I’ve been using LiSeller for about three weeks now, and I’m seeing decent open rates, but the reply rates are honestly disappointing. I thought the whole point of hyper-personalized AI messaging was that it would feel actually human, but when I read back through what got sent, a lot of it still has that AI tone—you know, the kind where it’s technically personalized but it doesn’t really sound like me.
I’ve been testing different approaches. First, I just let the AI generate everything from scratch. Then I started adding more context about the prospect and their company to the prompt. But here’s the thing—I’m wondering if the problem is that I’m not giving the AI enough real “me” to work with. Like, maybe the personalization is only surface-level because I haven’t actually trained it on how I actually talk.
I’ve also noticed that when I manually tweak a couple of these AI-generated messages, they do perform better. The hook feels sharper, the tone matches what I’d actually say. So either I’m not using the personalization feature correctly, or there’s a gap between what “personalized” means in the product and what actually converts.
How are you all structuring your prompts or initial setup to make sure the AI is capturing your actual voice? Is there a technique I’m missing, or is this just the nature of AI-generated outreach no matter what tool you use?
This is exactly the problem I see with a lot of AI tools. They personalize the details—name, company, recent news—but they never personalize the voice. That’s where the real conversion lives.
Here’s what changed everything for me: instead of just feeding the AI prospect data, I started including a brief description of my actual sales philosophy. Like, I wrote out: “I’m direct, I lead with curiosity, and I hate fluffy language. I ask one real question at a time.” Then when the AI generates, it uses that as the baseline tone.
Second thing—the hook is never going to feel like you unless you write at least the first line yourself. The AI can personalize the body and the ask, but that opening needs to come from your perspective. Something unexpected that makes the person sit up and read instead of scroll past.
What’s your typical opening line look like right now?
I actually solved this by building a simple system. I use LiSeller to generate the bulk of the message, but I pipe it through a Zapier workflow that flags messages with certain keywords—the AI tends to overuse phrases like “thought-provoking” or “mutual benefit”—and it prompts me to manually review high-value prospects before sending.
It’s an extra step, but you’re only actually reviewing maybe 20% of outreach if you set the rules right. The rest go out untouched. It’s like quality control on the front end instead of analyzing dead campaigns later.
You could also set up a Google Sheet that logs each message variant you send, so you can actually A/B test which AI-generated hooks your audience responds to best. Then the AI learns from that data over time.
I’ve been managing this by thinking about it differently. The personalization isn’t supposed to replace your voice—it’s supposed to replace the generic template research and basic context-pulling. You’re still the voice.
What I do is use the AI to handle the “thank you for X achievement” part and the “I noticed Y about your company” part. But the actual ask and the reason I’m reaching out? That comes from me. It’s more hybrid than fully automated, but it feels genuine because the core of it is.
With technical talent especially, people can smell when the entire message is AI-generated. They’ve seen it a thousand times. But when the personalization is clearly researched and the ask is clearly thoughtful? That’s when you get the meeting.
Maybe it’s worth separating: what do you actually need the AI to do, versus what do you need to keep as your own fingerprint?
One thing I’d also mention—sometimes the “generic” feeling you’re getting is actually a safety mechanism. If the AI gets too creative or uses language that’s way too unique, it can actually trigger spam filters more easily. LinkedIn’s algorithm looks for patterns, and something that’s weirdly personalized in a dozen different ways can actually flag as suspicious.
So there’s a balance. You want personalized enough to feel real, but not so wild that it looks like prompt injection. Make sure you’re testing your deliverability rates too. Sometimes “generic feeling” actually means “delivering reliably.”
What does your open rate look like versus your reply rate? That might tell you if it’s a deliverability issue masquerading as a voice issue.
Dude, I had the exact same frustration. Here’s what worked for me: I stopped trying to make the AI perfect and started using it as a first draft instead. Like, I treat it as content scaffolding. The AI does the research and context-matching, I do the actual selling.
I set up my LiSeller templates to be like 60% done. It gets the prospect details in there, opens a conversation, but I actually review-and-edit the last 30% before it sends. Takes maybe 30 seconds per message. Conversion rate went up like 40% compared to fully automated.
The real win is that I’m not spending hours researching—the AI does that—but I’m still putting my personality on it. Best of both worlds.
How many messages are you sending per day? If it’s a huge volume, maybe the hybrid approach isn’t practical for you.
Great question, and honestly this is one of the most common feedback points we hear. The key thing to understand is that the AI is matching patterns based on what you feed it. If you’re just giving it basic prospect data (name, company, title), that’s all it has to work with.
Here’s what power users do: they create a detailed “voice guide” in their settings. It’s basically a character description—how you actually pitch, what language you use, what kinds of questions you ask. You can also include past messages that did get replies and tell the AI “this is the tone I’m looking for.”
Also make sure you’re using the “tone adjustment” slider. A lot of people don’t realize it’s there. You can dial it toward “conversational” instead of “professional” if that’s more your style.
Have you set up a voice guide yet, or are you just using the default AI settings?
From a conversion standpoint, this is a real concern. Research shows that prospects can sense authenticity within about two seconds of reading. If the AI voice doesn’t match your actual communication style, it creates cognitive dissonance—they might respond to the message, but they’ll be skeptical when you actually talk.
The strategic play here is to think of personalization in layers: Layer 1 is prospect research (AI does this). Layer 2 is tone and approach (you define this). Layer 3 is the actual personalization (AI applies it within the framework you set).
The problem I see is when people use the tool as a complete automation—just point and shoot. That’ll get you scale, but not conversion. When you’re intentional about your voice framework first, then use the personalization layer, that’s when the magic happens.
What’s your current conversion rate, and how many touches are you typically getting before a reply?