I’ve been running campaigns on LiSeller for about three weeks now, and I keep hitting this weird wall. My conversion rate is stuck around 2-3%, and I’m pretty sure it’s because my personalized messages still feel… off. Like, they’re technically personalized—they mention the person’s recent job change or a company achievement—but the writing itself still screams ‘AI wrote this’.
I’ve been experimenting with different message templates and experimenting with the tone settings, but I’m wondering if I’m even approaching this right. Some messages I write come across as overly casual and forced. Others are too formal and stiff. I’ve noticed that when I keep things conversational—almost like I’m texting a colleague—I get better responses, but scaling that without sounding repetitive or robotic feels impossible.
The hook is usually something specific about their company or role, but maybe that’s not enough. I’m also wondering if the follow-up sequences are contributing to the problem. If my first message feels off, are the follow-ups making it worse by maintaining that same awkward tone?
Has anyone else struggled with this? What’s your actual process for balancing personalization depth with keeping the tone natural and human? Are you tweaking the AI prompts themselves, or is it more about hand-editing the output before it goes out?
This is my bread and butter. The problem isn’t your personalization—it’s your hook and your offer clarity. Most people bury the why they’re reaching out under layers of flattery. Try this: lead with ONE specific observation (not generic praise), then immediately pivot to the ask or value prop. Get to the point in 2-3 sentences max.
Example: ‘Saw your product just hit 10k users—congrats. I work with SaaS founders on scaling their GTM. Might be useful if you’re thinking about scaling next quarter. Open to a 15-min chat?’
That’s conversational, specific, and clear. The AI can help you personalize the observation, but YOU need to write the hook and offer. Don’t let the tool do all the thinking.
Pro tip: pull your LiSeller campaign data and pipe the message variants into a spreadsheet. Tag each one with ‘conversational’, ‘formal’, ‘casual’, etc. Then cross-reference with reply rate data in your CRM. You’ll see patterns.
Also, consider A/B testing the tone within LiSeller’s prompt settings. If you’re using the AI to generate messages, you can feed it instructions like ‘write like you’re texting a sales peer, not a prospect’ vs. ‘professional but warm’. The variance might surprise you. Then automate the winner into your sequences.
In technical recruiting, I’ve found that the best outreach mirrors how the candidate actually communicates. If they’re active on Twitter and use casual language, my tone shifts. If they’re LinkedIn-formal, I stay professional. The personalization piece isn’t just the info—it’s the voice.
One thing I’ve noticed: when I reference something they actually said or built (not just their job title), the response jumps significantly. It signals I actually know who they are, not just that I scraped their profile. That human touch makes the rest of the message feel less robotic, even if it’s AI-generated.
Good instinct flagging this. Account safety-wise, the more your messages look personalized to a human reviewer, the better your deliverability. Generic-feeling AI messages are also more likely to trigger spam filters, ironically. So fixing the tone issue also protects your account health.
Stay below 200 characters per sentence, keep slang minimal, and avoid ALL-CAPS or excessive punctuation. LinkedIn’s filters pick up on patterns, so if your bot detector sees too many identical structures across accounts, it flags you. Keep each message genuinely unique.
I had this same problem last month. Here’s what worked: stopped overthinking the AI output and started treating it like a first draft. I spend 30 seconds editing the hook, rewriting the ask, and maybe adjusting one or two phrases. Takes me longer upfront, but my reply rate jumped from 2.5% to 6% almost immediately.
Also deleted about 40% of my templates because I realized I had too many variations and no clear winner. Focusing on 3-4 high-performing hooks and just swapping in personalization details had way better results. Less is more, honestly.
Great question. So in LiSeller, when you’re setting up your message prompt, try framing it this way: ‘Write like you’re a knowledgeable peer who just noticed something cool about their company. Keep it under 150 words. Use casual language but stay professional.’ The specificity matters.
One feature we see people miss: the tone slider in advanced prompts. Sliding it toward ‘conversational’ without sliding all the way to ‘casual’ gets you that sweet spot. Also, longer does NOT mean better—constraint actually forces clarity. Try capping at 100 words and see if your engagement shifts.
From a sales psychology angle, you’re experiencing what we call ‘uncanny valley’ messaging. It’s personalized enough to feel targeted but generic enough to feel automated. The gap is killing your credibility.
Data shows conversational tone consistently beats formal by 30-40% in B2B cold outreach. But ‘conversational’ doesn’t mean casual slang—it means active voice, short sentences, and treating them like a peer, not a prospect. ‘Helping companies like yours…’ = instant turned off. ‘We’ve seen patterns in your space…’ = respectful and real.