I’ve been struggling with something that feels like a paradox: I want to automate my follow-up sequences to scale my outreach, but every automated follow-up I’ve tested sounds like a complete robot. And when it sounds robotic, people ghost harder.
So I tried the opposite—manually writing follow-ups for everyone. That worked in terms of tone, but it was eating up literally hours every single day. Not sustainable at all.
Then I started using LiSeller’s automated follow-ups but spent real time configuring the AI prompts properly. And I realized the problem wasn’t automation itself—the problem was how I was instructing the AI to write the follow-ups.
Here’s what changed things for me:
-
I stopped using templates. Every follow-up gets generated fresh based on the prospect’s reply (or lack thereof). If they replied with a question, the follow-up addresses it. If they ghosted, the follow-up is different—it’s a soft re-engagement, not a pushier version of the first message.
-
I inject personality through details. Instead of generic “hey, just following up!” I’m including specific references to what they said (or didn’t say), or something about their company that’s actually relevant to why I’m persisting.
-
I vary the length and tone across the sequence. Message 1 is short and curious. Message 2 (if they ghost) is slightly longer and offers more context. Message 3 is almost like a mini-case study. Message 4 is a soft exit. This keeps prospects from recognizing the pattern.
The weird part? My reply rates on follow-ups went up substantially. Not because I’m being more aggressive—because the automation is actually mimicking real human follow-up behavior, which means people respond to it like it’s real.
But I’m still tweaking it. Has anyone else figured out how to keep follow-ups feeling natural when you’re running 5-10 of them in sequence? Like, what’s the threshold before it starts feeling like harassing?
You’ve cracked something important: the rhythm matters as much as the message. Let me add a copywriting angle: each follow-up should have a different reason to open it.
Message 1: curiosity hook (“I noticed…”)
Message 2: scarcity or FOMO (“Quick question: are you guys…”)
Message 3: social proof or case study (“Similar company just achieved…”)
Message 4: soft urgency (“I’m ramping down my outreach to focus on…”)
If every follow-up is the same hook rephrased, people recognize it as a sequence and delete it. But if each one has a different psychological trigger, it feels like separate conversations with the same person.
Human follow-ups do this naturally. Your automation should too.
One more thing: stop after 4 touches max if they haven’t engaged. Anything beyond that looks desperate. And if they reply but aren’t ready yet? That’s a different sequence entirely—nurture track, not “close em now” track.
I’ve automated this beautifully by building conditional logic into LiSeller. Here’s the structure:
If prospect replies → pull their reply into a custom field → generate follow-up based on what they said → send it 2 days later.
If prospect doesn’t reply → send generic follow-up variation A at day 3, variation B at day 7, variation C at day 14.
Each variation is different (different angle, different length, different ask). And I’m pulling this directly into my Pipedrive CRM so the sales team sees the whole conversation thread.
The key is conditional logic. Don’t just send the same automation to everyone. Route it based on behavior. That’s what makes it feel human—real humans change their approach based on how someone responds.
In recruiting, follow-up fatigue is real—high-level talent will block you if you seem desperate. What keeps my sequences feeling human is genuinely backing off if someone says they’re not interested right now.
Like, if they reply “not looking at the moment,” my next message (weeks later) is just “Hey, I respect that. If things change, my door’s open.” Not “Are you sure?” Not “Mind if I check in in 3 months?” Just respect their answer.
Then maybe one very light touch point 2-3 months later if something’s changed on their profile (job change, new role, etc), but it’s contextualized by that change, not by my original ask.
People respond to follow-ups that feel like humans recognizing their answer, not ignoring it.
The threshold before it feels like harassment? It’s when you stop listening to what they’re actually saying. So make your automation respond to their inputs (or lack thereof), not just blast through a predetermined script.
From a safety perspective, be careful here. LinkedIn flags aggressive follow-up sequences, especially if they’re too frequent or too many touches. I’d cap it at 3-4 total touches (initial + 3 follow-ups) before moving on. And definitely space them out—minimum 3 days between touches, ideally 5-7.
If your automation is sending follow-ups too fast or too many times, LinkedIn’s algorithm will notice the pattern and either de-prioritize your messages or flag your account for “unusual activity.”
Make your automation feel human by including real delays between messages, not just “automated at X time.”
Also—and this is important—vary the sending time of your follow-ups. Don’t send all your day-3 follow-ups at 9 AM. Real people send messages at random times. Automation that looks mechanical gets filtered harder.
I’m running a 4-touch sequence, and here’s what actually works for me:
Touch 1: Problem-aware hook (“I help companies like yours solve X”)
Touch 2 (day 4): Social proof + slight reframe (“Latest case study shows Y improvement”)
Touch 3 (day 10): Soft urgency + different angle (“Noticed you got 15 new hires—wondering if you’re…”)
Touch 4 (day 18): Soft exit (“Seems like timing isn’t right, but keep me in mind if things shift”)
My reply rate on touch 3 is actually highest because it’s novel—different premise than touches 1-2, so if they didn’t bite before, this might trigger it.
I’m not hammering them. I’m giving them multiple reasons to care across the sequence. And the variation makes each message feel distinct, not like a reminder of the same message.
What’s your current spacing? Like, are you doing day 3, 7, 14, 21? Or something else? Because spacing definitely impacts whether it feels persistent vs. annoying.
This is a great use case for LiSeller’s AI configuration. The key is in how you prompt the AI for each follow-up. Here’s what I’d recommend:
For each follow-up in your sequence, write a specific prompt that explains the context and the goal for that specific touch. Don’t just say “write a follow-up.” Say:
Follow-up 1: “They haven’t replied. Write a brief message that reframes your value prop from a different angle.”
Follow-up 2: “Add credibility with a quick case study or example. Keep it short.”
Follow-up 3: “This is the soft exit. Be respectful, acknowledge they might not be interested, but leave the door open.”
When the AI understands why it’s writing each message, the tone naturally varies, and it stops sounding like a repetitive script.
Also, enable the “reply-aware” feature if your sequence has it. That way, if someone does reply before the automated follow-up fires, the system cancels the automation and switches to manual/custom routing. That prevents the robotic “here’s touch 3” sending after they already engaged.
The psychology here is important: people don’t get annoyed by follow-ups per se. They get annoyed by irrelevant follow-ups. So each touch needs to be contextually different.
Here’s what I recommend:
Touch 1: Problem + introduction (“I help X companies solve Y. Wondering if that’s relevant for you?”)
Touch 2: Social proof (“Recent data shows Z is working for [similar] companies”)
Touch 3: Different angle (“Following up because I know most teams are dealing with ABC right now”)
Touch 4: Soft disqualification (“Seems like maybe not the right time. I’ll stop here, but here’s something that might help regardless”)
Notice: each one is a different conversation, not a reminder of the first one. That’s how you keep it human.
The threshold before it feels like harassment is usually 4-5 touches. Much beyond that without a response, and you’re just objectively wasting their time.
What’s your current reply rate curve across your sequence? Most campaigns see highest reply on touches 2-3, then drop-off after. If you’re seeing high replies on touch 4+, you might actually be undersending earlier.