Stop Using Firmographics. Vectorize the Problem.

"We sell to manufacturing companies with 100-1000 employees."

That's 50,000 companies in Europe. Your sales team sends 1,000 emails and gets 10 responses. They're treating food packaging companies the same as windmill manufacturers, treating them all as if they have the same problem just because they fall into the same demographic box.

Firmographics tell you who could theoretically buy from you. They don't tell you who needs to buy from you right now.

The Problem with Demographic Filtering

Most B2B SaaS companies define their ICP with filters like industry, company size, location, and revenue. You run these through ZoomInfo and get 50,000 companies. Then what?

Your SDR burns through the list at 1-2% response rates. Half of these companies don't actually have your problem. A quarter already solved it. The rest aren't actively looking for a solution.

The issue is that you're filtering by company characteristics instead of problem intensity. Firmographics answer "who fits a demographic profile?" but the real question is "who has this problem acutely enough to pay us to solve it?"

Start with the Problem

Before you think about company characteristics, define the problem you solve. Not your product features. The actual problem. The manual process they're doing now. The risk they're managing. The inefficiency they're living with.

Then ask yourself: what makes this problem harder?

Don't ask "what industries have this problem?" or "what company sizes?" Just focus on the variables that intensify this specific problem.

Finding Problem Vectors

A vector is a variable that makes the problem harder or easier. It's a spectrum, not a binary yes/no.

Let's say you're building supply chain planning software. The core problem you solve is coordinating supply and demand across distributed networks. What makes this problem harder?

Scale is one vector. More locations in the network make coordination harder. Five warehouses are manageable. Fifty warehouses become exponentially harder to coordinate. Five hundred warehouses are impossible to manage manually.

Complexity is another vector. Product variety matters. Ten standard products are simple to plan for. One hundred products with customization options are harder to forecast. One thousand SKUs with custom configurations mean manual planning breaks down entirely.

Volatility is a third vector. How often does demand change? Annual contracts mean stable demand and easy planning. Seasonal fluctuations create moderate challenges. Weekly demand changes mean your forecasts become obsolete fast.

Notice that these three vectors came from analyzing this specific problem, not from applying a generic framework.

How to Find Vectors for Your Problem

Start by being specific about what problem you solve. "We help with sales" is too vague. "We automate the 60 minutes of manual research SDRs do per account before outreach" is specific enough to work with.

Then run through scenarios. When is this problem worse? More accounts to research makes it harder. More data sources to check makes it harder. More customization needed per account makes it harder. Faster turnaround time required makes it harder.

Next, find the boundaries using inversion. When does the problem not exist at all? If you're only researching three accounts per day, manual research works fine. If it's a simple transactional sale, no research is needed.

When does the problem become unsolvable for you? Maybe you can't handle 10,000 accounts per day because you'd need an enterprise data warehouse. Maybe you can't handle real-time trading decisions because they're too fast for your workflow.

Finally, check if your vectors are actually independent. Employee count and revenue usually move together, so they're measuring the same thing. Pick one. Scale and volatility are separate dimensions, so they're both useful.

Vectors Multiply

Here's why this matters. Imagine two companies, both with 500 warehouses.

Company A has 500 warehouses, 2,000 SKUs with custom configurations, and weekly demand shifts. All three vectors are high. This is a very hard problem.

Company B has 500 warehouses, 50 standard products, and annual contracts. Only one vector is high. This is a medium problem.

Same warehouse count, but Company A's problem is ten times harder because the vectors multiply together.

This is exactly why demographic filtering fails. Two companies can have identical firmographics but completely different problem intensity.

Making Vectors Observable

You need to be able to detect these vectors at scale, or they're not useful.

Some things are easy to observe from public data. Number of locations comes from LinkedIn or the company website. Product catalog size is on their website. Industry vertical comes from standard data enrichment.

Other vectors require inference. You can proxy volatility by checking if they're in consumer goods versus industrial B2B. Technical maturity shows up in job postings when they mention ERP systems versus Excel. Process complexity appears when they're hiring specialized roles.

If a vector is critical but you genuinely can't observe it, find a proxy. Volatility is hard to measure directly, but industry vertical works as a decent proxy since consumer goods generally means high volatility and industrial contracts mean low volatility.

