Foundations
Part of Social listening foundations for an England account model
How to estimate England's social listening opportunity so another analyst can repeat it
Estimate England's social listening opportunity from ONS business counts, explicit exclusions and a repeatable calculation, with a worked example and Ofcom context.
There is no official England figure for the social listening market. ONS data can define a business population, but it supplies no adoption rate, annual spend or 2027 demand total. A publishable estimate therefore needs a bottom-up account calculation with a visible denominator and an exclusion log.
What to take away
- Build a bottom-up account calculation, because no official source gives an England social listening market size.
- Define one customer unit and an England rule, then record the limitation that choice creates.
- Report eligible accounts as the denominator minus documented exclusions.
- Treat Ofcom research as context, never as an adoption rate.
- Publish a serviceable-account range, not revenue, until purchase frequency and comparable costs exist.
Set one unit and one geography
First decide what one potential customer means. For an account-based service, use an enterprise rather than a local unit. For a service aimed at individual sites, local units may be appropriate. Never add the two.
Choose unit and England rule
What does one potential customer mean?
Account-based service -> enterprise unit
Individual-site service -> local unit
Then select an England rule. The cleanest option for many models is an enterprise assigned to an English head-office region, although it misses UK groups buying centrally elsewhere for English operations. Record that limitation.
Use the ONS denominator
The ONS business bulletin counted 2.79 million VAT and/or PAYE businesses in the UK as of March 2026, up 1.9% on March 2025. It also counted 3.2 million sites (local units), up 1.6%. The source is the Inter-Departmental Business Register.
The largest industry group was professional, scientific and technical, at 15.3% of registered businesses. Table 4 of the bulletin shows the number of businesses in each region and its share of the UK total, to the nearest thousand, and Table 5 pairs businesses with their local units.
For England, take the nine English regional rows of Table 4 and leave out Wales, Scotland and Northern Ireland. Record the release, the table and the extraction date. London saw the largest rise between 2025 and 2026, at 2.6%. Treat the result as a source denominator, not a count of organisations buying listening services.
Screen for service relevance
Create one row per account and test the same fields:
- an English location under the chosen rule;
- an industry and size that the offer is designed to serve;
- a named decision that online conversation could inform;
- accessible and permitted sources relevant to that decision;
- enough material during a fixed sample window for meaningful review;
- an owner able to act on and correct the analysis;
- no duplicate parent-level buying record.
The serviceable count is simply: eligible accounts = ONS denominator minus documented exclusions. If the list is too large for complete review, draw a reproducible sample, publish the selection procedure and report uncertainty. Do not silently treat unreviewed accounts as eligible. The target audience guide shows how to define accounts by decision.
Example: from denominator to buying fraction
The figures below are invented to show the arithmetic. They are not ONS data for any English region.
- Extract120,000 enterprises in the chosen English region and industry filter.
- Exclusions84,000 fail the relevance fields, with each row logged.
- Eligible accounts120,000 minus 84,000 is 36,000.
- Sample200 eligible accounts are checked for a dated buying event, and 6 show one.
- Observed buying fraction6 divided by 200 is 3 per cent of the sample.
For the sample size, the usual formula for a proportion is n = z squared x p x (1 - p) divided by the margin squared. With 95 per cent confidence (z of 1.96), an expected p of 0.03 and a margin of 0.02, that gives about 280 accounts, so 200 is too few for that margin.
Apply 3 per cent to the 36,000 only if the sample design supports the inference, with statistical review before publication. Otherwise present the observed count of 6 alone. The demand signals article explains why web visits and mention volumes are not buying events.
Separate context from demand
Ofcom's Online Nation report is an annual report on what people are doing online, how they are served by content providers and platforms, and their attitudes to and experiences of using the internet. The page was last updated on 10 December 2025.
That makes it useful context for why digital conversation may be relevant. Its respondents are people, not purchasing organisations, and its geography is wider than England. It cannot supply an English buyer fraction, so keep it apart from the demand evidence.
Accept buying evidence, not interest
Review the same eligible-account sample for a dated buying event. Accept a published procurement, a signed research brief or an internally verified budget with a responsible owner.
Public sector purchases can be checked in Contracts Finder, which covers contracts worth over £12,000 including VAT, and in Find a Tender for high value contracts in the UK's public and utilities sectors. Exclude job adverts, page visits, mention volumes and generic technology interest unless the method defines a weaker signal category.
State the sample size, observation period and missing cases. A real purchase proves use by one organisation, not prevalence across the market.
Keep value out until cost is comparable
Do not multiply accounts by an assumed licence price. Projects, licences and managed retainers allocate analyst work differently.
A later spend model should normalise dated quotations to the same sources, history, seats, query capacity, analyst hours, reporting period and tax treatment. The pricing models article shows the main structures and the budget template records the inputs.
Until both a verified purchase frequency and a comparable cost unit exist, the defensible result is a serviceable-account range, not market revenue. Publish the ONS table, extraction date, filters, row-level exclusions, formula and scenario assumptions so another researcher can reproduce the estimate. The best providers longlist is a separate exercise from sizing.
Before you act
- Set one unit and one England geography rule.
- Record the limitation of your chosen England rule.
- Screen each account against the same relevance fields.
- Publish the selection procedure if you sample.
- Verify dated buying events and state the sample size.
- Hold back value until cost units are comparable.
Common questions
Why can't ONS data alone give an England social listening market size?
ONS counts registered businesses and sites by region, industry and employment size. That is a source denominator, not a count of organisations buying listening services. It supplies no adoption rate, annual spend or 2027 demand total, so a bottom-up account calculation is needed.
How should I use Ofcom's Online Nation in an estimate?
Use it only to explain why digital conversation may be relevant. Its respondents are people, not purchasing organisations, so it cannot supply an English buyer fraction. Keep it separate from demand evidence.
When can I turn account counts into a revenue estimate?
Only when you have both a verified purchase frequency and a comparable cost unit. Normalise dated quotations to the same sources, history, seats, query capacity, analyst hours, reporting period and tax treatment. Until then, publish a serviceable-account range, not market revenue.
