How to size UK B2B expert markets with ONS data

What to take away

Why SIC codes anchor B2B market sizing

A sector is a claim about who counts. SIC codes are the closest thing the UK has to an agreed answer. Standard Industrial Classification 2007 assigns every registered business a five digit code describing its main activity. Analysts use the first two digits for broad sectors and the full five for narrow ones.

Without a code boundary, a target list drifts. You end up counting recruitment agencies as marketing services, or software publishers as IT consultancies, because the company description sounded right. A code boundary makes the list reproducible. A colleague can rerun it and get the same firms.

The SIC code selection method

The method is a short sequence, not a research project. Work down from the sector you actually sell to, not the sector you wish you sold to.

Keep the list to roughly five to fifteen five digit codes. Longer lists stop being a boundary and start being a wish.

Where the code boundary breaks

SIC codes describe the main activity of a registered business. A holding company may sit under a code that says nothing about what its subsidiaries sell. A large group may file one code for hundreds of trading entities. A sole trader with no registration has no code at all.

This matters most in professional services and technology, where the same firm can be classified several plausible ways. Treat the code as the first filter, not the last. If your target audience definition depends on firmographics the code cannot express, say so before you quote a number.

Using ONS business counts to size a sector

The Office for National Statistics publishes UK business counts by industry, employment size band and region. These are the raw numbers behind most credible sector sizes. They are also free, which removes the excuse for quoting a supplier's estimate as fact.

Read the counts in two passes. First, the total number of businesses in your code boundary. Second, the split by size band. The second pass usually changes the plan more than the first.

A sector with 40,000 firms, of which 38,000 employ fewer than five people, is a different market from one with 4,000 mid sized firms.

Reading the size bands properly

ONS size bands run from zero employees through to 250 or more. Most B2B budgets cannot serve the micro end profitably. Filtering to firms with ten or more employees often cuts a headline count by more than half. That is not a problem with the data. It is the data doing its job.

Regional splits matter for UK fieldwork. Greater London, the South East and the North West carry a large share of professional and technical services. Scotland, Wales, the West Midlands, Yorkshire and the Humber and the South West each have distinct clusters worth naming in a plan. A national count hides all of this.

Getting the survey context right

The counts come from business surveys, not a census of every firm. The ONS publishes the instruments behind them, including the Business surveys - Office for National Statistics used to collect business data. Knowing the instrument tells you what the count can and cannot support.

If you need to know how firms in a sector buy, the counts will not tell you. They tell you how many firms exist and how big they are. Anything about behaviour, budget or intent needs primary work. That is where audience research takes over from desk data.

Companies House filings as a target list source

Companies House holds the register of UK companies. Every incorporated business files there, and much of the record is free to search and download. This is where a sector count becomes a list of names you can actually contact.

Search by SIC code, then filter by location, incorporation date and company status. Export the results and you have a working list. For each company you can pull the registered address, incorporation date, accounts filing history, confirmation statements and officer appointments.

The Companies House - GOV.UK register is the authoritative source for these filings. Use it rather than a scraped directory, which will be stale and may breach data protection rules.

What filings tell you beyond the name

Filing history is a demand signal in its own right. A company that filed its first accounts recently is young and likely still building its supplier base. A company that changed its SIC code last year has repositioned. A company with a new finance director appointed three months ago may be about to review spend.

These are weak signals individually. Stacked across a list of several hundred firms, they become a ranking. That ranking is more useful than a single firmographic score. It also feeds naturally into how you read demand signals across a sector.

Compliance when you build the list

Companies House data is public, but using it for marketing is not unrestricted. UK GDPR and the ICO's rules apply to personal data, including the names of directors and sole traders. Business to business contact is not exempt from the legislation.

Keep the lawful basis documented. Do not mix scraped personal emails into a Companies House export. If you buy a list, check whether the supplier can show its source and its basis for contact. The ICO has taken action against firms that could not.

Building benchmarks from business population estimates

The business population estimates are the annual official count of UK businesses, published by the Department for Business and Trade. They cover registered and unregistered businesses and split them by size and sector. That combination is what makes them the default benchmark.

The Business population estimates 2024 - GOV.UK release gives the current official picture by size band and industry. Use it to sanity check any sector count you built from a commercial database. If your list is twice the official count, you have a definition problem.

Building a benchmark table

A benchmark is only useful if it is comparable. Build a small table with the same columns each time, so year on year changes are visible rather than argued about.

The gap between the second row and the fourth row is your qualification rate. A large gap usually means the code boundary is too wide, or the list has not been deduplicated against groups and branches.

