Measurement
How ONS data builds keyword research benchmarks: the releases, the method and a worked example
Which ONS releases to use for UK keyword research benchmarks, how to turn a population into a searches-per-thousand ratio, and a worked example with clearly labelled illustrative figures.
What to take away
- ONS releases give a UK denominator, so benchmarks read as searches per thousand businesses rather than imported volumes.
- Three datasets do most of the workinternet access, business demography, and e-commerce and ICT activity.
- The method is three stepschoose the dataset, convert its category into a keyword group, then set a ratio.
- UK seasonality follows company formations, the tax year and the school year as much as retail peaks.
- ONS covers market size, not phrasing or intent.
Why imported US volumes mislead UK teams
Most keyword tools default to a US database. A term with six figures of monthly US searches can be a rounding error in the UK.
British searchers spell words differently, name institutions differently and look for rules that do not exist across the Atlantic.
The Office for National Statistics measures the size of the UK population that could be searching. Its internet access bulletin covers household and individual internet use. The linked edition is 2020, so check for a newer release before you use it.
If you want the logic behind baselines before you open a dataset, how benchmark baselines get set works through the method.
Imported volumes also flatten regional variation. A national figure hides the gap between Greater London, the West Midlands and West Yorkshire, and that gap matters when you plan local content.
US vs UK search volumes
US keyword tool
- Volume source
- US database
- Spelling
- US forms
- Institutions
- US names
- Regulation
- Ignored
- Regional detail
- Flattened
UK ONS-based
- Volume source
- UK population
- Spelling
- British forms
- Institutions
- UK names
- Regulation
- UK GDPR, ICO, CMA, ASA
- Regional detail
- ONS breakdowns
The ONS releases that map to keyword demand
Pick the release that describes your market, not the one with the most familiar name.
| ONS release | What it measures | Keyword groups it supports |
|---|---|---|
| Internet access | Household and individual internet use | Consumer advice and comparison queries |
| Business demography | Company births, deaths and survival by region | Supplier, software and startup terms |
| E-commerce and ICT activity | Business internet use and online selling | E-commerce platforms and digital tools |
| UK business: activity, size and location | Enterprise and local-unit counts by region, industry and size band | Regional and sector keyword groups |
An ONS category is a population, not a keyword group. Read the definitions before you map anything.
Each release answers a different question: who is online, how many organisations exist, and how digitally mature they are.
The releases go by these ONS titles. Search for them by name on the ONS site and note the edition and period each time:
- Internet access – households and individuals, Great Britain:a bulletin on who uses the internet, with breakdowns by age and region; check which edition is current.
- Business demography, UK:an annual bulletin on company births, deaths and survival.
- UK business: activity, size and location:an annual dataset of enterprise and local-unit counts by region, industry and size band. It is the one to use for the nine English regions.
- E-commerce and ICT activity, UK:a bulletin on business internet use and online selling.
The ONS release calendar lists the next publication date for each, so use it to anchor your review cycle. Each release also has a dataset page with downloadable tables; take the population from the table, not the summary text.
A ratio is only a benchmark if it can be reported monthly, so read what keyword research key metrics mean for reporting before you fix your columns.
Example: a worked UK benchmark
- Choose the dataset that matches your market. Limited companies point to business demography; consumer advice points to internet access.
- Convert the category into a keyword group, capped at a dozen terms drawn from the questions that population would search.
- Divide monthly UK searches by the ONS population and express the result per thousand.
| Keyword group | ONS source | Population | Monthly UK searches | Benchmark |
|---|---|---|---|---|
| Payroll software for new limited companies | Business demography, UK: 2024 release | 24,000 new companies (illustrative) | 1,800 (illustrative) | 75 searches per 1,000 |
The figures in that row are illustrative, not ONS outputs. The Business demography, UK: 2024 release, published on 20 November 2025, gives real denominators: about 317,000 business births and 280,000 deaths (provisional) in 2024, and about 2.86 million active businesses. These cover businesses registered for VAT or PAYE, not only limited companies, so a payroll-software ratio needs a narrower population than the headline count.
For scale, suppose your tool reported 3,170 monthly UK searches for a group of new-business terms (an assumed figure). Against 317,000 births, that is 10 searches per 1,000 births. Replace the assumed search figure with your own tool's number.
A ratio per thousand survives a change of tool and travels across regions.
Three-step ONS benchmark method
- Select dataset matching your market
- Convert ONS category into keyword group
- Set benchmark as share of ONS population
Timing seasonality with ONS releases
UK seasonality is several curves, not one. Business demography shows when formations and closures cluster, and demand for accountancy, insurance and recruitment follows with a lag.
Internet access data is annual and changes slowly, so it sets a baseline rather than showing seasonality. The school year and the tax year show up in your own search data instead.
Retail peaks are a separate curve again, and ONS retail sales figures help date them.
Use at least two release periods. One shows a pattern, two show whether it repeats.
When a seasonal swing looks like an opportunity, when to read keyword research signals lists the tests worth applying first.
Building a benchmark table your team can reuse
- One row per keyword group, with its terms listed in a single cell.
- The ONS release name and period in every row, not just the first.
- A population figure taken from the release, never from memory.
- A ratio per thousand, so rows compare like with like.
- A review date tied to the next publication.
When a ratio moves sharply, check the dataset before you check the content. Business demography refreshes annually, so an annual review cycle is realistic for most teams.
Where ONS stops and primary research starts
ONS tells you how many. It does not tell you why, or what people type.
A release can show a sector growing, not what its buyers type. Search Console queries, site search logs and sales call notes fill that gap.
Use ONS for the denominator and primary research for the numerator. The monthly UK search figure comes from a named tool: Google Keyword Planner set to the United Kingdom, Ahrefs or Semrush with the UK database selected, or your own Search Console queries. Note the release period on every row, because ONS publishes after the period it describes.
Common questions
Which ONS release should a content team start with?
Start with internet access if you publish consumer content, and business demography if you sell to organisations.
Can ONS data replace keyword tools entirely?
No. It gives market size and structure, not phrasing or volume. Use it for the baseline and tools for what sits inside it.
How often should UK benchmarks be reviewed?
Match the cycle to the ONS publication calendar. Annual works for most teams, with an interim check when a major release lands.
Why not just scale US volumes by population?
Population scaling ignores spelling, regulation and regional variation. It produces a smaller number rather than a British one.
