---
title: "Entity resolution guides and playbooks"
summary: "19 articles on entity resolution: what it is, how accuracy is measured, what each company identifier identifies, and how to deduplicate a file."
canonical_url: https://entitybind.com/guides
last_updated: 2026-10-15
copy: full
---

# Entity resolution guides and playbooks

> 19 articles on entity resolution: what it is, how accuracy is measured, what each company identifier identifies, and how to deduplicate a file.

Canonical page: https://entitybind.com/guides
Last updated: 2026-10-15

## Learn entity resolution well enough to defend a merge

19 articles for the person who has to sign off the merge. Read them and you can take a vendor's accuracy claim apart in five minutes, say what an LEI proves and what it does not, and work a CRM export or a supplier master down to a merge list you would defend in a review. They split two ways: 9 on the field — measurement, matching, calibration — and 10 beside a real file, with the playbooks, the identifiers and the dated rules that all fail on the same thing. Start with a reading path below, or take the shelf that matches what you have open.

- 19: articles — 9 on the field, 10 on the file open in front of you
- 105: questions answered across the library
- 4: reading paths, three articles each, about half an hour
- 0: false binds in the 317-company recording of 2026-08-29 that these articles quote

## Concepts

What the field is, how accuracy is measured, and what a published rate has to carry before it means anything. Written to be checked rather than believed: the arithmetic is on the page and the failing numbers are beside the winning ones.

- [Choose a company matching tool you can defend](https://entitybind.com/guides/entity-resolution-buyers-checklist) (Foundations): Evaluate company matching on your records. Download a scorecard for identifier conflicts, unresolved rows, evidence, coverage, controls and pricing. (7 min read)
- [What is entity resolution?](https://entitybind.com/guides/what-is-entity-resolution) (Foundations): Entity resolution decides which records refer to the same real-world thing. What it is, why company identity is hard, deterministic versus probabilistic matching, and why fuzzy matching is not entity resolution. (9 min read)
- [Entity resolution vs record linkage: which term is right](https://entitybind.com/guides/entity-resolution-vs-record-linkage) (Foundations): Record linkage, entity resolution, deduplication, entity matching, entity linking and identity resolution name overlapping problems from different fields. What each term means and when each one is correct. (8 min read)
- [How to measure entity resolution accuracy](https://entitybind.com/guides/how-to-measure-entity-resolution-accuracy) (Measurement): Precision, recall, abstention, pairwise versus B-cubed cluster metrics, Wilson intervals and out-of-fold evaluation. Why a single accuracy number for entity resolution is always meaningless, with worked arithmetic. (13 min read)
- [Precision, recall and abstention](https://entitybind.com/guides/precision-recall-and-abstention) (Measurement): Entity resolution is a three-way trade, not a two-way one. Why abstention is a design choice rather than a failure, what an abstention has to carry to be useful, and how to measure a system that declines. (8 min read)
- [Fellegi-Sunter explained, with worked arithmetic](https://entitybind.com/guides/fellegi-sunter-explained) (How matching works): The probabilistic record linkage model behind most matching engines: m and u probabilities, match weights, the three decision regions, and a worked example you can check with a calculator. (10 min read)
- [Calibration and confidence scores](https://entitybind.com/guides/calibration-and-confidence-scores) (How matching works): What a calibrated confidence actually means, how expected calibration error is computed, why a similarity score is not a probability, and why an LLM's self-reported confidence is neither. (10 min read)
- [Blocking and candidate generation](https://entitybind.com/guides/blocking-and-candidate-generation) (How matching works): You cannot compare every pair. How blocking keys work, why rarity weighting matters, how to measure retrieval with recall@k, and the silent failure mode that mints duplicates. (9 min read)
- [Survivorship and golden records](https://entitybind.com/guides/survivorship-and-golden-records) (After the match): Merging records is only half the job. Field-level survivorship, source precedence, temporal facts, per-field provenance, and why last-confirmed beats first-seen. (10 min read)

## Applied

The same engineer two weeks later, with a CRM export or a supplier master open. Playbooks including the steps that are somebody else's job, what each company identifier actually identifies, and the dated rules whose failure mode is an identity problem wearing a compliance hat.

