office, business, notebook, statistics, technology, digital, computer, internet, marketing, accounting, corporate, job, communication, creative, table, meeting, businessman, ideas, woman, project, coworking, brainstorming, loft, plan, documents, discussion, group, man, team, mobile, mock, work, laptop, professional, startup, management, teamwork, network, account, paperwork, hipster, young, cooperation, object, papers, global, millennials, living room, sofa, blue business, blue computer, blue office, blue technology, blue laptop, blue work, blue meeting, blue paper, blue marketing, blue team, blue mobile, blue network, blue community, blue room, blue internet, blue digital, blue table, blue group, blue job, blue company, blue conference, blue global, blue businessman, blue teamwork, blue plan, blue planning, blue creative, blue document, blue accounting, blue idea, blue management, statistics, statistics, marketing, marketing, marketing, accounting, accounting, accounting, accounting, accounting, project, management
Photo by Goumbik on Pixabay

Measurement

Part of When to review customer success software measurement

How to brief customer success software benchmark research

Customer success software benchmark research works only when the decision, sample and metric are fixed before fieldwork, and results are reported with ranges.

Editorial Desk · · 5 min read

What to take away

  • Decide what the customer success software benchmark will change — a renewal, a target or a budget line — before you commission it.
  • Name the platforms and the decision each one informs, rather than benchmarking an unnamed market.
  • Weight evidence quality above headline percentages when you compare studies, and say so in the brief.
  • Fix the sample definition and the metric formula before fieldwork starts, not during the write-up.
  • Give the study a shelf life by setting the re-run date on the day you publish.

What should a benchmark study prove before you fund it?

A benchmark earns its budget when it changes a decision. If the output is a chart nobody opens during a renewal, the work was a research exercise rather than an input to planning. Write the decision on a single page before you commission anything.

Most studies fall into one of three jobs: setting a target, testing a supplier claim, or sizing a market. Each job needs a different sample. A target-setting study can lean on your own book of business, while testing a supplier claim needs comparable accounts held elsewhere.

Where a study touches personal data about named accounts, accountability under the UK GDPR applies from the design stage. You need a lawful basis before account data enters the study; legitimate interests is often the basis teams rely on for benchmarking their own book, and it needs a recorded assessment. The ICO's guidance on accountability and governance explains how to record those decisions and show compliance, which is the same discipline a benchmark needs.

If the study feeds a board pack, our guide to customer success software measurement and reporting sets out how to keep metric definitions stable between quarters so comparisons hold.

How do you score the evidence so it survives scrutiny?

Set the rubric before you see any results. Weights force a conversation about what matters, and they stop the loudest number winning by default.

Scoring rubric criteria and weights

  • Sample definition20% weight
  • Metric definition20% weight
  • Source transparency15% weight
  • Comparability15% weight
  • Repeatability15% weight
  • Commercial independence15% weight

Score each criterion from 1 to 5, then multiply by the weight. Two reviewers scoring the same study should land within a few points of each other. Where they do not, the criterion description is too loose.

Set a minimum score for every criterion, not only for the total. A strong sample cannot rescue a metric definition that nobody can reproduce.

Name the products before you score anything. A customer success software benchmark a UK team can act on covers named platforms — Gainsight, ChurnZero, Totango, Planhat, Vitally or Catalyst — on the same accounts and the same metric definitions. An unnamed 'market average' cannot be checked, and it should not reach a renewal meeting.

Score the metric, not only the study. Net and gross revenue retention are different measures: net revenue retention is a metric that reflects a company's ability to retain and grow revenue from its existing customer base, calculated as NRR = ((Recurring Revenue at End of Period + Upsell and Expansion Revenue During the Period - Revenue Churn During the Period) ÷ Recurring Revenue at Start of Period) × 100, while gross revenue retention excludes any upsell, cross-sell or expansion revenue; in 2025 the median NRR across all software-as-a-service (SaaS) companies was 102% and the median GRR was 91%.

Vendor material such as Agentforce customer stories shows how a platform behaves in named deployments. Score it as context rather than independent evidence, because the supplier chose which accounts to publish.

Which data sources keep a benchmark credible?

Aim for two independent sources behind each claim. Internal telemetry gives you definitions you control. A survey reaches accounts you cannot see. Neither is sufficient alone for a market-wide statement.

Official guidance works as a neutral baseline for expansion questions. The GOV.UK pages on growing your business cover finance, marketing and expansion without promoting a product.

Keep the recruitment record. Who was approached, who declined and when the fieldwork ran should all be reconstructable by someone who was not on the project.

When a benchmark sits beside revenue figures, the causation problem appears: our guide to customer success software attribution methods sets out how to separate influence from contribution before you publish a retention claim.

How do you report benchmark results without overclaiming?

Report a range with the sample attached. For example, a report stating that between 18 and 26 per cent of mid-market accounts renewed early, from 140 responses, beats a single figure with no denominator.

Report ranges, not single figures

18-26% renewed early

From 140 responses

Date every number you quote

Name the weakest segment

Date every number you quote. A 2024 median repeated in a 2026 board pack misleads even if it was accurate when collected.

Name the weakest segment. Aggregate results hide the accounts where the product struggles, and those accounts usually drive churn.

Keep one page of limitations next to the findings. Reviewers trust a study more when the authors state what it cannot show.

Common questions

How many responses does a benchmark need?

How many responses a benchmark needs depends on the decision and the number of segments you plan to report. A study cut into six segments needs far more responses than one reporting a single total.

Should you publish the full methodology?

Publish enough that a sceptical reader could repeat the study. That usually means the sampling method, field dates, question wording and the metric formulas.

How often should a benchmark be repeated?

Annual is a sensible default for pricing and retention metrics, because supplier packaging moves. Move to quarterly if you are tracking a live renewal cohort.

Filed undercustomer success softwarebenchmark researchmeasurementvendor due diligence

More in Measurement