CRM Data Cleaning First: The Checks to Run Before Enrichment
A routine for RevOps and Sales Ops. What to check on a HubSpot export, a Salesforce report or a CSV before you enrich it, which record survives a merge, and what goes back to the CRM.
CRM data cleaning is the pass you make on a list before you pay to fill it in: duplicates, outdated employers, invalid and role emails, people you must not contact, formats and gaps. The order decides the bill. Work-email and mobile lookups are charged when a result is found, so a hit for a duplicate, a shared inbox or a person who has left is a bad row you paid for.
This guide is about the CRM side: a HubSpot or Salesforce export, the routine you run on it, and what goes back into the portal. If you enrich in a Clay table, the companion piece shows where those credits go.
Start with the export: keep the record ID
Export the record ID with every other field. Emails change and names have typos, but the ID names one CRM record without doubt. That matters when the cleaned file goes back by import. HubSpot matches an imported contact by its email and a company by its domain. A row without an email, a domain, a record ID or a custom unique property creates a new record, and a row whose email matches several contacts fails. A cleaned CSV without IDs can bring back the duplicates you just removed. When you connect HubSpot or Salesforce instead, ListPlus reads the records with their source IDs.
The checks that change who you enrich
Standard checks use 0 credits, optional AI checks show their usage first, and nothing changes until you apply a fix.
Duplicates, including near-duplicates
Two rows for one person become two lookups and, later, two sequences. Exact matches are the easy half: the same email, phone or LinkedIn URL. The expensive half is the near-duplicate: “Acme Inc” beside “Acme”, a typo in the last name, an accent. In ListPlus, Duplicate Email, Duplicate Name + Company and Duplicate LinkedIn flag exact pairs, and Duplicate Name + Company ignores the legal form. Similar Name + Company flags the near misses for you to review rather than merging them on its own.
Which record survives a merge
Write one rule down before anyone merges: which record stays (the most active, the oldest or the most complete, the same choice every time), and which value wins per field (recently checked over old, filled over empty, typed in by hand over a guess). HubSpot says plainly that merged records cannot be reverted. In Salesforce, matching rules find likely duplicates, and duplicate rules decide whether saving one brings a warning or a block.
In ListPlus, the fix for a duplicate group keeps the most complete row, copies over the values only the other row had, and marks the other row ignored. Ignored rows leave the working view but are not deleted. An address or phone shared by two different names is never merged: that is info@ or a switchboard. Emails, deals and notes hang on the CRM record, so the record itself is merged in the CRM.
Outdated employers and job changes
A work email for a company the person has left is the wrong email. Former/Ex Role flags a title that already says “Former” or “Ex-”. Left the Company flags a row when a stored LinkedIn profile marks the position at the company in your list as ended, while the company cell still names it. Possible Job Change flags weaker signs, such as a company that no longer shows as current. The Job change column uses 0 credits and shows Yes only in the first case. A profile URL alone has no positions to compare.
Detect Company Changes judges profiles the list already holds, at 0 credits. Detect Job Changes loads missing profiles and shows its steps and estimated price first: the email check costs 0.3 credits for a clear result, loading a LinkedIn profile 0.5 credits when one is found, the other steps 0 credits. A profile loaded in the last 30 days is reused at 0 credits, so a very recent change can be missing. ListPlus does not repeat the check on a schedule or send job-change alerts. Neither check fixes the row. Decide first whether it follows the person or stays with the account.
Invalid emails and role inboxes
Invalid Email Format and Email Domain Typo catch values that are not addresses. Disposable Email Domain and Test Domain Email catch mailboxes that should stay out of a sequence. Role-Based Email marks shared mailboxes such as info@ and sales@, where a reverse lookup has nobody to find, so ListPlus does not run one. Do-Not-Contact Email marks addresses such as noreply@. Personal Email Domain marks a personal provider: keep those rows for a campaign aimed at personal addresses, hold them when you are buying work emails.
A found address is not yet a deliverable one. Verify Email Deliverability costs 0.3 credits for a clear result (valid, invalid or disposable). Catch-all and unknown results are not charged, and an obviously invalid, disposable, role or test address is marked at 0 credits before the verification service is called.
Block lists and existing customers
Some rows are correct and still must not be enriched or sequenced: customers, competitors, partners, unsubscribes. Match them before the lookup. On the block list flags an email or a company domain you have blocked. Reference lists, part of Pro, flag Existing customer, Competitor, Partner and In a list with a role. The check flags the row. You decide whether it stays out of the hand-off.
