CRM data cleanup and hygiene: a checklist and best practices

In a CRM, data hygiene covers the records themselves: one record per person and company, email addresses that still work, the fields your process depends on filled in, values written the same way down each column, an owner who still works at the company, and records that still describe someone you can reach.

It is not data security. Who can see records, permissions and backups matter, but they are a different job with a different owner.

Data quality describes the records: how accurate, complete, consistent and current they are when you look. Data hygiene is the work that raises those numbers and stops them slipping back, which is what the rest of this page covers.

To put numbers on it first, see how to measure CRM data quality.

Data decay is the share of records that go out of date without anyone touching them. People change jobs, so their email stops working and their title and company are wrong. Companies rename, merge and close. Phone numbers change with the person.

People move fastest. 20.6% of US wage and salary workers had been with their employer a year or less in January 2026 (US Bureau of Labor Statistics). Leaving out people new to work, our research estimates that job changes alone put just under a fifth of B2B contacts out of date each year.

Decay adds to what new data brings in. Imports, web forms and manual entry keep adding duplicates and inconsistent values, so a CRM cleaned once drifts back within months.

A full cleanup is worth the effort when one of these is true. If none of them are, a smaller fix is a better use of the afternoon.

Run these checks on an export of your CRM. The steps below take them in order, starting with which records to keep.

These decide how the records go back, so they are worth knowing before you export.

A CRM is built to work one record at a time, which is the wrong shape for fixing forty thousand records at once. A spreadsheet is good at the opposite: doing the same thing to every row. So export, clean the file, and push it back over the originals. The steps are the same on HubSpot, Salesforce and Pipedrive, and the order matters, because each one makes the next cheaper.

The pictures are real runs in Operelio on one sample export of 5,875 contacts; the names and companies are made up.

Give it one owner. That person doesn't fix every record, but they set the rules (which fields are required, how values are written, what counts as a duplicate), run the reviews, and turn back imports that break them. When everyone owns the data, the duplicates come back with the next list.

Clean every import before it lands, since much of the damage arrives with lists: trade show scans, prospecting exports, a spreadsheet from a rep. When to clean in the file and when in the CRM is in cleaning data before vs after a CRM import.

Duplicates come in with imports, forms and integrations, from a list with someone under a second address or an import that isn't matched on email. If the pairs from step 3 share a source, fix that source too, or they come back with the next file.

Save the mapping and the workflow from this cleanup, so the next file is one run, and screen every file against your Do Not Contact list, not only the first. For files that arrive every week, a trail with the Remove suppressed contacts step does this for you: it watches a folder, cleans each file that lands there and pushes it into HubSpot, Salesforce or Pipedrive.

Review what is already in the CRM once a quarter: the checklist above, on a fresh export. Between reviews, keep a saved view or report in your CRM for the checklist's counts, such as contacts with no company or no owner, so a jump shows the week it happens.

Let your own numbers say when quarterly isn't enough. Run Data Health Check on each review's export and note the score. If it has fallen since the last review, records are going bad faster than you fix them: check more often, and look at what came in since, because the cause is usually one source.

If nobody in-house can own it, a cleanup service or a cleaning tool can take the work; how they compare is in best CRM data cleaning tools.

HubSpot deduplicates as records come in: contacts by email address and companies by domain, on imports and form submissions. An import row whose email matches an existing contact updates that contact instead of creating a new one. An import with no email or Record ID column creates a new contact for every row, as HubSpot's Deduplicate records in HubSpot article says. Two things slip through: the same person under two email addresses, and companies created through the API, which are not deduplicated by domain.

For records already there, HubSpot's Data Quality page (Data Management, Data Quality) shows duplicates and formatting issues, on Starter and above of most hubs. Reviewing and merging duplicate pairs one at a time needs a Professional or Enterprise subscription, and merging them in bulk needs Data Hub Professional or Enterprise. A merge cannot be undone, so decide which record is primary before you click. How to find, merge and prevent duplicate contacts in HubSpot covers it step by step.

Some contacts cost money as well as accuracy. HubSpot's Understand marketing contacts article says only marketing contacts count toward the contact tier you pay for, on Marketing Hub accounts that have marketing contacts (every account created after September 2022); contacts set as non-marketing don't count, and you aren't billed for them. Contacts who hard bounced, unsubscribed or marked an email as spam are already dropped from future marketing emails, so keeping them as marketing contacts pays for people you can't email.

The quickest route is Account & Billing, Marketing Contacts, Clean up contacts, which lists unsubscribed and bounced contacts by default and sets the ones you pick as non-marketing. To delete them instead, or to find them yourself, use active segments. For bounces, the Email tools menu in the marketing email tool creates an active segment of hard-bounced contacts, and HubSpot's Delete contacts who no longer receive your marketing emails article walks through deleting them from it. For unengaged contacts, filter on engagement dates. HubSpot's own example is Last marketing email open date is unknown and Marketing emails delivered is greater than a number you pick; Last email clicked, Last reply and Last activity date work the same way.

Give the unengaged one re-engagement email, then set the rest as non-marketing, from the Usage & Limits page or the contacts list. The change takes effect on the next update date, and fewer marketing contacts don't move you to a lower tier until your subscription renews.

Then stop the next mess at entry. On Starter and above you can make properties, and the link to a company, required on the form your team uses to create records by hand, as Customize the create form for each object explains. Forms can make a field required too, and a dropdown select property makes people pick a value rather than type one, so a country is written one way.

Salesforce's Duplicate Management, included in Essentials, Professional, Enterprise, Performance, Unlimited and Developer editions, works in two parts. Matching rules say what counts as a duplicate for accounts, contacts and leads. Duplicate rules say what happens when someone creates one: show an alert, or block the save. What they do to imports and pushes from other tools is in Salesforce duplicate rules.

Validation rules do the same job for formats and required fields: a record that breaks the rule cannot be saved. Both act when a record is created or edited, so they keep new mess out but leave the records already in your org as they are. Those are what the steps above are for.