Hierarchy data maps how companies connect to their parents, subsidiaries, and affiliates. Kernel AI builds this map from entity resolution across corporate filings, M&A records, and ownership disclosures, then assigns each company a position: the entity itself, its immediate parent, and its global ultimate parent.
A single account rarely stands alone. YouTube reports up through Alphabet. Blizzard reports up through Activision Blizzard to Microsoft. Without a hierarchy layer, a CRM treats YouTube and Alphabet as two unrelated accounts, and a sales team can end up running two separate deals into the same buying committee.
A CRM without hierarchy data cannot answer a basic question: does this account belong to a company we already sell to? That gap shows up in three places.
Territory design. Reps get assigned by account name, not by corporate family. A rep in one territory works a subsidiary while a rep in another territory works the parent, and neither knows the other exists.
Deal routing. A new lead from a subsidiary gets routed as a net-new account instead of an expansion opportunity inside an existing relationship. The deal takes longer to close and the account team never gets credit for the relationship they already built.
CRM accuracy. Duplicate and orphaned accounts accumulate because the system has no concept of corporate structure. Reporting on total pipeline by parent company becomes a manual spreadsheet exercise instead of a query.
Kernel AI resolves this at the entity level, not the account name level, so a subsidiary and its parent are linked automatically instead of requiring a rep to notice the connection.
Once accounts are linked by corporate structure, a few things become possible that were not possible before:
Territory maps that follow ownership, so one team owns a parent and every subsidiary beneath it
Account scoring that rolls up to the parent, so a small subsidiary with high intent surfaces the buying signal to the team already working the parent account
Whitespace analysis across a corporate family, showing which subsidiaries have no coverage at all
Consolidated reporting that separates entity-level activity from the parent's total footprint
These use cases are covered in depth on the Kernel Account Hierarchies page, including how territory allocation and account scoring work against a resolved hierarchy.
Most providers stop at one parent field. Kernel AI tracks three distinct positions for every entity:
Immediate parent. The company that owns this entity directly.
Global ultimate parent. The top of the ownership chain, including holding companies and investment vehicles.
Global operating parent. The top entity that actually runs the business, skipping past holding companies that exist for tax or legal structure alone.
That third position is where Kernel AI differs from Dun & Bradstreet and ZoomInfo. Both report the global ultimate parent and stop there. If the ultimate parent is a holding company with no operations, that field tells a rep nothing useful about who actually runs the account. Kernel AI's global operating parent skips the holding layer and surfaces the entity a rep can actually sell into and reference in a conversation.
This resolution runs on KERN ID, Kernel's entity identifier, which stays stable as companies get acquired, renamed, or restructured. The hierarchy behind it is built and refreshed continuously from the same data pipeline that processes more than 30 billion tokens a day across 1.8 million AI agents, so the parent-child map a rep sees reflects a recent ownership change, not a snapshot from last quarter's data refresh.
Full technical documentation on how entities and parent relationships are resolved is available at docs.kernel.ai/data/hierarchies and docs.kernel.ai/data/hierarchies/parent-relationships.
Corporate hierarchy data maps the ownership relationships between a company and its parents, subsidiaries, and affiliates, so a CRM can treat related entities as one corporate family instead of unrelated accounts.
The global ultimate parent is the top entity in the ownership chain, even if it is a holding company. The global operating parent is the top entity that actually runs the business, skipping past holding or investment layers that exist for legal structure alone.
It fixes territory assignment, deal routing, and CRM accuracy by linking accounts that belong to the same corporate family, so reps and reporting see the real relationship instead of disconnected account names.
ZoomInfo's company universe is anchored on the web domain, so subsidiaries and regional entities that share a corporate domain collapse toward a single record, and accounts with no website or a name and URL mismatch may not resolve at all. Kernel AI is not tied to domains: multiple entities can share a domain and still stay unique through their own KERN ID, and Kernel also resolves a global operating parent, a level ZoomInfo does not provide.
D&B's hierarchy is built on the D-U-N-S Number, assigned per registered site, and covers only registered legal entities using levels like Domestic Ultimate and Global Ultimate. That Global Ultimate is frequently a holding company with no employees or buying center, so it does not show where buying power actually sits, and D&B's linkage refreshes on a monthly batch cycle. Kernel AI adds a global operating parent, a level the D&B model does not have, resolves unregistered and brand-only entities from web and operating signals, and updates continuously rather than on a monthly cycle.
