One customer in two systems with no shared id

Goal Match ERP customers and CRM contacts, without ever merging by chance.

An ERP, a CRM, no shared id. Email matches the records. A shared mailbox points at two customers.

matching by a written rule;
two candidates: an ambiguity, not a merge;
a human decides, and the decision holds on replay.

lumnik Open
Master data
Sales
Connector
Entity

erp-accounts.csv

account_code,company_name,email,city
AC-100,Atelier Boréal,commandes@atelier-boreal.example,Lille
AC-200,Métal Azur,achats@metal-azur.example,Nice
AC-300,Rhône Filtration,compta@groupe-rhone.example,Lyon
AC-301,Rhône Levage,compta@groupe-rhone.example,Lyon
AC-400,Capteurs Armor,contact@capteurs-armor.example,Rennes

crm-contacts.csv

contact_code,display_name,email,phone
CT-9001,ATELIER BORÉAL,Commandes@Atelier-Boreal.example,+33 3 20 00 00 01
CT-9002,Métal Azur,achats@metal-azur.example,+33 4 93 00 00 02
CT-9003,Groupe Rhône,compta@groupe-rhone.example,+33 4 72 00 00 03
CT-9004,Transmission Nord,,+33 3 28 00 00 04
CT-9005,Pneumatique Atlas,ventes@pneumatique-atlas.example,+33 5 61 00 00 05

manifests/10-erp-accounts.yaml

apiVersion: connectors.lumnik.io/v1kind: Connectormetadata:  name: scenario-resolution-erp  connector_type: csv-file  scopes: [entity-resolution]spec:  transport:    kind: local    path: /tmp/lumnik-scenario/entity-resolution/erp-accounts.csv  parser:    has_header: true  after_process: none

manifests/20-crm-contacts.yaml

apiVersion: connectors.lumnik.io/v1kind: Connectormetadata:  name: scenario-resolution-crm  connector_type: csv-file  scopes: [entity-resolution]spec:  transport:    kind: local    path: /tmp/lumnik-scenario/entity-resolution/crm-contacts.csv  parser:    has_header: true  after_process: none

$ lm validate -f manifests/10-erp-accounts.yaml
✓ 10-erp-accounts.yaml — valid

$ lm apply -f manifests/10-erp-accounts.yaml
Applied connector "scenario-resolution-erp" — 1 endpoint

$ lm validate -f manifests/20-crm-contacts.yaml
✓ 20-crm-contacts.yaml — valid

$ lm apply -f manifests/20-crm-contacts.yaml
Applied connector "scenario-resolution-crm" — 1 endpoint

$ lm connector run scenario-resolution-erp
run scheduled: 6afa0588-9d94-4533-be68-5621c1e4d4a2

$ lm run get 6afa0588-9d94-4533-be68-5621c1e4d4a2
ID        CONNECTOR                ENDPOINT  STATUS     IN  OUT  SKIPPED  DEAD-LETTERS  STARTED           ENDED
6afa0588  scenario-resolution-erp  default   Completed  5   5    0        0             2026-11-03 16:24  2026-11-03 16:24

$ lm connector run scenario-resolution-crm
run scheduled: b37554a1-10c3-49ec-831a-aaf04d166c6b

$ lm run get b37554a1-10c3-49ec-831a-aaf04d166c6b
ID        CONNECTOR                ENDPOINT  STATUS     IN  OUT  SKIPPED  DEAD-LETTERS  STARTED           ENDED
b37554a1  scenario-resolution-crm  default   Completed  5   5    0        0             2026-11-03 16:24  2026-11-03 16:24

manifests/30-customer.yaml

kind: Entityname: CustomerOfRecorddescription: "One customer across an ERP account and a CRM contact that share no identifier."scopes: [entity-resolution]sources:  - table: t_scenario_resolution_erp    map:      account_code: account_code      company_name: name      email: email      city: city  - table: t_scenario_resolution_crm    map:      contact_code: crm_id      display_name: name      email: email      phone: phoneresolution:  identity:    t_scenario_resolution_erp: account_code    t_scenario_resolution_crm: crm_id  cascade:    - id: email      exact:        fields: [email]        normalize: [lower]

