1. What CRM data cleansing is, and why CRMs decay
CRM data cleansing, also called CRM data cleaning or CRM data hygiene, is the work of finding and fixing bad records in your CRM. It is different from CRM enrichment, which adds what is missing: cleansing fixes what is there. In practice you do both, cleansing first.
CRMs decay for two reasons. The first is internal: people. Every form, import and sales rep adds records in its own way, and nobody checks whether the company already exists. The second is external: the world moves on. Companies relocate, change their name, merge into a group or stop trading, and nothing in your CRM tells you.
| Problem | Typical cause | Visible from inside the CRM? |
|---|---|---|
| Duplicate companies | Forms, imports and reps creating the same account | Partly, with duplicate tools |
| Inconsistent formats | Free-text fields, no picklists | Yes |
| Empty fields | Short forms, quick manual entries | Yes |
| Closed or merged companies | Bankruptcy, dissolution, acquisition | No |
| Moved companies | New office, new registered address | No |
The last two rows are why cleaning only with the tools inside your CRM always leaves something behind.
2. Measure before you clean
Pull five numbers before you touch anything. They tell you where to start and, later, whether it worked.
- Duplicate rate: the share of companies that exist more than once. Your CRM's duplicate tool gives a first estimate.
- Complete records: the share of companies with every required field filled (country, industry, size, owner).
- Stale records: companies with no activity, deal or update in two years.
- Bounce rate: from your email tool, as a proxy for contact decay.
- Unmatched companies: the share you cannot link to a registered company. You only know this after step 5, so note it then.
3. Back up and agree the rules
Export a full backup of companies, contacts and deals with their IDs. Then write down, with sales and marketing, three rules that every cleaning decision follows:
- Which record survives a merge: usually the one with the most activity or the oldest deal, so reporting does not break.
- What counts as the same company: the same registration number. Two subsidiaries of one group are two companies.
- What happens to dead records: archive rather than delete, so history and attribution survive.
4. Standardise formats
Formats first, because every later step matches on them. Fix these once and lock them with picklists and validation rules:
| Field | Standard |
|---|---|
| Country | ISO 3166 code or a fixed picklist, never free text |
| Company name | Legal name in one field, trading name in another |
| Legal form | One spelling per form (B.V., GmbH, Ltd), or a separate field |
| Phone | International format with country code (+31 20 ...) |
| Website | Domain only, lowercase, no https:// or www. |
| Industry | One code system (SIC or NACE) from a picklist |
5. Find and merge duplicates on the registration number
Duplicate tools compare names and domains. That catches "Heineken" and "HEINEKEN N.V.", but it also merges two different companies with similar names, and it misses the duplicate that was entered under a trading name. The reliable key is the number the register gave the company.
Match every company to the register
Send name plus country, or the website domain, and store the registration number that comes back. Our docs on matching a list of companies show what to send per country.Group records with the same number
Every group of two or more is a set of duplicates. No fuzzy threshold to tune, no false merges between namesakes.Merge by your survivor rule
Keep the surviving record from step 3 and merge the rest into it, with contacts, deals and activities.Link groups, do not merge them
Subsidiaries keep their own record and get a parent association. Merging a group into one account breaks territories and invoicing.
6. Remove closed, merged and moved companies
With every company matched, the register tells you what your CRM cannot: whether the company still exists, and where.
- Dissolved or bankrupt: archive the company and its open deals, and stop sequences to its contacts.
- Merged into another company: move the history to the surviving entity, often already in your CRM as a separate account.
- Moved: update the registered address and check whether the territory owner changes.
- Renamed: update the legal name, keep the old name as an alias so search still finds it.
Our register data covers 377M+ companies in 190+ countries, with the status of each, so one check covers a CRM spread across many markets.
7. Cleaning in HubSpot, Salesforce, Pipedrive and Zoho
Every CRM has built-in help for formats and obvious duplicates. Use it for steps 4 and 5, and add the register check for step 6.
| CRM | Built in | Add for the rest |
|---|---|---|
| HubSpot | Duplicate management for companies and contacts; property validation; data quality tools in Operations Hub | The CompanyData Enrichment app to match companies to the register |
| Salesforce | Duplicate and matching rules; validation rules; Data Loader for bulk updates | An export cleaned against the register, loaded back on the Salesforce ID |
| Pipedrive | Merge duplicates for organisations and people; required fields | An export cleaned against the register, or the API through Zapier or Make |
| Zoho CRM | Deduplication on chosen fields; validation rules | An export cleaned against the register, or the API through Zoho Flow |
Salesforce data hygiene deserves one extra note: matching rules run on records as they are saved, so a fuzzy rule set too loose blocks reps from creating real new accounts. Matching on a registration number field avoids that.
8. Keep it clean: CRM data hygiene rules
A cleaned CRM stays clean for months, not years, unless you change how records get in. Five rules do most of the work:
- A matching key on every company: the registration number, so every import and integration can check for an existing record.
- Enrich on creation: new companies are matched and filled the moment they are created. See CRM enrichment.
- Required fields with picklists: country, industry and owner cannot be skipped or typed freely.
- A re-check two to four times a year: the whole base against the register, for status and address changes.
- One owner: someone who decides on fields and rules, and reads the five numbers from step 2 every month.
Data cleansing services
We run all eight steps on your export and send back the cleaned records with a change report.
Get a free cleansing auditHubSpot data enrichment
Match HubSpot companies to the register from the record itself.
See the HubSpot app