- [How to deduplicate a CRM](https://entitybind.com/guides/how-to-deduplicate-a-crm) (Playbooks): Export, preview, tier the file, run it, review the abstentions, merge with receipts, measure. The full sequence for deduplicating a CRM account object without making a wrong merge you cannot undo. (12 min read)
- [How to deduplicate a supplier master](https://entitybind.com/guides/how-to-deduplicate-a-supplier-master) (Playbooks): Duplicate vendors, payment leakage, remittance conflicts and parent concentration. The accounts payable playbook for finding which vendor records are one company, with a citation for every finding. (12 min read)
- [Entity resolution for data analysts](https://entitybind.com/guides/data-analyst) (Playbooks): Profile the columns, normalise without manufacturing merges, resolve, work the exceptions, then ship a crosswalk and a golden-record view. The warehouse-side playbook for company rows with no shared key. (25 min read)
- [Company identifiers explained](https://entitybind.com/guides/company-identifiers-explained) (Identifiers): Company number, LEI, CIK, ticker, SIREN and SIRET, DUNS, VAT, EUID and ISIN. What each identifies, who issues it, what it does not guarantee, and which ones actually decide a match. (13 min read)
- [What is an LEI?](https://entitybind.com/guides/what-is-an-lei) (Identifiers): The Legal Entity Identifier explained: the twenty-character structure and its check digits, GLEIF and the LOUs, Level 1 and Level 2 data, annual renewal and lapse, and what a lapsed LEI means for matching. (11 min read)
- [Entity resolution for AI agents](https://entitybind.com/guides/entity-resolution-for-ai-agents) (Agents): An agent about to pay or contract a counterparty needs a decision with a receipt, not a similarity score. Why abstention is the safe default for an autonomous caller, and what the MCP surface exposes. (11 min read)
- [Entity binding failures in AI agents](https://entitybind.com/guides/entity-binding-failures-in-ai-agents) (Agents): An entity binding failure is an agent calling the right tool on the wrong real-world entity. What a 2026 study of 1,800 runs found, and four safeguards. (8 min read)
- [KYB data requirements](https://entitybind.com/guides/kyb-data-requirements) (Regulatory deadlines): What a know-your-business onboarding flow actually needs, field by field, and which step is identity resolution. Explicitly not a screening, scoring or eligibility tool. (11 min read)
- [ISO 20022 structured address requirements](https://entitybind.com/guides/iso-20022-address-requirements) (Regulatory deadlines): CBPR+ payment messages are to stop accepting fully unstructured postal addresses. Swift delayed the 14 November 2026 date on 27 August 2026. What changes, and the identity work it forces. (11 min read)
- [The DORA register of information](https://entitybind.com/guides/dora-register-of-information) (Regulatory deadlines): What the register demands, why only 6.5% of almost 1,000 dry-run registers passed all 116 data quality checks, and the provider identity problem that explains most of the failures. (11 min read)

## Four ways in, depending on what you have open

Each path runs three articles in the order they answer one problem, and takes about half an hour.

### You are auditing somebody's accuracy claim

Three articles and you can say what a published rate leaves out: the denominator, the interval, and which of the two metrics a vendor picked when they published only one.

1. [How to measure entity resolution accuracy](https://entitybind.com/guides/how-to-measure-entity-resolution-accuracy) (13 min read)
2. [Precision, recall and abstention](https://entitybind.com/guides/precision-recall-and-abstention) (8 min read)
3. [Calibration and confidence scores](https://entitybind.com/guides/calibration-and-confidence-scores) (10 min read)

### You have a file with duplicates in it

A CRM export or a supplier master, and a merge you cannot take back. The playbooks run in order and include the steps that are somebody else's job.

1. [How to deduplicate a CRM](https://entitybind.com/guides/how-to-deduplicate-a-crm) (12 min read)
2. [How to deduplicate a supplier master](https://entitybind.com/guides/how-to-deduplicate-a-supplier-master) (12 min read)
3. [Survivorship and golden records](https://entitybind.com/guides/survivorship-and-golden-records) (10 min read)

### You are building or buying the matcher

What the problem actually is, the scoring model most of the field is built on, and the step that picks which pairs get scored at all — which quietly caps every accuracy number above it.

