Firmographic segmentation means grouping target companies by attributes such as industry, size, revenue, location and growth stage, then treating each group differently in messaging, offers and sales motion. Every B2B team does some version of it. Most do it badly: the boundaries come from a data provider's default dropdown ("51-200 employees") instead of from the team's own wins and losses.
This post is the practical how-to. If you need the definition of the underlying data first, read our glossary entry on firmographic data. Here we cover which firmographics actually predict fit, how to set segment boundaries from your closed-won data with a worked example, what to show each segment on your website, where the data comes from, how accurate it is, and how many segments to start with.
Which Firmographics Actually Predict B2B Fit
Dozens of firmographic fields exist. Only a handful usually change who buys, how much they pay and how long it takes. Rank candidates by two questions: does the attribute change the buyer's problem or process, and can you know it reliably before a sales call?
| Attribute | What it predicts | Reliability before first contact | Use it when |
|---|---|---|---|
| Industry (sub-industry) | Problem, vocabulary, regulation, proof needed | Medium: classification schemes disagree | Your case studies or compliance story differ by industry |
| Employee count band | Buying process, number of stakeholders, plan fit | Medium to high | Pricing tiers or sales motion change with size |
| Revenue band | Budget capacity | Low for private companies (often estimated) | Headcount misleads, e.g. lean high-revenue firms |
| Geography | Language, currency, data residency, sales coverage | High at HQ level, lower at office level | You sell or comply differently by region |
| Growth and funding signals | Urgency, budget availability, hiring needs | High for funded startups (public announcements) | Your product helps teams that are scaling |
| Tech stack overlap | Integration fit, switching cost, competitor usage | Medium: detection works for web-visible tools | Your product depends on or replaces specific tools |
The last row is strictly technographics rather than firmographics, but for segmentation it behaves the same way: a fixed attribute of the company that changes what you should say. Our view: for software companies, "uses tool X" is often a sharper segment than industry, because it maps directly to an integration message.
How to Set Segment Boundaries From Closed-Won Data
Do not start from a vendor's size bands. Start from your own outcomes. The method:
- Export closed opportunities (won and lost) from the last 12 to 24 months with company headcount, industry, region, deal value, created date and close date.
- Enrich missing fields so every row has the attributes you want to test. Blank rows bias the result.
- Bucket finely first. Use narrow bands (for example 1-20, 21-50, 51-100, 101-200 and so on) so you can see where outcomes change.
- Calculate per bucket: win rate, average contract value, median sales cycle, and expected value per opportunity (win rate times average contract value).
- Merge adjacent buckets that behave alike and draw boundaries where a metric jumps.
- Sanity-check sample sizes. A bucket with five deals is an anecdote. Merge it with a neighbor or mark it as "unknown".
Worked example
Example (illustrative numbers for a hypothetical B2B software company with 300 closed opportunities over 18 months):
| Headcount band | Closed opps | Won | Win rate | Avg contract value | Median cycle | Value per opp |
|---|---|---|---|---|---|---|
| 1-50 | 120 | 18 | 15% | €6,000 | 30 days | €900 |
| 51-200 | 90 | 22 | 24% | €14,000 | 55 days | €3,400 |
| 201-1,000 | 60 | 15 | 25% | €32,000 | 95 days | €8,000 |
| 1,001+ | 30 | 3 | 10% | €70,000 | 180 days | €7,000 |
Value per opportunity is win rate times contract value: 0.24 x €14,000 = €3,360 (rounded to €3,400), 0.25 x €32,000 = €8,000, and so on. What the table says:
- 1-50 produces the most opportunities but the least value per opportunity. That points to a low-touch, self-serve or fast-demo path, not more SDR time.
- 51-200 and 201-1,000 win at almost the same rate, but contract value more than doubles. Keep them as separate segments for pricing and proof, even though the sales process is similar.
- 1,001+ has high value per deal but a low win rate based on only three wins. That is too few to conclude much. Treat it as a named-account ABM motion and revisit when you have more data.
The boundary at 1,000 here comes from the data, not from a convention. Your boundaries might sit at 30 and 400. Repeat the exercise for industry and region, and you have an evidence-based ideal customer profile instead of a guessed one.
Segment Table: What to Show Each Segment on Your Website
A segment is only useful if something changes for it. Continuing the example, here is how the segments translate into website messages and calls to action.
| Segment | Hero message angle | Proof to show | Primary CTA |
|---|---|---|---|
| Small (1-50) | Fast setup, low effort | Short testimonial from a small team, setup time | Start now / book a 15-minute walkthrough |
| Mid-market (51-200) | Outcome for a growing team | Case study from a similar-size company, integrations | Book a demo |
| Upper mid-market (201-1,000) | Scale, control, ROI | ROI case study, admin and permissions, security summary | Talk to sales / get a tailored ROI estimate |
| Enterprise (1,001+) | Security, compliance, rollout support | Recognizable logos in their industry, security documentation | Contact enterprise team |
| Industry overlay (e.g. healthcare) | Industry-specific problem and terms | Industry case study, relevant compliance | Same as size segment, with industry wording |
| Unknown company | Default message | Broad logo bar | Default CTA |
Company size and industry are best handled as two layers: size picks the CTA and the proof weight, industry picks the vocabulary and the case study. Our posts on personalizing your website by company size and industry-specific website personalization show page-level examples for each.
