, and only these 5. One, the unrendered token or the greeting that gives away the template. Two, the first line that talks about me and not about them. Three, the empty personalization line that doesn't prove I've read anything of theirs. Four, more than one ask in the same message. Five, the default AI tone, that cadence with no person behind it. For each message, give me back three things and nothing else. First, which of the 5 mistakes it makes, by number, quoting the exact sentence where it makes them. Second, which of them is costing it the reply the most and why. Third, the message rewritten without those mistakes, in my words and my voice, not yours. Two rules. If a message doesn't make any of the 5, say so and don't invent a flaw just to have something to say. And if, when rewriting it, you're missing information only I have, leave the gap marked instead of filling it in yourself. My messages:
The rest of the resource opens when you leave your details.
Other resources you might like
A résumé doesn't sell. Your profile, as a landing page
As a résumé, it attracts work-anniversary congrats. As a landing page, it attracts clients. We give you the 7 pieces to convert it and analyze your profile to tell you what to change.
The 8 B2B signals with their real multiplier
The difference between a cold list and a booked meeting isn't the copy. It's when you write. Here's 1 full signal with its window; the other 7 are in the playbook.
I handed the keys to my LinkedIn over to Claude
The complete guide to B2B posts: the 3 prompts I use with Claude, 9 hook archetypes, 5 structures, the 24-check checklist and the rule of ONE.