What benchmarks do
Benchmarks compare observed, source-backed company signals with a relevant cohort so reviewers can make a relative comparison.
BENCHMARK GUIDE
How to read company benchmarks
Benchmarks compare observed company signals with an appropriate cohort. They support review judgement, but they are not credit scores, risk ratings or standalone decisions.
Search is the entry point.Review is the outcome.
01What this page explainsBenchmark dimensions, cohort selection, coverage, example companies and the caveats that keep the view in review context.
02Who it is forReviewers using Review Workspace to compare observed company signals with an appropriate cohort.
03After readingYou should know how to interpret the benchmark snowflake, cohort, dimensions, example companies, coverage and caveats before relying on the view.
Quick answer
At a glance
The short version reviewers should understand before reading the detailed methodology.
What they do not do
They do not make credit decisions, risk ratings, compliance decisions or claims of complete truth about a company.
What users should verify
Check source evidence, cohort fit, freshness, coverage notes and review context before relying on the benchmark view.
Method
How it works
Simple steps first, with the detailed source and caveat text below.
- 01
Read the snowflake as shape, not verdict
The benchmark snowflake shows relative differences across dimensions. It helps reviewers see where a company looks denser, thinner or similar to the cohort without turning that pattern into a score.
- 02
Confirm the cohort
Benchmarks depend on the selected comparison group. Use the most specific cohort that still has enough companies to support a meaningful comparison.
- 03
Interpret each dimension as an observed signal
Dimensions summarise source-backed evidence such as regulatory depth, organisational scale or web activity. They are relative comparison aids, not conclusions.
- 04
Check coverage before weighting the view
Coverage varies by source availability, company disclosure, matching confidence and refresh timing. Sparse or no-match dimensions should stay visible rather than disappearing.
- 05
Use examples and source evidence to follow up
Reference companies and linked evidence help reviewers understand why a benchmark looks the way it does and what should be inspected next.
Product surfaces
What users see in the product
The methodology connects directly to surfaces where users inspect evidence or record review work.
Review WorkspaceThe workflow area for review cases, checks, owners, notes, monitoring and evidence links.Evidence modeA company-profile view for inspecting source-backed rows, grouped evidence and coverage notes.Company OverviewThe starting view for identity, filings, status, source coverage and profile context.Operational FootprintMapped and unmapped location evidence, address/place context and operational signals where covered.
Labels
What the labels mean
Short definitions for terms that appear across evidence, footprint and review views.
| Label | Meaning for reviewers |
|---|---|
| Benchmark snowflake | The multi-dimension visual shape for the selected company against its comparison cohort. |
| Cohort | The selected group of companies used as the comparison baseline. |
| Dimension | A grouped area of observed company signals shown as a relative comparison. |
| Coverage | How much source-backed evidence supports the visible benchmark dimensions. |
| Reference company | An illustrative company drawn from the cohort to help explain the comparison context. |
| Sparse or no-match dimension | A visible limit showing thin or absent observed evidence in the covered view. |
Detail
Method detail and caveats
Use these sections when you need the source-level detail behind the quick answer.
Benchmark snowflake purpose
The benchmark snowflake is a multi-dimension view of observed company signals against a selected cohort. It is designed to show relative shape across covered dimensions, not to deliver a single verdict.
Use it to understand where a company looks broadly similar to its cohort and where the visible, source-backed evidence appears denser, thinner or meaningfully different. It should guide review attention rather than replace source inspection.
Benchmarks show observed signals against a selected reference cohort. They are a review aid, not a scoring system or compliance conclusion.
How cohort selection works
Benchmarks are only meaningful relative to the cohort shown on screen. The preferred comparison uses the most specific available cohort that still meets minimum cohort-size rules, so the baseline stays relevant without becoming too thin.
Depending on the company and available lens, the comparison may be based on SIC, SIC section, region, size or a broader active-company population. Changing the cohort changes the relative comparison, not the underlying company evidence.
Most specific lens first
TradeDesk favours the narrowest relevant cohort that still has enough companies to support a useful comparison.
Fallback cohorts stay explicit
If a very narrow cohort is not viable, the comparison can step back to a broader but still visible cohort such as SIC section, region, size or the wider active-company population.
