CRM data cleanup before migration: what to fix, what to keep, and what to assess first

CRM data cleanup before migration should merge duplicates and delete junk but keep closed-lost deals, old activities and notes until someone assesses their value. Export everything with record IDs first, then run a short screen with the owner, because long deal histories are what AI data buyers look for.

What should you clean up in a CRM before migration, and what should you keep?

Clean up duplicates, broken fields and obviously dead test data. Do not purge closed-lost deals, old activities, notes or inactive accounts until someone has assessed whether that history has value. The rule is simple: assess before you purge.

Pre-migration cleanups usually aim to cut record counts to reduce migration cost and speed up the cutover. That instinct is sound for junk, but the same pass often deletes the longest, most connected history in the company: the trail of how deals were won and lost. For a RevOps consultant, a short screen with the owner before deletion costs an hour and keeps an option open.

What is safe to remove, and what is not?

Separate hygiene from history. Hygiene fixes the data; history is the data.

Record typeDefault actionWhy
Exact duplicate contacts and accountsMerge, keep the merge logMerges are reversible only if you record the surviving ID
Test records, sandbox imports, spam leadsDeleteNo operational value
Closed-lost opportunities with loss reasonsKeep, archive if neededOutcomes are what make deal history informative
Old activities, emails and call logsKeep or archive to a flat exportThe sequence of touches shows how work really happened
Notes and attachments on accountsKeep or export with record IDsContext lives here and nowhere else
Inactive accounts and former customersKeepChurn history is part of the story
Fields nobody usesHide first, delete after exportField metadata explains old data

The assess-before-purge screen in five questions

Run this with the owner or revenue leader before any bulk delete. If any answer is yes to the first two, pause the purge.

  • History: does the CRM hold several years of deals, activities and cases, going back before the current sales team?
  • Outcomes: are won, lost and escalated outcomes recorded, with reasons?
  • Breadth: does the CRM connect to other systems such as finance, support or marketing automation?
  • Rights: did the company create these records, and do customer contracts and privacy notices allow them to be licensed?
  • Sponsor: can you reach an owner, CEO, CFO or authorized representative who could decide?

The company must also meet the baseline: 50+ full-time employees at peak (contractors excluded), several years of documented operations and the right to license. The who qualifies page has the details.

How do you run the cleanup without destroying value?

Follow this order. The point is that archival copies exist before anything is deleted.

  1. Take a full export of every object, including attachments, with record IDs and a field dictionary, before the cleanup begins.
  2. Agree the delete list in writing with the owner, naming each object and the filter used.
  3. Do duplicate merges first, and store the mapping of merged IDs to surviving IDs.
  4. Delete only the categories marked safe in the table above.
  5. Migrate what the new CRM needs for daily work, and archive the rest to storage the company controls.
  6. Keep the export until the owner has decided whether the history is worth screening, even if the new CRM never imports it.

For a specific move off Salesforce, what history gets left behind in a Salesforce to HubSpot migration shows where the gaps usually appear. If the same project retires a second system, the Jira and Confluence migration page covers the engineering side.

What mistakes do cleanups make most often?

MistakeWhy it hurtsFix
Deleting closed-lost deals older than a cutoffRemoves the outcomes that give history meaningArchive them in the export instead
Importing only the last 12 months to cut costLeaves years of history in a system being cancelledExport the rest before cancelling
Merging duplicates without logging IDsBreaks links to finance and support recordsKeep a merge map
Stripping notes to save storageLoses context found nowhere elseExport notes with their parent IDs
Cleaning without telling the ownerDecision about the archive is made by defaultPut the screen on the project plan

Illustrative: a mid-sized services firm

Illustrative and fictional. A 180-person IT services firm is moving from one CRM to another. The migration plan proposes dropping opportunities older than three years and all activities older than 18 months. The consultant proposes one extra step: export first, then ask the owner whether the history is worth a screen. The firm keeps the export, imports the recent year, and the owner later chooses to check eligibility. Nothing is promised; the company decides.

What do you say to the owner?

Use the introduction email builder to draft it with the owner's approval. You never handle or describe the records.

Which questions should the owner answer before you start?

Ask these in the kickoff call, so the archive decision is made on purpose and not by whoever writes the filter.

  • Who owns the old CRM contract, and when does access end?
  • Is any business unit, brand or acquired company history stored in the same instance?
  • Do customer contracts restrict how account and contact data can be used or shared?
  • Which integrations write into the CRM, such as billing, support or marketing tools, and will they be retired too?
  • Who inside the company can run exports and confirm they are complete?

Answers to the rights questions matter most. If customers or an outsourcer own the data, or the instance is mostly consumer personal data, the screen ends there and the cleanup proceeds as normal.

How do partner rewards work for a RevOps consultant?

Partners earn 25% of the eligible platform fees SourceX actually collects from the referred company's licensing deals, capped at $100,000 per referred company. Payment happens only after the buyer pays and SourceX receives its fee; a screen, meeting or signed agreement alone does not trigger it, and no reward is guaranteed. The reward is never deducted from what the company receives. Consultants should also disclose the arrangement to clients where their own engagement terms require it. The program terms govern the details.

When is it not worth raising?

Skip it for a CRM under a few years old, one that is mostly consumer contacts, or a company well below 50 full-time employees at peak. Also skip it when a court, trustee or assignee controls the assets and has not been involved. The day 1 readiness guide covers when to hold legacy systems during acquisitions, and cutting cloud storage costs covers the same assess-first rule for storage.

Next step

Add the five-question screen to your next migration plan. If a client passes, register as a partner and make the introduction, or point the owner to sourcex.si/apply. The RevOps consultant page explains the role in more depth, and data licensing during a CRM migration covers timing.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Buyer selects and paysThe buyer selects and pays for the data and SourceX receives its fee.
  4. Step 4You get your rewardYour share of SourceX fees becomes payable.

Common questions

Should I delete old leads before a CRM migration?

Delete spam, test records and exact duplicates, but keep old leads that were worked, because their activity trails show how prospects moved or stalled. If cost is the concern, archive them in an export with record IDs instead of deleting. Decide with the owner in writing before any bulk delete.

How do I deduplicate without losing history?

Take a full export first, merge records, and keep a map of merged IDs to surviving IDs. Merging in the CRM can re-parent activities, but the map lets finance, support and marketing systems still reconcile. Never run bulk merges without a backup that includes attachments and notes.

Do closed-lost deals have any value after migration?

Yes, they often matter more than expected. A closed-lost opportunity with a stage trail, loss reason and activity history is a labeled outcome, which is useful for analysis and for AI training and evaluation. Whether a dataset is licensable depends on rights, size, history and the owner's decision.

Does the RevOps consultant touch the data when making an introduction?

No. The consultant makes the introduction and shares basic fit information only. The company completes the inventory with SourceX, redaction rules are agreed before any work starts, and data moves only after an executed agreement and the company's authorization.

What if the company is too small or the CRM is too new?

Then it likely does not qualify. The baseline is 50+ full-time employees at peak with contractors excluded, several years of documented operations and the right to license. A young CRM can still be exported for safekeeping, and the company can revisit eligibility as history accumulates.

Free resources

By SourceX Partnerships Team · Published 2026-10-09 · Updated 2026-10-09

Know a US company with valuable proprietary data?

Become a referral partner from anywhere we support, get your link and introduce an owner or authorized decision-maker.

Refer a company →

I own a business

Explore licensing your company's data to AI developers worldwide. Start a short assessment; no uploads needed.

Start an assessment