I just spent about four hours building out what I thought was a really dialed-in ICP. Job titles (3 different ones), company size (50–500 employees), industry (tech and SaaS), location (US only), recent hiring activity, and funding stage.
Then I opened up LiSeller’s lead filtering, and I realized I could dial this in even more—but I’m not sure if I should.
Here’s my thinking: the broader I make my filters, the more leads I get to message, which means more volume to test with. The narrower I go, the fewer leads, but theoretically they’re way more likely to convert because they actually match my ICP perfectly.
But there’s a middle ground issue: I don’t want to waste time building a hyper-specific filter only to realize I’m filtering out tons of actual good-fit prospects. On the flip side, I don’t want to spray generic outreach at 2,000 people who are “kind of close” to my target.
I’m currently running one campaign with fairly broad filters (just to get volume) and it’s converting at about 1.5%. My instinct tells me that if I filter smarter before I even send the first message, I could probably double that.
But here’s what I’m really asking: at what point does adding more filter criteria start hurting your conversion rate instead of helping? Is there a sweet spot where “good enough” filters beat “obsessively specific” filters?
This is honestly backwards from how most people think about it. You don’t filter before you personalize—you filter to enable better personalization.
Here’s the psychology: a hyper-specific filter means you have enough intel to write a killer hook. A 1.5% conversion rate usually means your hooks are weak because you’re messaging too broad an audience and there’s no one insight that lands.
Build your filter around criteria that give the AI something concrete to reference. Did they recently hire? Launched a feature? Shifted their messaging? Those filter criteria should generate intel that feeds your personalization.
So the answer: as specific as you can go while still having enough prospect pool to test message variants. If your filter is so tight you only get 50 prospects, that’s probably too tight. If you get 500+ and conversion stays 1.5%, your filter is too broad.
Test this way: build three filter variants—broad, medium, tight. Send 100 messages from each with identical personalization. If the tight one converts at 3–4% and the broad at 1.5%, then yes, filtering was your problem. If they’re all the same, your problem is the copy, not the filtering.
From a workflow standpoint, here’s what I do:
I build my initial filter with what I call “hard filters” (the drop-dead criteria—company size, industry, geography) and then set up separate data enrichment layers in Google Sheets that pull in the “soft” criteria (recent funding, hiring, tech stack changes) via APIs.
That way, LiSeller’s filtering focuses on quantity and fit, and then my secondary layer in Sheets scores and ranks. So I’m not limiting upfront, but I’m channeling the best prospects into my first message variants.
The sweet spot for me: 300–500 ranked leads per campaign beats 2,000 unqualified leads every time. And technically, you’re not filtering out prospects—you’re just messaging the best ones first.
Pro tip: use LiSeller’s API to export your filtered list before you message, then run that list through a second scoring layer (Pipedrive, HubSpot, or even a simple Zapier setup). It transforms filtering from binary (yes/no) to ranked (70 points, 55 points, 40 points). Way more flexible.
In recruiting, I’ve learned that the “sweet spot” filter depends entirely on two things: conversion goal and message strategy.
If your goal is to get high-intent conversations, narrow filters beat broad ones—always. You’re messaging people who need what you offer, so your copy can be more direct.
If your goal is to gather data (like A/B testing messaging), broader filters work fine because you’re not optimizing for conversion—you’re optimizing for insights.
For my recruitment campaigns: I use core filters (active job search, right seniority, right industry), but I never filter on company size or other soft criteria. Those vary too much. Hard criteria → narrow; soft criteria → test across variants.
Does that distinction make sense for your use case?
Here’s the safety angle nobody talks about: too broad a filter with high outreach volume flags your account faster than a narrow filter does.
LinkedIn detects spray-and-pray behavior. If you’re messaging 2,000 loosely targeted people, your engagement patterns look spammy. If you’re messaging 300 highly targeted people, your conversion rate and reply rate stay higher, which protects your account health.
So from a risk standpoint: narrow filters are safer. Your engagement looks intentional instead of automated. You hit daily connection limits more slowly. Your reply rates stay healthier, so the algorithm doesn’t ding you.
The sweet spot: specific enough that your reply rate stays above 2–3%, loose enough that you have enough prospects to test. That’s your safety zone.
Also: if you’re using a proxy or warming up an account, start with narrow filters. Prove you can get replies from a small, targeted pool before you scale to broader outreach. It’s the same principle as account warmup—prove intent before scale.
Also, the platform can help: use LiSeller’s filter preview to see what percentage of your addressable market each criterion removes. If adding one criterion cuts your audience by 75%, that’s a signal it’s either too strict or poorly calibrated to your actual ICP.
Strategically, here’s the principle: filter to the level where you have enough context to personalize, but enough volume to test.
Your 1.5% conversion rate suggests one of two things: either your targeting is too broad (you’re messaging people who aren’t really a fit) or your personalization is weak (you’re targeting right but your hook is generic).
If it’s the first problem, narrow filters solve it. If it’s the second, narrow filters won’t help—better copy will.
Running the diagnostic: send 50 messages to your broadest filter segment with really sharp, personalized hooks. Send 50 to your narrowest segment with generic hooks. If narrow + generic beats broad + sharp, filtering is your lever. If the opposite, copy is your problem.
Deal-breaker filter criteria: anything that disqualifies someone from actually being interested in your offer (wrong company size, wrong industry, wrong seniority). Nice-to-haves: anything that improves your intel for personalization (recent funding, team growth, product changes).
One more framework: always build your filter around intent signals first (what indicates they need your solution), then fit signals (what indicates they’re your ideal customer). That priority usually corrects for over-filtering.