Data Quality and Reliability Report

Executive Summary

This report defines what can be trusted, what should be qualified, and how to publish insight pages responsibly at scale. High entity volume alone does not guarantee publication quality. Quality comes from coverage depth, intersection depth, identifier reliability, and transparent caveats.

The dataset performs strongly on relationship depth. More than 252k companies are already linked to at least two related source families, and nearly 189k are linked to at least three. This supports large-scale generation of rich pages where each profile can include multi-dimensional context instead of single-field summaries.

Coverage varies by metric type. Structural fields such as country and sector are highly complete, while funding-event amount values are less complete than funding-event counts. The practical rule is simple: counts and rankings can usually be reported with high confidence, while amount totals require explicit coverage notes.

Quality Diagnostics Charts

Company Intersection Depth

How many companies are linked across multiple source families

• >=1 source280,500
• >=2 sources252,299
• >=3 sources188,841
• >=4 sources118,347
• all 9428

FasterCapital

Critical Coverage Ratios

Coverage by field family (counts converted to percentage)

• Funding date93.8%
• Funding amount16.3%
• Company country98.8%
• Company sector97.9%

FasterCapital

Duplicate Name Groups

Text-name duplicates are expected; IDs remain stable

• Investor name groups78,206
• Company name groups6,941

FasterCapital

Production-Ready Profile Tiers

How much data is available for richer pages

• Deep intersection132,349
• Premium tier70,254

FasterCapital

Confidence Framework and Publishing Rules

Publication policy should reflect confidence level. High-confidence fields can appear in headlines and summaries. Medium-confidence fields should remain in chart or analysis sections with caveats. For strict quality control, entity-level pages should pass minimum intersection rules before publication (for example: funding + investors + people + core descriptive fields).

Identifier policy is equally important. Do not merge or deduplicate records by display names alone. Use stable identifiers (`_id`, profile slug, public ID) and treat display names as presentational attributes. This prevents false merges and preserves analytical reliability across large-scale report generation.

• Show coverage notes directly beside amount-based charts.
• Enforce ID-based identity logic for all profile-level pages.
• Use minimum quality tiers before generating deep profile reports.
• Retain methodology and caveat sections on every indexable report page.

Year-Dimension Reliability Note

Year dimension quality is strong when dates are available, which supports robust cycle analysis. For year-based reporting, prioritize date-complete series and avoid constructing hard conclusions from amount-only subsets without indicating coverage ratio.

Related Reports (Quality Tier Exploration)

Quality tiers are designed to control confidence at scale. The fastest way to understand impact is to keep the scenario fixed and vary only the quality tier. The examples below are generated for the most recent years in the dataset (2025, 2024).

FAQ