Using the time series

The Business population estimates - GOV.UK collection holds earlier releases, which lets you show direction of travel rather than a single year. A sector growing at a steady rate supports a different plan from one that has plateaued.

For macroeconomic context, the Gross Domestic Product (GDP) - Office for National Statistics data gives the wider picture on output. If a sector's business count is rising while its output is flat, you are looking at fragmentation, not growth. That distinction should change how you price and package work.

Turning ONS data into a B2B influencer marketing target list

A target list for expert and creator led work is not the same as a sales list. You are looking for firms whose buyers are reachable through credible named practitioners. That changes the filters.

Start from the sector boundary you built with SIC codes. Apply the size band filter from the business population estimates. Then rank the remaining firms on signals that suggest they buy advice, not just products. Firms that publish research, sponsor events or employ in house marketing teams are stronger candidates than firms that do none of these.

A worked example

Take a mid sized professional services sector in the North West. ONS counts suggest roughly 9,000 firms, of which about 1,200 employ ten or more people. Companies House returns around 1,000 active companies in the chosen codes with a registered address in the region.

Filter to firms incorporated more than three years ago, with accounts filed on time, and at least one marketing or communications officer appointed. That leaves about 300. Of those, perhaps 120 publish content or speak at events. Those 120 are the realistic universe for an expert led programme.

That number is the one to quote. Not the 9,000. Not the 1,200. The 120 is defensible, sourced and small enough to work properly.

Ranking and sequencing

Work the list in clusters rather than as one flat file. A cluster of thirty firms in the same code and region will share trade press, events and often the same handful of credible voices. That is where expert led activity compounds.

Choosing the voices

Once the list exists, the creator or expert side needs the same discipline. Check each practitioner's registration at Companies House if they trade through a limited company. Check disclosure practice against the CAP Code and CMA guidance. Check that their audience actually contains the job titles on your list.

For a fuller view of what credible numbers look like in this market, see our notes on market size in England and on where to find benchmark research. Both are useful when a client asks why your universe is smaller than a competitor's.

Limits of official data and how to state them

Official data is the best starting point and never the whole answer. Naming its limits protects your work and your client's budget. It also makes you the person in the room who is trusted with numbers.

Lead with the lag. Business population estimates are annual and published after the period they describe. ONS business counts update on their own schedule. Companies House filings arrive when companies file, which for small firms can be many months after the fact. A target list built today reflects a market that has already moved.

Then cover classification. SIC codes are self reported and only lightly checked. Firms pick codes that match their self image, not their revenue. A code boundary is a convenience, not a truth.

Coverage and suppression

Unregistered businesses, including many sole traders, sit outside Companies House entirely. Very small counts get suppressed in published tables to protect confidentiality. Both gaps hit the smallest and newest firms hardest, which are often exactly the firms an emerging category wants to reach.

Group structures distort everything. One filing can represent dozens of trading entities, or one entity can appear across several codes. Deduplicate before you quote, and say how you did it.

A short statement you can reuse

Write one paragraph and put it in every deck. It should name the source, the release date, the code boundary, the size filter and the known gaps.

For example: this sizing uses the current business population estimates, ONS business counts for the named region, and a Companies House extract on a stated date, filtered to firms with ten or more employees in the listed SIC codes. It excludes unregistered businesses and does not adjust for group structures.

That paragraph does more for your credibility than a bigger number would. It also stops the number being reused in contexts it cannot support.

Common questions

Why use SIC codes rather than a commercial database?

SIC codes are free, stable and reproducible, so anyone can audit your boundary. Commercial databases add contact detail and technographics but change their coverage without notice. Use codes for the boundary and a supplier for enrichment, not the other way round.

How current are ONS business counts?

They are published on a scheduled basis and describe a reference period that has already passed. Treat them as a structural picture, not a live count. For anything time sensitive, pair them with Companies House filings and your own pipeline data.

Can I build a target list from Companies House alone?

Yes, and many agencies do. You get names, addresses, codes and filing history. What you do not get is turnover, headcount or buying intent. Plan for a qualification step before the list reaches a campaign.

Do UK GDPR rules apply to a Companies House list?

They apply to personal data within it, including director names and sole trader details. Business to business marketing is covered. Document your lawful basis, keep the source recorded and follow ICO guidance on contact.

How many firms should a realistic target universe contain?

Fewer than most briefs assume. A well filtered universe of a few hundred firms in one sector and region is workable. If your number runs into tens of thousands, the boundary is too wide to support expert led activity.

What should I do when official data and a supplier's figure disagree?

Show both, explain the definitional gap and say which you rely on and why. Suppliers often count differently, for example including branches or unregistered traders. The explanation is the valuable part, not the reconciliation.