I spent the first six months of my startup selling to anyone who would take a meeting. SaaS companies, agencies, e-commerce brands, a logistics firm, two nonprofits. My pipeline looked healthy. My close rate was 8%.
The problem wasn't the product. It was that my ideal customer profile was a fiction — broad enough to include everyone, which meant it actually described no one. If your ICP reads 'B2B SaaS companies between 10 and 500 employees,' you don't have an ICP. You have a NAICS code.
Here's how to fix it.
What 'Too Broad' Actually Looks Like
Most founders define their ICP at the firmographic level and stop there. Company size, industry, geography. That's a starting point, not a profile. A profile answers the question: 'Of all the companies that match these firmographics, which ones are the ones that desperately need what we sell right now?'
Broad ICP: 'Mid-market SaaS companies in North America.'
Tight ICP: 'B2B SaaS companies with 50–200 employees, $3M–$15M ARR, selling into enterprise accounts, running on Salesforce + HubSpot, with a VP of Revenue hired in the last 12 months who is under pressure to hit pipeline targets.'
The second one tells a story. You can picture the person picking up the phone. You know what they're worried about. That specificity is what turns outreach from noise into signal.
The Real Cost of a Vague ICP
When your ICP is too broad, every downstream decision gets worse. Your messaging becomes generic because it has to speak to too many different contexts. Your sales cycle lengthens because you spend time educating buyers who were never a fit. Your product roadmap fragments because you're taking feature requests from wildly different customer segments. Your churn goes up because the customers you acquired don't get enough value.
I've seen teams where salespeople spend 60–70% of their time on prospects that never had a realistic chance of closing — simply because no one drew a clear line around who to pursue. That's not a sales performance problem. That's an ICP problem wearing a sales costume.
The 5-Dimension Framework for Narrowing Your ICP
Most ICPs only cover firmographics. A useful ICP covers five dimensions:
1. Firmographics (the basics): industry vertical, company size (headcount and revenue), geography, business model (B2B vs B2C), and sales motion (SMB vs mid-market vs enterprise). These narrow the universe but don't close it.
2. Technographics: What tools are they already running? If your product integrates deeply with Salesforce, companies on HubSpot-only stacks are a harder sell. Knowing the tech stack isn't optional — it predicts both fit and implementation friction before you ever get on a call.
3. Behavioral signals: Are they growing? Hiring? Raising capital? Expanding into new markets? Behavior is a proxy for urgency. A company in growth mode has budget pressure and a mandate to move. A company in cost-cutting mode has a very different buying posture, and most B2B tools are a poor fit for them right now.
4. Pain specificity: Not just 'they have the problem we solve,' but 'they have this specific manifestation of the problem, at this severity level, with this failed previous attempt to fix it.' Customers who have already tried and failed to solve the problem are your best buyers. They've validated the problem is real, they've burned time on inadequate solutions, and they're motivated to get it right.
5. Economic buyer profile: Who signs the check? A 200-person company where the VP of Engineering owns the budget is a completely different sales motion from one where the CFO does. Get this wrong and you'll spend weeks selling to a champion who can't actually say yes.
Use Trigger Events, Not Just Profiles
The highest-converting addition you can make to any ICP is a trigger event — a specific, observable thing that happened recently to make this account likely to buy right now.
Trigger events I've seen work reliably for B2B SaaS: a recent Series A or Series B raise (new money, new pressure to grow), a new VP of Sales or Marketing hired in the last 90 days (they need to show results fast), an acquisition (tech stack consolidation creates purchasing decisions), a competitor's product being discontinued (forced migration), or three or more open job postings for a function your product serves (they're scaling that team, which means they feel the pain).
When you layer a trigger event onto your ICP, you're not just targeting a profile — you're targeting a moment. The difference in reply rates is not small. Outreach to accounts showing buying signals converts 2–3x better than identical outreach to accounts that match the firmographic profile but show no active buying motion.
How to Derive Your ICP From Data, Not Intuition
The ICP mistake I see most often from early-stage founders is that they write it once during a pitch deck session, based on gut feel, and never revisit it. Your ICP should be derived from your closed-won data — the customers who actually paid you, stayed, and got value.
If you have 10 or more paying customers, run this exercise: list your top five (most revenue, best retention, most enthusiastic) and your bottom five (most churned, most support tickets, lowest NPS). Write down every attribute you can observe — size, industry, tech stack, the role that championed the purchase, the trigger that started the conversation. Compare the two lists. The pattern in the top five is your real ICP. The pattern in the bottom five is who you should stop pursuing.
If you have fewer than 10 customers, make up the gap with prospect interviews. Talk to 10 people who said no and understand why. Talk to 10 who said yes and understand exactly what made them buy. This interview data is more valuable than any market research report.
The FOMO Trap (And Why Narrowing Actually Grows Revenue)
The reason founders resist narrowing their ICP is fear. Fear that if you only target Series A SaaS companies with 50–150 employees in North America using Salesforce and HubSpot, the market will be too small. It's almost never actually too small.
There are roughly 15,000 SaaS companies in North America that have raised a Series A. If even 10% match your tighter profile, that's 1,500 accounts. You don't need to close all 1,500 to build a $5M ARR business. You need to close about 50–100 of them, and to do that, you need messaging and targeting so specific that when the right buyer reads your outreach, it feels like you wrote it just for them.
Tight targeting doesn't shrink your market. It concentrates your energy where the probability of winning is highest. You can always expand later — once you've dominated your initial wedge and have the revenue and case studies to open the next segment.
When to Expand Your ICP
Expand when you've hit one of three signals: you're closing more than 30% of qualified pipeline from your current ICP (which means you've saturated easy wins), you've hit $1M ARR and have 15+ strong reference customers, or inbound demand is coming consistently from a segment you haven't targeted that converts as well as your core ICP.
Don't expand out of boredom or board pressure. Expand when the data tells you the wedge is won and the adjacent market is already showing up in your pipeline.
The One Thing to Do This Week
Pull your last 20 closed-won deals. Write down every observable attribute for each. Find the three attributes that appear most consistently in your best customers. Make those the center of your ICP and spend the next 30 days exclusively targeting accounts that match all three.
If you don't have 20 closed-won deals, start with the 5 customers who are happiest with your product and work backwards. They are your ICP. Everything else is hypothesis.