How to define your ICP when you sell a horizontal B2B product

By Dave Curran, Co-Founder, Firmbase | March 2026 | 10 min read

Every ICP framework you'll find online starts from the same assumption: you know what industry you sell to.

That works if your product is genuinely sector-specific. Payroll software for restaurants, compliance tools for financial services - if your ICP is clearly bounded by industry, defining it is mostly an exercise in being precise about which slice of that sector you want.

But what if you sell to everyone? An MSP sells IT services to any company with a network. A recruiter sells to any business that is hiring. A £10k SaaS product that solves a workflow problem sells to any company with that workflow, regardless of what they actually do. A business consultancy sells to any MD who is wrestling with the specific problem they solve.

For these sellers, the standard ICP frameworks fall apart. Sector-based filtering doesn't work when your best customers are spread across a dozen industries, and the advice to 'niche down' ignores the reality that the product genuinely does work for a very wide range of companies.

This article is for that seller. It covers what actually works when you're trying to define and map a horizontal ICP: the filters that matter, why the tools you're probably using are missing a significant chunk of your market, and how to build a complete picture of your target universe using UK public data.

What a horizontal ICP actually is (and why it needs a different approach)

An ICP - Ideal Customer Profile - is a description of the type of company most likely to buy from you, get value from what you sell, and retain or expand over time. The standard advice is to be as specific as possible: the more precise your ICP, the less time you waste on accounts that will never close.

That logic holds. The problem is that most ICP frameworks define 'specific' in terms of industry or sector, because that is the easiest filter to apply. Sector is a clean, categorical variable. You can filter by it in any CRM or prospecting tool without thinking.

For horizontal products, this approach fails for two reasons.

First, sector is genuinely not the defining characteristic of your best customers. A cloud infrastructure company's best customers are not 'all companies in the professional services sector.' They're companies at a particular stage of technical maturity, above a certain headcount in their engineering team, with a particular growth profile that signals increasing infrastructure complexity. Those companies exist across dozens of sectors.

Second, the tools available for sector-based filtering are significantly worse than most people assume. We'll come back to this.

The alternative is to define your ICP around the three variables that actually predict fit for a horizontal product: revenue band and trajectory, growth signals, and director or leadership profile. None of these require you to know what industry your customer operates in. All of them are available for every UK registered company.

The three filters that work for horizontal ICPs

Filter 1: Revenue band and trajectory

Every UK company that files accounts with Companies House gives you at minimum their net assets over time, and often their turnover. This is the starting point for any horizontal ICP, because it tells you two things that sector filtering cannot: whether this company has the budget to buy what you sell, and whether they are in a growth phase that creates the kind of problems your product solves.

A company with £2M in turnover that has been flat for four years is fundamentally different from one at the same revenue that has doubled its headcount and is filing new accounts every year showing a bigger cash position. The second company is going somewhere. They're hiring, investing, taking on new clients, building out processes. Those are the conditions under which companies buy new tools and services.

The specific band matters and varies by what you sell. A £10k SaaS product has a different sweet spot than a £50k consulting engagement. But the principle holds regardless: define your revenue floor (below which companies almost never buy), your revenue ceiling (above which you lose to enterprise alternatives), and look for growth trajectory within that band rather than just a static revenue snapshot.

What to look for in the data: Turnover or headcount growing year-on-year within your target revenue band. Consistent growth over two or three years is a more reliable signal than a single good year - it tells you the company is in a sustained expansion phase, not just recovering from a bad one.

Filter 2: Growth signals

Beyond the filed accounts, there are real-time signals that tell you whether a company in your revenue band is currently in a growth phase - and therefore currently in a buying phase.

Director appointments are the most reliable. A company that has just hired a new Head of Operations, a first CFO, or a commercial director is signalling that they're building out their leadership structure. That is a growth company. New leaders evaluate existing processes and tools in their first 90 days, making them more receptive to sales conversations than at almost any other point.

Recruitment patterns are the second most useful signal. What a company is hiring for tells you exactly what problem they're trying to solve. A company posting three SDR roles is scaling their sales function and probably needs everything that a growing sales function needs. A company hiring their first Finance Director is reaching an inflection point where financial complexity is increasing. A company advertising for 'Head of IT' for the first time is moving past ad-hoc infrastructure management.

