B2B demand generation: four checks before quoting a benchmark

What to take away

Benchmark research is comparing your own funnel numbers against an external reference point such as a published sector average, a peer survey or your own historic baseline.

What does benchmark research actually measure?

Most benchmark work tracks conversion between funnel stages, cost per qualified lead, sales cycle length or win rate. Each of those depends on how a lead, a qualified lead and a win are defined at the point of collection.

Two studies can report the same metric and disagree widely. The gap usually comes from definition, not performance. A form fill counts as a lead in one dataset and a marketing qualified lead in another.

Sample composition matters too. A benchmark drawn mainly from enterprise software firms rarely transfers to a ten-person agency selling locally. Cost per qualified lead also moves with channel mix, so compare like with like before drawing a conclusion.

As an illustration, a team paying £400 a month per channel might see a 2% enquiry-to-opportunity rate where a published study reports 5%. The gap usually comes down to how each side counts a webinar sign-up.

Where should you source benchmark data?

Start with your own records, because your CRM holds the only dataset built on your definitions and your sales cycle. External sources add context once that baseline exists. Then rank those sources by how close their sample sits to your market, company size and buying cycle.

Platform documentation shows how a vendor defines its own funnel metrics. HubSpot sets out its definitions in a lead generation use case covering generating and qualifying leads, useful when you check what a platform counts.

Sector bodies often publish survey data with a stated methodology. The IAB UK maintains an SME digital advertising toolkit that helps smaller teams frame channel and spend questions before comparing themselves with anyone else.

Trade associations, regulators and universities often publish method notes that show how a figure was produced, which is the part most summaries leave out.

Where a supplier publishes aggregate customer data, treat it as marketing until you read the methodology. A vendor dashboard is a tool, not an impartial study, even where the product is widely used, as with Salesforce sales cloud automation.

How do you judge whether a benchmark is usable?

Work through four checks before a figure enters a report.

If a source fails two or more checks, cite it as context rather than evidence. If it passes, record the checks so a reader can trace the figure back. The wider measurement and reporting guide sets out how benchmarks sit alongside targets and forecasts.

Date matters most in fast-moving categories, where a two-year-old survey can describe a market that no longer exists.

A worked example: HubSpot's State of Marketing

Run a well-known study through the four checks to see how they work. HubSpot's 2026 State of Marketing report draws on data from more than 1,500 global B2B and B2C marketers, asked across industries, regions and company sizes.

The result: cite it as context on what marketers say their priorities and challenges are, but do not set a UK enquiry-to-opportunity target from it. A UK B2B conversion benchmark needs a sample of UK B2B firms and a stated lead definition, and until you find one that passes all four checks, your own CRM trend is the stronger comparison.

How should benchmark figures appear in reporting?

Show three lines side by side: your current period, your previous period, and the external benchmark with its source and date. That stops a reader treating a sector average as a target.

Label every external figure with its origin. A number without a source invites challenge, and a challenged number weakens the report. Keep the source and date beside it.

State plainly when a benchmark comes from a different sector, country or company size, so nobody reads it as a like-for-like comparison.

Where two benchmarks disagree, present both and explain the likely cause. Often it sits in how each study treats multi-touch journeys. Attribution choices change the numbers you compare, a problem covered in the guide to attribution methods.

Common questions

How often should benchmark research be refreshed?

Quarterly suits most teams. Refresh sooner if you change a lead definition, enter a new sector or switch a major channel, because those break comparability.

Can you benchmark without external data?

Yes. Your own historic trend is a valid benchmark, often more useful than a sector average, provided your definitions have not changed. Note any change in the report.

Who should approve a benchmark before publication?

Agree one owner, usually the person accountable for the metric: in an agency the account lead, in-house revenue operations. Approval covers source, date and definition.