Formats, gaps and company rows
The rest is mechanical, and it comes before the duplicate match: two spellings of one company are how near-duplicates survive. Name Capitalization, Legal Suffix Normalization, Missing Country Code and Country Format Inconsistency give the next match one spelling. Empty Company, Empty Job Title and No Email for Campaign show the gaps worth filling. LinkedIn Company Page and Email-Company Mismatch catch a company where a person belongs. On a company list, person columns, checks and enrichment stay hidden until you switch it to contacts, and nothing is deleted.
How to clean CRM data before you enrich it
Run these steps in order on every export. Each one removes rows the next would pay for. One person, usually in RevOps, should own the field rules and the merge rule.
- Export with the record ID, then bring the file into a list or connect the CRM. A CSV or Excel file works on every plan.
- Normalize names, companies, countries and phones.
- Ignore exact duplicates, review near-duplicates, keep the more complete row.
- Flag block list, customer, competitor and partner matches.
- Hold invalid emails, disposable domains and role inboxes.
- Flag former employers and decide the company before an email lookup there.
- Read the price, then enrich only the missing fields on the rows you keep.
- Send the changes back, and merge the remaining duplicates in the CRM.
What goes back to HubSpot or Salesforce
A list imported from HubSpot, Salesforce, Google Sheets, Notion or Attio can write its changes back to the source. Only fields that came from the source go back: a name you corrected, or an empty cell you filled in an imported column. Columns you added in ListPlus and the enrichment data behind the cells stay in ListPlus.
The sync sends the list’s open changes in one run. You do not pick rows for it. If someone changed the same field in the CRM in the meantime, the CRM’s value wins and the cell takes it. Ignoring a row changes nothing in the CRM, so a duplicate you ignored in the list is still two records in the portal until you merge it there. Records that should reach another place, such as a sequencer, leave by export or by a pipeline.
CRM data enrichment, after the list is clean
Before a paid run, ListPlus shows the steps, the expected credits and a credit ceiling you accept. A step that would pass it is not started. Work-email and mobile lookups are charged when a result is found. Without a match the field stays empty. A run on a selection fills empty cells only. A rule that enriches new records can also replace invalid values or overwrite everything, including values typed by hand.
Connecting the CRM: on Basic, one connected tool costs 50 credits per 30 days, renewed while the balance covers it. A second tool needs Pro, which includes every connected tool. Each request ListPlus makes to the tool for import, reload, sync back or export costs 1 credit. Looking at a source in the import wizard uses 0 credits. Your own agent, such as Claude or ChatGPT, can run the same checks and enrichment through the Agent API, within the day budget you set on its connection.
Then hand the rows on
The cleanup is done when every row you hold is one person, at a company they still work for, with fields a matcher can read and no reason to suppress them. Send the corrections back to the CRM and the kept rows on to the sequencer.
Questions before you clean a CRM export
What is CRM data cleaning?
It is the work of making a CRM export or a contact CSV match reality before you use it. Cleaning decides which rows are worth a lookup. Enrichment then fills gaps on the rows you keep.
How often should I clean CRM data?
On every export, right before a paid run, because that is when a bad row costs credits. For the CRM itself, keep a schedule you will hold, and add a pass after a large import, a purchased list or a new form.
Does HubSpot remove duplicates automatically?
Partly. HubSpot matches a new contact to an existing one by email and a new company by its domain, but companies created through the API are not matched by domain, and look-alike records with different emails stay two records. Its duplicate management tool, on Professional and Enterprise plans, suggests those pairs for review. Bulk merges and custom duplicate rules need Data Hub Professional or Enterprise.
Does cleaning in ListPlus change my CRM?
Only through the sync, and only for fields that came from the CRM. A list imported from HubSpot or Salesforce sends its open changes back in one run: a corrected name or a filled cell in an imported column goes back, columns you added in ListPlus and enrichment data stay in ListPlus. Ignoring a row changes nothing in the CRM, so a duplicate you ignore in the list is still two records there until you merge it in the CRM.
How do I keep enrichment cost under control?
Read the expected credits and the ceiling before the run, and enrich only rows that passed the checks. Work emails and mobile numbers are charged when a result is found, a miss is not, and standard checks use 0 credits.