For horizontal products, the specific recruiting signal that matters will be different for every seller. Define it precisely: not 'they're hiring' but 'they're hiring for X role, which tells us they have Y problem, which is exactly what we solve.'

What to look for in the data: Leadership appointments in the last 90 days. Recruitment for roles that directly signal the problem your product solves. Both are publicly available and verifiable - no probabilistic guessing required.

Filter 3: Director and leadership profile

The third filter is the one most horizontal sellers underuse: who is actually running the company, and does that person match the profile of your best customers?

This sounds like it's about seniority (it isn't) or industry background (it isn't that either). What it's actually about is the specific combination of experience, company stage, and role that predicts whether this is a person who will engage with you, understand the value you're offering, and have the authority to say yes.

For a lot of horizontal B2B products, the best customers are founder-led or owner-managed businesses at a particular stage of professionalising - moving from informal processes to proper systems. The person running that company is often a managing director who has built something from scratch and is now wrestling with the complexity that comes from success. That person has a very different profile and a very different set of concerns from the Head of Sales at a PE-backed firm or the procurement lead at a large corporate.

You can see director tenure, previous appointments, and directorship history through Companies House. This won't tell you everything, but it can be a powerful filter when combined with revenue trajectory and growth signals.

Why SIC codes are the wrong tool for horizontal ICPs

Standard Industrial Classification (SIC) codes are the UK government's taxonomy for categorising business activity. Every company registered at Companies House has at least one SIC code. They are the most widely used filter in UK prospecting tools, and for horizontal ICPs, they are almost useless.

The problem is not that SIC codes are inaccurate - it's that they were designed to classify economic activity for statistical purposes, not to identify sales prospects. The codes are broad, inconsistently self-assigned, and frequently out of date. A technology consultancy might file under 'Information Technology Consultancy Activities' (62020) or 'Management Consultancy Activities' (70229), depending entirely on how their accountant interpreted their business at the point of filing.

More importantly for horizontal sellers: your target market is by definition distributed across many SIC codes. A typical horizontal B2B ICP - a product that sells to growing SMBs with a particular problem profile - maps to between 8 and 14 SIC codes. Most filter-based prospecting tools that require you to select sectors miss between 30% and 60% of your actual target market as a result.

The companies you're missing aren't bad-fit accounts. They're companies that do exactly what your best customers do, employ exactly the kind of people who get value from what you sell, and are at exactly the right stage to buy - they're just classified under a code you didn't think to include.

This is why natural language search matters so much for horizontal ICPs. Describing your ICP in plain English - 'growing professional services firms with a commercial director, between £1M and £10M in revenue, that have expanded their headcount significantly in the last two years' - captures far more of your actual target market than any combination of SIC filters.

How to use UK public data to map your complete universe

The good news is that the UK has one of the best public company datasets in the world. Companies House has records for every registered company - over 5 million of them - including filings, director appointments, and registered addresses. It is free, it is reasonably complete, and it is updated continuously.

The limitation is scale. Manually reviewing Companies House filings to find companies that match a precise ICP is not viable when you're looking at millions of records. The data is there; the infrastructure to use it at scale is not, if you're working directly with the raw filings.

This is the gap Firmbase fills. Rather than asking you to select SIC codes or filter by sector, Firmbase lets you describe your ICP in natural language and then searches across Companies House data - combined with web scraping, job posting data, and funding round information - to surface accounts that match your description. The output is not a static list. It's a continuously updated view of all the companies in the UK that fit your criteria, with the most recent signals surfaced at the top.

For a horizontal ICP, this means you can define your universe by describing the company profile you're looking for rather than selecting categories you hope your best customers fall into. You'll find accounts that a filter-based approach would have missed entirely.

A worked example: building an ICP for a £10k SaaS product with no vertical focus

Let's make this concrete. Imagine you sell a project management and client reporting tool for professional services firms. Your product is horizontal - it works equally well for a legal practice, a marketing agency, a management consultancy, or an IT services company. Your best customers tend to be 15 to 50 people, growing, with a director-level person who is frustrated by the amount of time being spent on internal coordination rather than client work.