Mapping Your Solution to the Vectors

Once you have your vectors, you need to map where your solution actually creates value.

For each vector, identify three points. The lower bound is where the problem doesn't exist yet. For scale, maybe that's fewer than ten nodes. For complexity, maybe fewer than ten products.

The sweet spot is where the problem is hard enough that they'll pay to solve it, but you can still solve it effectively. For scale, maybe that's 50 to 500 nodes. For complexity, maybe 100 to 2,000 SKUs.

The upper bound is where the problem becomes too hard for your solution. Maybe more than 5,000 nodes means they need SAP or a custom enterprise system. Maybe aerospace-level customization is a completely different domain.

Your ICP lives in the sweet spot across all your vectors.

Why This Beats Demographics

Consider two companies with identical firmographics. Both are in manufacturing, both have 500 employees, both generate €50M in revenue.

The first is a food packaging company. They have 50 distribution centers, 200 SKUs that are mostly standard, and high volatility because retail demand changes weekly. This is a strong fit for supply chain planning software.

The second is an industrial valve manufacturer. They have five production facilities, 2,000 highly custom products, and low volatility because they work on multi-year contracts. The scale is too low and the complexity is about engineering rather than planning. Wrong fit.

Same demographics. Completely different problem intensity.

Building a Scoring System

Start by weighting your vectors based on what actually predicts problem intensity. For supply chain planning, maybe volatility is 40% of the problem, complexity is 35%, and scale is 25%. Test this against your existing customers to see which vectors your best customers score highest on.

Create a scale for each vector. For scale, maybe 0-2 means fewer than ten nodes where no problem exists, 3-5 means 10-50 nodes where the problem is emerging, 6-8 means 50-500 nodes which is your sweet spot, and 9-10 means more than 500 nodes which is too complex for you.

Calculate a composite score by multiplying each vector score by its weight. A company with volatility at 8/10, complexity at 7/10, and scale at 6/10 would score 3.2 + 2.45 + 1.5 = 7.15 out of 10.

Create tiers based on these scores. Tier 1 might be 8-10 for immediate outreach. Tier 2 might be 6-7.9 for outreach within 48 hours. Tier 3 might be 4-5.9 for weekly nurture. Tier 4 below 4 gets excluded.

Making This Operational

In week one, build the infrastructure to detect your vectors. Set up data collection for scale indicators. Analyze public signals for complexity. Use enrichment tools for proxy indicators.

In week two, automate the scoring. Score each company on each vector using your 0-10 scale. Apply the weights you determined. Calculate the composite score. Assign them to a tier. This should run automatically, not manually.

In week three, implement different treatment for each tier. Tier 1 gets immediate outreach with custom messaging that references their specific vector challenges. Tier 2 gets outreach within 48 hours using templates with some personalization. Tier 3 gets weekly educational content. Tier 4 gets excluded.

In week four, validate everything. Check if Tier 1 companies actually convert better than Tier 2. See which vector best predicts closed deals. Adjust your weights if needed.

Continuous Refinement

Your vectors will evolve as you learn more about your market.

Every quarter, review whether these vectors still predict success. Check if your weights are still right. Look for new vectors emerging that you didn't see initially.

Track your confidence levels explicitly. Note whether something is based on 5 customers or 50 customers over 12 months. State your assumptions clearly. If you're assuming industry vertical proxies for volatility, write that down and then test it.

The Results

Before vectorization, you have 10,000 companies in your ICP, 1-2% response rates, generic messaging, and random coverage of your total addressable market.

After vectorization, you have the same 10,000 companies but now scored 0-10. You focus on the top 500. Response rates improve to 3-5% because you're targeting companies with acute problems. Your messaging references specific vectors like their volatility challenges or complexity pain. You cover your market efficiently by prioritizing highest intensity first.

We've built this approach for supply chain SaaS, workforce planning tools, ESG software, and development agencies. The pattern holds: companies that vectorize their problems outperform companies that spray-and-pray based on firmographics.

TL;DR

Your ICP is not "companies that look like this demographically." Your ICP is "companies where this specific problem is painful enough that they'll pay us to solve it."

Demographics describe companies. Vectors describe problems.

Find what makes your problem harder. Map where you can create value on each spectrum. Build systems to score companies at scale. Focus your efforts there.