$ lm entity apply -f manifests/30-customer.yaml
Applied entity "CustomerOfRecord" → connector."v_customerofrecord"

$ lm entity resolve CustomerOfRecord
rows seen: 10
resolved: 8
ambiguous: 1
unresolved (no rule): 1
unresolved (provider): 0
newly minted: 6
provider calls: 0
provider failures: 0
rule hits:
  customerofrecord:email: 9

$ lm entity ambiguous CustomerOfRecord -o json
[
  {
    "identityRef": "2f7844dc-3739-457a-968a-e4a7e8c4b5f0",
    "source": "t_scenario_resolution_crm",
    "sourceId": "CT-9003",
    "candidates": [
      "2f0553b1-1829-4350-98d5-e4234266135a",
      "59727d7b-3b4d-4b2b-8a45-5bb8fe51a910"
    ],
    "discardedCandidates": [],
    "setAside": [],
    "decidedBy": "automatic",
    "occurredAt": "2026-11-03T15:24:34.823862Z"
  }
]

$ lm entity get CustomerOfRecord --filter crm_id=CT-9004 -o json
[
  {
    "account_code": null,
    "name": "Transmission Nord",
    "email": null,
    "city": null,
    "crm_id": "CT-9004",
    "phone": "+33 3 28 00 00 04",
    "_conflicts": "{}"
  }
]

$ lm entity get CustomerOfRecord --filter crm_id=CT-9001 -o json
[
  {
    "account_code": "AC-100",
    "name": "Atelier Boréal",
    "email": "commandes@atelier-boreal.example",
    "city": "Lille",
    "crm_id": "CT-9001",
    "phone": "+33 3 20 00 00 01",
    "_conflicts": "{\"name\": [{\"src\": \"t_scenario_resolution_erp\", \"value\": \"Atelier Boréal\"}, {\"src\": \"t_scenario_resolution_crm\", \"value\": \"ATELIER BORÉAL\"}], \"email\": [{\"src\": \"t_scenario_resolution_erp\", \"value\": \"commandes@atelier-boreal.example\"}, {\"src\": \"t_scenario_resolution_crm\", \"value\": \"Commandes@Atelier-Boreal.example\"}]}"
  }
]

$ lm entity history CustomerOfRecord t_scenario_resolution_crm CT-9003 -o json
[
  {
    "ruleDeclaredId": "email",
    "ruleVersion": "ce968f5902d73417@engine-v1@hmac-k1@lower:1.0.0",
    "provider": "",
    "candidates": [
      "2f0553b1-1829-4350-98d5-e4234266135a",
      "59727d7b-3b4d-4b2b-8a45-5bb8fe51a910"
    ],
    "discardedCandidates": [],
    "setAside": [],
    "outcome": "AMBIGUOUS",
    "decidedBy": "automatic",
    "runId": "",
    "occurredAt": "2026-11-03T15:24:34.823862Z",
    "assertedBy": null,
    "assertedByLogin": null,
    "reason": null
  }
]

$ lm entity assert-same CustomerOfRecord --left t_scenario_resolution_crm:CT-9003 --right t_scenario_resolution_erp:AC-300 --reason "The CRM billing contact belongs to Rhône Filtration"
recorded: t_scenario_resolution_crm:CT-9003 and t_scenario_resolution_erp:AC-300 are same

$ lm entity resolve CustomerOfRecord
rows seen: 10
resolved: 9
ambiguous: 0
unresolved (no rule): 1
unresolved (provider): 0
newly minted: 0
provider calls: 0
provider failures: 0
rule hits:
  customerofrecord:email: 8