Here's what a SIC-code-based approach gives you: a list of companies classified under professional services codes - probably 50,000 to 150,000 companies depending on how broadly you define it - with no way to distinguish the growing ones from the stagnant ones, the right-sized ones from the wrong-sized ones, or the ones with the coordination problem from the ones that have already solved it with something else.

Here's what a signal-based approach gives you: companies in the relevant revenue band (say £1M to £8M in turnover) that have been growing consistently - more staff, bigger client base, more complexity - that have made at least one director or senior hire in the last 12 months, and that do not yet have a Head of Operations or COO on the Companies House register, suggesting the coordination problem hasn't been addressed at leadership level yet.

That second list is dramatically smaller and dramatically better. Every company on it has the profile of a company in a buying window. The outreach you do from that list will produce more conversations, faster conversions, and less wasted effort.

How Firmbase approaches this

Firmbase was built specifically for horizontal ICP mapping. You describe the kind of company you're looking for in plain English, and Firmbase searches across Companies House filings, job posting data, and funding round information to find UK companies that match.

The result is a continuously updated view of your ICP - not a list you pull once and let go stale - with the companies showing the strongest current signals surfaced at the top. New director appointments, growth in filed accounts, and relevant hiring all feed into the signal score automatically.

For teams with a horizontal product, this is fundamentally different from trying to assemble the same picture from SIC code filters, LinkedIn searches, and manual Companies House lookups. The coverage is better, the signals are richer, and the system stays current without manual maintenance.

Start your free trial at app.firmbase.co/signup

Frequently asked questions

What is an ICP for a horizontal B2B product?

An ICP (Ideal Customer Profile) for a horizontal B2B product is a description of the company characteristics that predict a good fit when you can't use industry or sector as a primary filter. Because horizontal products sell across many industries, the most useful ICP variables are financial (revenue band, growth trajectory), behavioural (hiring patterns, leadership appointments), and structural (company stage, director profile). For UK companies, all of these signals are available through Companies House filings and job posting data.

How do I find all the companies in my ICP in the UK?

The most complete way to map a horizontal ICP in the UK is to combine Companies House data with job posting signals and funding round information, and search across them using company profile criteria rather than sector filters. Companies House alone covers every registered UK company, giving you access to financial trajectory and director appointment data for over 5 million businesses. Tools like Firmbase aggregate this data and allow natural language search across the full dataset, so you can find all companies matching your ICP criteria without manually filtering by SIC code.

What data should I use to define my ICP if I don't sell to a specific industry?

The three most predictive data points for a horizontal B2B ICP are: revenue band and trajectory (available from Companies House filed accounts), growth signals including director appointments and recruitment patterns (both publicly available), and company age relative to growth rate (indicating whether a company is in a scaling phase). These variables work regardless of industry and are measurable across the entire UK company register.

How many companies are in a typical UK B2B ICP?

A well-defined horizontal B2B ICP in the UK typically contains between 2,000 and 15,000 companies, depending on revenue band and how precisely the growth signal criteria are set. A typical horizontal ICP maps to between 8 and 14 SIC codes, meaning filter-based approaches using sector selection miss 30% to 60% of the actual addressable market. Defining the ICP by company profile and signals rather than by sector significantly expands coverage without sacrificing precision.

Why are SIC codes unreliable for B2B prospecting?

SIC codes were designed for statistical classification of economic activity, not for sales prospecting. They are broad, frequently self-assigned without specialist guidance, and often out of date relative to a company's current activity. More importantly, horizontal products have target markets that are genuinely distributed across many SIC codes - so any filter-based prospecting that requires sector selection will systematically miss a significant proportion of target accounts. A company doing work identical to your best customers may simply be classified under a different code.

How often should I update my ICP?

An ICP should be reviewed whenever you have meaningful new data about which customers have retained, expanded, or churned, and whenever your product evolves to serve a new or different buyer. In practice, most teams review their ICP quarterly. The more important question is whether your ICP mapping is static or live. A static ICP you refresh quarterly will always lag behind the market. A live ICP that continuously surfaces new accounts matching your criteria means you never miss a company entering your target window.