Where Firmographic Data Comes From
You need firmographics at two moments: offline, when analyzing deals and building target lists, and in real time, when a visitor lands on your site.
- Your CRM. HubSpot, Salesforce or Pipedrive records for existing customers and opportunities. The most trusted source for your own accounts, but often incomplete for prospects.
- Enrichment providers. Databases that append company attributes to a domain or record, such as ZoomInfo, Apollo, Cognism, Dun & Bradstreet and HubSpot Breeze Intelligence (formerly Clearbit). Coverage differs by region and company size, so test on your own records before committing.
- Reverse IP lookup and visitor identification. Tools that match the network a visitor comes from to a company, such as Leadfeeder, Demandbase and 6sense. This is what makes real-time website segmentation possible. See how reverse IP lookup works.
- Technographic sources. BuiltWith and Wappalyzer detect technologies visible on a company's website; HG Insights and similar vendors infer wider installed bases.
- Public sources. Company registries (for example Companies House in the UK), SEC filings for US public companies, Crunchbase for funding rounds, job postings and LinkedIn company pages.
- Declared data. A form field asking for company size, or a domain you can enrich. The most accurate source for the individual visitor, and the one most teams over-ask for.
For using the same attributes to build outbound lists, see this guide to firmographic data for sales targeting.
Accuracy Caveats Most Teams Ignore
- Headcount differs by source. One provider counts LinkedIn profiles, another uses filings, another estimates. Pick one source as canonical for segmentation so a company does not jump between bands.
- Industry codes disagree. NAICS, SIC and LinkedIn industries classify the same company differently, and multi-line businesses fit several. Map provider codes to your own short list of 5 to 10 industries.
- Revenue for private companies is usually modeled. Use it as a rough band, never as a precise filter.
- IP matching misses people. Remote workers on home broadband, VPNs and mobile networks often cannot be resolved to a company, and small companies are harder to match than large ones. Match rates vary widely by audience, so measure your own: take a sample of visitors who later filled a form and check how many were matched to the right company beforehand.
- Location means HQ. A visitor from a Berlin office of a US company is often tagged as US. Use visitor geolocation for language and currency, and HQ for sales territory.
- Data ages. Funding rounds, headcount and tech stack change. Re-enrich active accounts at least quarterly.
How Many Segments to Start With
Fewer than you think. Data is rarely what limits you; content and measurement are. Each segment needs its own copy and proof, someone to keep it current, and enough traffic to tell whether it works.
The measurement side is unforgiving. Example: a segment that converts at 2% and where you want to detect a lift to 2.6% (a 30% relative improvement) needs roughly 9,800 visitors per variant at 95% confidence and 80% power, about 19,600 identified visitors for a two-variant test. The math is in our guide to A/B test sample size. Most B2B segments will never reach that in a quarter.
Our decision rule:
- Start with three segments that cover most of your won revenue, plus the default experience for unidentified visitors.
- Formally A/B test only the largest segment. For the others, ship the personalized experience based on sales and customer evidence, and track pipeline and win rate per segment quarter over quarter.
- Add an industry overlay only for industries with their own case studies and at least a few won deals.
- Add a new segment only when you can name the different message, the different proof and the owner.
Once the rules are defined, a tool like Markettailor's segmentation can evaluate them on each visit, but the hard part is the analysis above, not the software.
Frequently Asked Questions
What is firmographic segmentation?
Firmographic segmentation divides B2B prospects and customers into groups based on company attributes such as industry, employee count, revenue, location, ownership and growth stage. Each segment then gets messaging, offers, pricing emphasis and sales coverage that match how companies of that type buy. It is the B2B counterpart of demographic segmentation, which groups individual consumers by attributes like age or income.
What are examples of firmographics?
Common firmographics are industry or sub-industry, number of employees, annual revenue, headquarters country and region, legal or ownership structure (public, private, PE-backed, non-profit), company age, funding stage and recent funding, and growth signals such as hiring. Technology stack is usually classed separately as technographics but is often used alongside firmographics in segmentation.
How do you choose firmographic segment boundaries?
Use your own closed-won and closed-lost data. Bucket opportunities finely by headcount, industry or region, calculate win rate, average contract value, sales cycle and value per opportunity for each bucket, then merge buckets that behave alike and place boundaries where a metric changes sharply. Avoid relying on a data provider's default size bands, which rarely match your buyers.
How many firmographic segments should you start with?
Three, plus a default experience for visitors you cannot identify. Each segment needs its own message, proof and owner, and enough traffic to measure. Most B2B sites lack the traffic to A/B test many segments, so test the largest one formally and track pipeline per segment for the rest. Add segments only when there is a clearly different message to show.