# Foundational data Kernel's foundational dataset combines the grounding in consistent entity data, detaile entity memory research, and the Kernel research agent to produce highly accurate firmographic data for all entities. Unlike static data vendors, our research actively investigate multiple sources to provide accurate, up-to-date information with full transparency into methodology. ## HQ Determining a company's true headquarters location requires more than pulling a single address field. Kernel's AI agents analyze multiple signals including company websites, regulatory filings, news reports, and business registrations to identify the actual headquarters location. This is particularly valuable for companies with multiple offices, those that have relocated, or where LinkedIn data may be outdated or incorrect. Like all our analyses, the HQ determination comes with transparent reasoning. ## Headcount To determine a company's employee count, we do not rely on a single, static data point. Instead, Kernel triangulates information from multiple primary data sources. This process involves analyzing and comparing data from a company's LinkedIn profile, its official website, public filings, and news reports to arrive at the most accurate figure possible. Like all our analyses, the headcount figure comes with a clear explanation of how it was determined, providing full transparency into our process. If you disagree with our finding, you can use the feedback field in your CRM to flag it for review and have it reviewed within 48 hours. ## Revenue Kernel provides either an observed revenue figure backed by cited sources or a model-based estimate when no public figure exists. Every company receives a revenue figure, a confidence level, and an explanation of how it was derived. #### How it works: 1. **Evidence search**: Our engine searches for reported revenue across public filings (SEC 10-K, Companies House, annual reports), earnings releases, credible news sources, and official company disclosures. When multiple sources exist, they are cross-referenced and the most reliable figure is selected. 2. **Currency normalisation:** Revenue reported in foreign currencies is converted to USD using verified annual average FX rates for the relevant fiscal year. 3. **Forward adjustment**: If the most recent reported figure is from a prior fiscal year, it is adjusted forward to the target year using observed company and industry growth signals. The target year is always the most recently completed fiscal year. 4. **Model-based estimation**: When no public revenue figure can be found, our proprietary model estimates revenue using the company's headcount, industry classification, revenue model, operating model, geography, company maturity, and funding profile - among other signals. These estimates are clearly labelled as model-based. #### Definition: Revenue is defined as annual consolidated total revenue - the GAAP or IFRS top line. Accepted labels include total revenue, net revenue, net sales, turnover, and operating revenue. #### Transparency and feedback: Every revenue figure includes reasoning that explains the source, the methodology, and any adjustments. This allows your team to understand the basis for the number. If you have information that suggests a different figure, the built-in feedback mechanism lets you request a review and correction.
{% embed url="" %}