1. [What is entity resolution?](https://entitybind.com/guides/what-is-entity-resolution) (9 min read)
2. [Fellegi-Sunter explained, with worked arithmetic](https://entitybind.com/guides/fellegi-sunter-explained) (10 min read)
3. [Blocking and candidate generation](https://entitybind.com/guides/blocking-and-candidate-generation) (9 min read)

### You have a dated obligation and a register to file

Three rules that fail on the same thing: being unable to say which company a record refers to. Each one is attributed to the body that publishes it, with its dates.

1. [KYB data requirements](https://entitybind.com/guides/kyb-data-requirements) (11 min read)
2. [ISO 20022 structured address requirements](https://entitybind.com/guides/iso-20022-address-requirements) (11 min read)
3. [The DORA register of information](https://entitybind.com/guides/dora-register-of-information) (11 min read)

### Written for

Anyone who owns a company master and has to defend it — a stewardship queue, a CRM after a migration or a merger, a supplier file before a payment run, or counterparty onboarding that has to survive an audit. No account, no call in between.

### Not covered here

Entity Bind resolves companies and organisations only. It does not resolve people, vessels, bank-account ownership, sanctions exposure, emissions or geolocation, and its results may not be used for eligibility decisions about individuals, including employment, credit and tenancy.

Self-service. Create your workspace online and verify your email to start matching company records. Free starts with 300 credits and no card. Paid monthly plans are available in workspace Billing; VAT is added where applicable. Usage is tracked without automatic overage charges. [Trust and security](https://entitybind.com/trust)

Or start with your own rows. Create your Free workspace, verify your email and upload a company CSV to inspect its results. Contact us if you want an assisted measurement beyond the shallow Free plan. [Start free →](https://app.entitybind.com/app?signup=1)

## No figure without its denominator, its interval and its fixture.

### Every percentage shows its fraction

A percentage with no denominator is not a measurement. Every rate in these articles is quoted with the fraction behind it, the 95% interval where one is published, and the fixture it was recorded on — 317 live GB companies for the per-tier table, 1,176 labelled company pairs for the gate.

### Two recordings, never averaged

The live-adjudicated recording of 2026-08-29 and the continuous-integration gate of 2026-09-05 measure different things over different fixtures. Averaging them would produce one rounder number and a false one.

### A precision figure needs its automation rate

The gate's precision is 1.0000, on 357/357 automated merges, and it never appears on its own. It decides 63.10% of its pairs without a person, and on its own record-level fixture the groups it forms match the real ones at 0.7584 — its cluster F1. Perfect precision at no automation is what you get for deciding nothing at all.

Historical results on an internal UK fixture with a preloaded catalogue. The September 5 replay has no full-prompt-verified calls and cannot establish current calibration. No independent customer holdout or global coverage claim.

[Every figure, its population and what it does not show →](https://entitybind.com/accuracy)

The same rule holds for a number about your own file. Every figure that comes back names the rows it was measured over. [Start free →](https://app.entitybind.com/app?signup=1)

## Shorter than an article, and longer than one

Three families either side of this library: one sentence per word, one page per alternative, and one answer per question.

- [One sentence per word](https://entitybind.com/glossary): The field's vocabulary one term at a time: the definition first, then what Entity Bind does about it — including the techniques it does not implement.
- [The named alternatives, with their wins counted](https://entitybind.com/compare): Entity Bind beside the alternatives, with the rows where the alternative wins counted rather than omitted, and the one category this product refuses to be sold into.
- [One answer per question](https://entitybind.com/faq): Every question this site answers, clustered by the decision it belongs to, each with its answer in the page rather than behind a search.

## The basis for every figure in this library

Internal fixtures only. The historical 2026-08-29 per-tier recording covers 317 GB companies with identifier-rich arrivals seeding the catalogue first; the context tier includes country and registered-town prose. Its legacy replay has no full-prompt-verified calls and cannot certify current calibration. The September 5 CI reports cover 1,176 labelled pairs and 705 records over 250 companies using offline stubs. These are not independent customer holdouts or live production validation; the 99% minimum-input product target remains unmet. Scope is company and organisation identity only — not people, vessels, bank-account ownership, sanctions or emissions. Results may not be used for eligibility decisions about individuals, including employment, credit and tenancy.

Figures read off the product repository on 2026-09-05. Entity Bind holds no SOC 2, no ISO 27001 and no other certification, and claims none. Where an article quotes somebody else’s figure — a vendor’s price, a regulator’s dry-run result — it is attributed in the body as theirs and has not been audited by us.