$ lm entity history CustomerOfRecord t_scenario_resolution_crm CT-9003 -o json
[
  {
    "ruleDeclaredId": "email",
    "ruleVersion": "ce968f5902d73417@engine-v1@hmac-k1@lower:1.0.0",
    "provider": "",
    "candidates": [
      "2f0553b1-1829-4350-98d5-e4234266135a",
      "59727d7b-3b4d-4b2b-8a45-5bb8fe51a910"
    ],
    "discardedCandidates": [],
    "setAside": [],
    "outcome": "AMBIGUOUS",
    "decidedBy": "automatic",
    "runId": "",
    "occurredAt": "2026-11-03T15:24:34.823862Z",
    "assertedBy": null,
    "assertedByLogin": null,
    "reason": null
  },
  {
    "ruleDeclaredId": "",
    "ruleVersion": "",
    "provider": "",
    "candidates": [
      "59727d7b-3b4d-4b2b-8a45-5bb8fe51a910"
    ],
    "discardedCandidates": [],
    "setAside": [],
    "outcome": "RESOLVED",
    "decidedBy": "asserted",
    "runId": "",
    "occurredAt": "2026-11-03T15:24:35.05687Z",
    "assertedBy": "1",
    "assertedByLogin": "integrator",
    "reason": "The CRM billing contact belongs to Rhône Filtration"
  }
]

$ lm entity ambiguous CustomerOfRecord -o json
[]

$ lm entity get CustomerOfRecord --filter crm_id=CT-9003 -o json
[
  {
    "account_code": "AC-300",
    "name": "Rhône Filtration",
    "email": "compta@groupe-rhone.example",
    "city": "Lyon",
    "crm_id": "CT-9003",
    "phone": "+33 4 72 00 00 03",
    "_conflicts": "{\"name\": [{\"src\": \"t_scenario_resolution_erp\", \"value\": \"Rhône Filtration\"}, {\"src\": \"t_scenario_resolution_crm\", \"value\": \"Groupe Rhône\"}]}"
  }
]

$ lm entity resolve CustomerOfRecord
rows seen: 10
resolved: 9
ambiguous: 0
unresolved (no rule): 1
unresolved (provider): 0
newly minted: 0
provider calls: 0
provider failures: 0
rule hits:
  customerofrecord:email: 8

$ lm entity get CustomerOfRecord --filter crm_id=CT-9003 -o json
[
  {
    "account_code": "AC-300",
    "name": "Rhône Filtration",
    "email": "compta@groupe-rhone.example",
    "city": "Lyon",
    "crm_id": "CT-9003",
    "phone": "+33 4 72 00 00 03",
    "_conflicts": "{\"name\": [{\"src\": \"t_scenario_resolution_erp\", \"value\": \"Rhône Filtration\"}, {\"src\": \"t_scenario_resolution_crm\", \"value\": \"Groupe Rhône\"}]}"
  }
]

$ lm entity assert-same CustomerOfRecord --left t_scenario_resolution_crm:CT-9999 --right t_scenario_resolution_erp:AC-300
Error: API error 404: entity 'CustomerOfRecord' has no row t_scenario_resolution_crm:CT-9999 — check it with lm entity history CustomerOfRecord t_scenario_resolution_crm CT-9999, or run lm entity resolve CustomerOfRecord if the row is new

Limits

Email is a declared signal, not proof of identity.
Exact rules only; no candidate provider.
Synthetic data; the company names are French.

Customers · ERP and CRM

The same customer lives in two systems.

With no shared id.

Five accounts, five contacts

The problem

Who is who?

AC-300 in the ERP, CT-9003 in the CRM.
Neither knows the other.

What is not changed

The ERP and the CRM stay as they are.

Two exports, read as they are.
No key to add at the source.

The trap

A shared mailbox.

compta@groupe-rhone points at two accounts.
Merging by chance is getting it wrong.

What had to be written

normalize: [lower]

One line.

Email matches, in lower case.
Two candidates: an ambiguity.
A 52-line scenario, three manifests.

The decision

A human decides.

CT-9003 joins Rhône Filtration.
The reason is written, it holds on replay.

The limit

Email is a signal, not a proof.

Exact rules only.

Try it

The scenario replays locally.

Every step is published with its output.

This scenario runs on lumnik Open, free and open source.

Try lumnik Open →