Clean customer data without merging the wrong people
Two customer records can look like the same person while belonging to two different buyers at the same company. A matching rule that is too cautious leaves your CRM full of duplicates; a rule that is too aggressive merges separate trade buyers, scrambles order history and exposes one customer's quotes to another. - Sales reps see multiple conflicting records for the same contact with different phone numbers and notes. - Off-the-shelf deduplication tools threaten to merge separate buyers who share a company name or business park postcode. - Invoices, credit limits and order history are attached to individual people rather than the parent trade account. - Form imports and manual entries keep recreating the same duplicates weeks after a clean-up finishes. Agentix helps UK eCommerce, B2B and trade businesses audit, cleanse and match customer records safely. Part of our wider [customer data and systems work](/services/customer-data-systems), we test matching rules on a snapshot of your data, separate individual contacts from company accounts, and route uncertain matches for human review before any live record is changed.
Why "remove duplicates" goes wrong on B2B and trade databases
In our experience, automated deduplication tools cause damage when run blindly against B2B and trade databases because they confuse a **natural person (Contact)** with a **business entity (Account)** or a **delivery site (Location)**. On a trade account, a procurement director, a site manager and an accounts clerk can share the same company name, corporate email domain and delivery postcode. If an automated script merges those records on company name or postcode alone, individual login credentials break and private pricing quotes go to the wrong inbox. Here is an illustrative breakdown with invented round numbers (not a client result) showing how **10,000 unmaintained CRM records** separate once proper identity rules are applied: | Record category in a 10,000-record database | Records | Share | What is happening in the data | What happens if you run a naive auto-merge | | -- | -- | -- | -- | -- | | 1. Clean, unique customer records | 6,200 | 62% | One verified contact linked cleanly to the right account | Left untouched. | | 2. Exact deterministic duplicates | 1,400 | 14% | Identical email address, mobile number or customer ID entered twice | Safely consolidated into 700 clean r
Where customer data quality breaks down in practice
When records duplicate or conflict across your systems, the cost shows up quickly in your sales team, finance office and marketing campaigns: | What you see in your CRM or reports | What is happening underneath | What it costs your business and team | | -- | -- | -- | | Sales reps see three records for the same buyer with different notes | Website forms, eCommerce checkouts and manual entries create new contacts instead of checking existing primary keys | Reps duplicate outreach, ask questions the customer already answered yesterday, and waste hours cross-checking accounts | | Separate trade buyers were merged into a single profile | A bulk cleaning tool matched records on company name or postcode without checking individual identity | Individual order histories and portal logins are lost, and private quotes land in the wrong person's inbox | | Invoices and credit limits sit against individuals rather than the business | Your database treats Contacts and Accounts as the same thing, logging orders against whichever person placed them | Finance cannot see true company-wide credit exposure, and account spend is fractured across multiple profiles | | Marketing emails bounce while a ne
How we cleanse and match your records safely
In enterprise CRMs such as HubSpot, Salesforce and Microsoft Dynamics, merging two contact records re-parents historical activities, deals and orders permanently. Once background triggers have run, there is no simple one-click undo. That is why we never run bulk matching scripts straight against your live production database: 1. **Snapshot audit and entity rules first.** We take a secure snapshot of your records, agree what defines a Contact versus an Account or Location in your business, and decide which system owns each field. 2. **Dry-run testing with human exception queues.** We run proposed deterministic rules (using verified email addresses, mobile numbers and UK Companies House registration numbers where applicable) on the snapshot first, and show you the exact before-and-after counts and sample matches before touching live data. 3. **Documented execution and root-cause prevention.** Once you approve the rules and rollback plan, we execute or brief the cleanse with a full audit log of source IDs, and point out the form or import step that was creating the duplicates so they don't pile right back up.
Frequently asked questions
### Why shouldn't we just turn on our CRM's built-in auto-deduplication tool? In our experience, off-the-shelf deduplication tools rely on loose text similarity that works poorly for B2B and trade data. When multiple buyers share a company switchboard, `accounts@` email address or industrial estate postcode, automated tools can merge distinct people or separate companies into one record. We use strict deterministic rules and keep ambiguous records in a human review queue so you never lose a real buyer relationship to a false merge. ### How do you handle trade accounts where several people buy for the same company? We separate the individual buyer (Contact) from the contracting company (Account) and delivery site (Location). Where several people share a company domain or address, we link them as distinct contacts under one parent account rather than merging them into a single person. We only merge contact records when verified primary identifiers prove they are the same individual. ### Can a customer data merge be undone if something goes wrong? In many live CRMs, merging contacts overwrites fields and re-parents history in a way that cannot be reversed with a native undo button.
Speak to us about customer data cleansing
Whether you need to clean up years of duplicate CRM contacts, structure B2B trade accounts properly, or prepare a database before a system move, we can help you do it safely.