Benchmark dimensions
Dimensions group related observed signals so reviewers can compare broad themes rather than isolated rows. Each dimension should be read as a relative comparison backed by covered sources, not as a final statement about the company.
Regulatory depth and enforcement exposure
These dimensions compare visible regulatory and enforcement-related signals where covered sources support them. A denser pattern should prompt evidence review, not a risk conclusion.
Public-sector dependency and change velocity
These dimensions compare observed public-sector activity and the pace of visible company change. The signal should be read alongside sector norms and source timing.
Web activity and corporate complexity
These dimensions compare observable digital presence and structural complexity. Neither is a standalone proxy for legitimacy, stability or concern.
Organisational scale and financial profile
These dimensions compare visible size and financial characteristics where covered data supports them. Sparse evidence can narrow what the view can say.
Innovation, IP footprint and social impact
These dimensions compare observed innovation-related records, intellectual-property context and social-impact signals where covered. Presence or absence often reflects business model as much as performance.
Coverage and sparse dimensions
Coverage varies by source availability, company disclosure, matching confidence and refresh timing. Lower coverage means the benchmark rests on a thinner evidence base, not that a company is better or worse.
Sparse or no-match dimensions should remain visible so reviewers can see where the comparison is thin. Those limits should not be silently hidden or treated as proof that the underlying activity is absent.
Coverage varies. Sparse or no-match dimensions are shown so evidence limits stay visible.
Cohort examples
Reference companies are drawn from the selected cohort to help reviewers understand the comparison context. They are examples, not endorsements or claims that the companies are direct competitors, suppliers or customers.
Use cohort examples to sanity-check the cohort fit and to understand what kinds of company profiles are shaping the comparison baseline.
Caveats and review context
Benchmarks are based on observed records and derived features from covered sources. They should be read alongside source evidence, review notes and the commercial or policy context of the review.
Interpret with extra care when a company is recently incorporated, dormant, dissolved, unusually structured, thinly disclosed or represented by a cohort that is itself narrow or mixed.
Review aid only. Not a scoring system, compliance conclusion or recommendation.
Refresh cadence
Benchmark rows and cohort statistics refresh as source coverage updates. The visible benchmark reflects the latest available TradeDesk view for the selected company and comparison lens, but different source families can update on different timetables.
If timing matters to the review, inspect the linked source evidence and note any freshness concerns in the review record.
Limits and caveats
What the view cannot prove on its own
These limits keep methodology pages readable without hiding uncertainty.
Benchmarks support review judgement; they do not replace source evidence, policy checks or business context.
Changing the cohort changes the comparison, not the underlying company evidence.
Coverage varies by source, disclosure, matching confidence and refresh timing.
Sparse or no-match dimensions should be read as visible evidence limits, not proof that a company lacks the underlying activity.
Reference companies are illustrative cohort examples, not endorsements, peers in every commercial sense or trading relationships.
FAQ
Common questions
Are benchmarks scores or ratings?
No. Benchmarks are relative comparisons of observed signals against a selected cohort. They are not credit scores, risk ratings or compliance decisions.
Does a denser benchmark decide the outcome?
No. The visible shape depends on business model, sector, cohort selection, coverage and source timing. It should prompt evidence review rather than determine the outcome.
Why does the cohort matter?
Because the benchmark only makes sense relative to the comparison group shown on screen. A different cohort can produce a different relative picture.
What should I do if coverage is limited?
Read the linked evidence, check cohort fit and source dates, and treat the comparison cautiously. Limited coverage means less can be said with confidence.
Related pages
Keep reading by review question
How TradeDesk WorksThe product model, evidence links and review workflow.Data Sources & CoverageWhat sources are covered, what no match means and why recency varies.AI TransparencyWhere AI may assist and how users should verify outputs.Evidence MethodologyHow scattered source records become review-ready evidence.Operational Footprint MethodologyHow mapped locations, source rows, group scope and footprint caveats should be read.Review Workspace GuideHow reviews, checks, owners, notes and evidence links work together.Corrections & DisputesHow to ask TradeDesk to review something that appears wrong or incomplete.Trust CentreSecurity, AI transparency, legal and governance resources.