CRM and contacts

Keeping CRM Data Clean: Duplicates, Numbers and Stale Records

A practical routine for accurate CRM contact data: prevent bad entries, find and merge duplicates, standardise phone numbers and retire stale records safely.

Call editorial team · Published · 5 min read

Dirty data costs a phone team in small, constant ways: the same person called twice in a day, a callback sent to the wrong number, a report that counts one lead as three. Nobody notices until trust in the CRM has already gone. The good news is that most data problems follow a few patterns, and a short routine keeps them under control.

An analyst checking a list of contact records on a monitor
Illustrative image

This guide covers how to prevent, find and fix the common problems. It follows from Logging Calls in a CRM Without Slowing Agents Down, because clean data begins with clean entry. For the lists you build on top, see Tags and Segments for Contact Lists.

The usual suspects

Where CRM data goes wrong

  • DuplicatesThe same person entered twice, or after an import, or by two agents.
  • Number formatsNumbers with and without country codes, spaces or leading zeros.
  • Missing basicsNo company, no owner, no stage.
  • Stale recordsPeople who have moved on, closed accounts or invalid numbers.
  • Free-text chaosIndustries or sources typed ten different ways.
  • Consent gapsNo note of whether a person may be contacted.

Do a quick audit first. Pull a sample of fifty records and count the problems by type. You will usually find that two or three issues account for most of the mess, and those are where to start.

Prevent problems at entry

It is far cheaper to stop a bad record than to fix it. Set a standard and make it easy to follow.

Decide who may create and edit records. Fewer people with the right to create contacts means fewer accidental duplicates. Agents can still add notes and tasks to anything.

Finding and merging duplicates

Duplicates usually come from imports, from different agents adding the same person, and from the same person calling from two numbers. Match on phone number and email first, since they are the strongest signals, and review name-and-company matches with care because people share names.

When you merge, keep the older record as the main one and move everything onto it: calls, notes, tasks and consent. Keep the second number as a secondary one so you can still recognise calls. Before you merge, check which record will be kept, because most systems cannot undo a merge.

A safe merge routine

  1. FindList records that share a number or email.
  2. CompareCheck that they really are the same person.
  3. Choose the keeperUsually the older, more complete record.
  4. MergeMove calls, notes, tasks and consent onto it.
  5. CheckOpen the merged record and read the history.

Phone numbers and do-not-call status

For a phone team the number is the key. Bad numbers waste calls and good numbers on the wrong list create risk. Mark invalid numbers and wrong numbers so they are not called again. Make do-not-call status visible on the record and keep it when merging. If someone asks not to be contacted, the record must show that immediately and the number should stay blocked even if the contact is later re-imported.

After an import or a system change

Imports and migrations are the biggest single source of dirty data. Before a large import, clean the source file: remove blank lines, fix obvious number errors and decide how to treat rows that match existing contacts. Import a small test batch first and inspect it. Afterwards, check how many rows were imported and how many were skipped and why. If the numbers differ from what you expected, find out before the next import, not after.

When you change anything that touches many records, such as renaming a stage or merging two accounts, note what changed and when. A short change log makes it far easier to explain a strange report later.

Dealing with stale records

Contacts go out of date: people change jobs, companies close and numbers are reassigned. Review records that have had no activity for a long time, and decide what to do with them. Archive rather than delete when unsure, so history is preserved. Where you hold personal data, you also have an obligation not to keep it longer than needed; your retention policy decides how long.

Archive, update or remove

When it fitsWhat to do
Old but still relevantCustomer who has not called for a whileKeep; add a reminder to check in
Out of date detailsNumber or company changedUpdate; note the source of the change
No longer usefulClosed company, invalid number, wrong personArchive or remove under your retention policy
Asked to be removedFormal request to erase dataFollow your data request process

Do not forget the people side. Tell agents why clean data matters, using a real example such as a customer who was called three times in one morning. Praise accurate records, and make it easy to report a bad one. A team that sees data quality as shared pride, rather than a chore handed down, keeps it up without being chased.

Make it a routine

Data quality is not a project; it is a habit. Choose a small set of checks and run them on a schedule. A weekly duplicate review, a monthly look at missing owners or stages and a quarterly review of stale records is more than enough for most teams. Assign each check to a named person and report the results in the same meeting as the call metrics.

Monthly data quality checks

  • Duplicates by number and email reviewed and merged
  • Records without an owner or stage listed
  • Invalid and wrong numbers marked
  • Do-not-call records confirmed visible
  • Latest import checked for problems
  • Stale records reviewed for archive
  • Any free-text fields that need standardising

Cleaning data in Call

In Call, the CRM setup page under Settings lists possible duplicates, showing why two records look like the same contact, for example a shared number. For a pair, you can press Merge; the older record is kept and the calls, notes and appointments move onto it. Merge is offered for pairs only, and cannot be undone from that page. On the Revenue page, the Data hygiene tab finds groups of records that share a phone number or email, up to fifty groups of each, and lets you choose which to keep. Merging there moves call history, notes, tasks and consent, keeps the other number as a secondary, and fills blank fields on the kept record. The same tab lists accounts at risk of leaving, with days since last contact.

Extra fields on a contact can be marked required, and accounts can be added so contacts from the same company link together by company name. When you import a CSV, numbers already on your do-not-call list come in at the stage Do not call instead of being dropped, and the page tells you how many contacts were skipped. See Importing Contacts Without Making a Mess.

Frequently asked questions

How often should we check for duplicates?

Weekly for active teams, and always after a large import. A quick check on new records is cheaper than a big clean-up later.

Which record should I keep when merging?

Usually the older and more complete one. Check that call history, notes and consent end up on it before you finish.

Should we delete old contacts?

Not automatically. Archive when unsure, and follow your retention policy and data request process for personal data.

Which phone number format should we use?

Choose one and use it everywhere, including the country code. The exact format matters less than consistency.

Who is responsible for data quality?

Everyone enters it, but one named person should own the routine checks and report on them.

How do I stop duplicates at the source?

Warn on matching numbers and emails, limit who can create contacts, and use one standard way to add people.

Find and merge duplicate contacts in Call

In the demo console you can try incoming calls in a simulation.

Related guides