## Your file, measured the same way.

Create your workspace, verify your email and run your first company file. The Free plan needs no card. Create API keys in your workspace when you need an integration.

[Start free](https://app.entitybind.com/app?signup=1) · [Plans and prices](https://entitybind.com/pricing)

300 credits. No card.

For a guided review of your file, explore [the Merge Safety Sprint](https://entitybind.com/audit-pack).

## The questions these articles open with

One question from each of 8 articles, in the words the article uses. All 105 are answered in full on the pages themselves.

### What is entity resolution in simple terms?

It is working out which records refer to the same real-world thing, and joining them under one identifier. If your CRM has "Acme Ltd", "ACME Limited" and "Acme Holdings Ltd", entity resolution decides which of those are the same company and which is a different legal entity that happens to share a name stem. The output is a set of clusters with stable identifiers, not a list of similarity scores.

### Is record linkage the same as entity resolution?

They solve the same problem and come from different traditions. Record linkage comes from statistics and public health, usually means joining two defined files, and is measured over pairs. Entity resolution comes from the database community, assumes multiple or continuously arriving sources, produces a persistent partition with stable identifiers, and is measured over clusters. If you have two files, say linkage. If you need canonical entities, say resolution.

### Why is a single accuracy number for entity resolution meaningless?

Because the population of possible decisions is every pair of records and almost all of them are non-matches. A system that says "not a match" to everything scores above 99.99% accuracy on a ten-thousand-record file while resolving nothing. Accuracy is not a slightly weak metric here, it is a disqualified one. What you want instead is precision on automated decisions, the automation rate it was drawn from, cluster metrics, an interval, and the input shapes it was measured over.

### Is abstention just a way of hiding a weak model?

It can be, and the defence is publishing both numbers. A system that abstains on everything hard shows high precision and a low automation rate, and the two figures together make that visible immediately. Perfect precision at zero automation is trivial, which is why neither number is ever published here without the other. What abstention is legitimately for is the population where the evidence genuinely does not discriminate — a truncated name whose missing tokens were the ones that mattered cannot be resolved by a better model.

### What does a calibrated confidence score mean?

It means the score is a frequency you can act on: among all the answers returned at 0.90, about 90% are correct. Calibration is a property of a population rather than of an individual answer, so a calibrated confidence has to name the population it was derived from and its size. A score with no stated population cannot be checked and should not be thresholded against.

### Is a company registration number unique?

Within its own register, yes. Within a country, usually not, and that is the assumption that produces wrong merges. A company number plus a country is not a unique key and never was: 1,071 registers serve 232 jurisdictions, 220 of those jurisdictions run more than one, German numbers are qualified by the court that holds them, and Poland runs three parallel number spaces. Treating the pair as unique produces false positives, which is why the qualifier is carried on the record. The fix is to carry the issuing authority alongside the number rather than to infer it from a country column.

### What does LEI stand for?

Legal Entity Identifier. It is a twenty-character alphanumeric code defined by ISO 17442 that identifies a legal entity participating in a financial transaction. Global: roughly 3.4 million LEIs assigned across all registration statuses, about 2.0 million of them with registration status ISSUED and most of the rest LAPSED (GLEIF, golden copy of 2026-09-25), issued by GLEIF, which issues them through accredited Local Operating Units, and the reference dataset is published openly.

### Will you write the merges back into our CRM?

No. Automatic CRM and ERP writeback is explicitly off the roadmap rather than a gap on the way to being closed. We propose the clusters in your own record ids, with a receipt per decision, and you apply them under your own change control in the system that owns the record. A tool that writes irreversible merges into a production CRM on the strength of its own confidence is one whose failures you hear about from a customer.

---

The 0.90 confidence floor. It is an acceptance contract, not calibration evidence: it says what we refuse to return, not how often an accepted match is right. Read it with the coverage it costs.

This is the Markdown copy of https://entitybind.com/guides, assembled from the same data the page renders and carrying every sentence, table row, figure and question the page's main content carries. The HTML page is canonical; it adds the layout, the navigation and the interactive parts (the ask panel, the forms), which a copy can name but not run. Every page on the site, with a line on each: https://entitybind.com/llms.txt.