[GTM glossary](/glossary.md) Entity resolution

Contextual Intelligence & AI

# Entity resolution

In short

Entity resolution is the process of identifying and merging records that describe the same real-world entity, such as one company that appears as "Acme Inc.", "ACME Incorporated", and "acme.com" in different systems. It is also called record linkage or identity resolution.

For GTM teams it is the foundation under reporting, routing, and AI. If one account is split into three records, scores are wrong, reps collide, and agents reason over fragments.

## How entity resolution works

### 1. Standardize

Clean names, domains, addresses, and country codes into consistent formats.

### 2. Block

Narrow the comparisons to plausible candidates, such as records sharing a domain or similar name, so you do not compare every pair.

### 3. Match

Score pairs using rules (same domain) or models that weigh name, location, and other fields.

### 4. Merge or link

Create one golden record, or keep records and link them under a shared ID. Linking is safer when you may need to undo a decision.

---

## Hard cases in B2B

* **Parents and subsidiaries:** Decide whether a subsidiary is its own account or part of a parent.
* **Shared names:** Many unrelated companies have similar names; domains and locations disambiguate.
* **Rebrands and acquisitions:** The same company can change name and domain.
* **Personal vs. work emails:** People appear under several addresses.

---

## Example

Say your CRM has 1,200 accounts, and enrichment shows 150 of them are duplicates of others. After resolution, open opportunities, contacts, and activity roll up under one account, and a rep no longer emails the same company twice in one week.

---

## Common mistakes

* **Matching on name alone:** This creates false merges that are painful to unpick.
* **Merging irreversibly:** Keep source IDs and a merge log.
* **Doing it once:** New records arrive daily; run matching continuously.
* **Ignoring hierarchy:** Without parent links, territory and ABM reporting break.

---

## Entity resolution FAQs

**What is entity resolution in data?**

Entity resolution is deciding which records across databases refer to the same real-world entity, such as one company or person, and linking or merging them under one identity. It removes duplicates and connects fragmented data.

**Why does entity resolution matter for CRM data?**

Duplicate and fragmented records split activity, distort scoring and reporting, and cause reps to contact the same account twice. Resolving them gives each account one history, which improves routing, analytics, and any AI using the data.

**What is the difference between entity resolution and deduplication?**

Deduplication removes repeated records within one dataset. Entity resolution is broader: it also matches records across different systems and formats, links related entities, and often keeps a persistent shared identifier rather than just deleting rows.

**How do you match company records without a shared ID?**

Standardize names, domains, and locations, then compare candidates using rules or a trained model. Domain is usually the strongest key. Use several fields together, set a confidence threshold, and send uncertain pairs to human review.

Share term

---

## More related phrases

* [Golden record](/glossary/golden-record.md)
* [Business ontology](/glossary/business-ontology.md)
* [Data deduplication](/glossary/data-deduplication.md)

## Similar phrases

* [Knowledge graph](/glossary/knowledge-graph.md)  
A knowledge graph stores entities such as companies and people as nodes and their relationships as links, giving search and AI a structured, queryable map of facts.
* [Context graph](/glossary/context-graph.md)  
A context graph links accounts, people, events, and decisions into one connected map, so GTM teams and AI agents can see how each fact relates to the others.
* [Context layer](/glossary/context-layer.md)  
A context layer is the shared data and meaning layer between your systems and your AI tools, giving them consistent definitions, relationships, and fresh facts to work from.
* [AI hallucination](/glossary/ai-hallucination.md)  
An AI hallucination is output from a language model that sounds confident and plausible but is false, invented, or unsupported by any real source, a real risk in sales research and outreach.
* [Data enrichment](/glossary/data-enrichment.md)  
Data enrichment is the process of adding missing or updated information, such as firmographics, contact details, and technographics, to records you already have from internal or external sources.