Sales sets up territory routing by country. Leads from the United States should go to the Americas team. Within a week, a third of American leads are landing in the default queue. The routing rule checks for “United States”. The CRM also contains “US”, “USA”, “U.S.”, “united states” and “America”, because one form uses a dropdown, another uses a free-text field, and an old import used two-letter codes. The workflow is fine; it’s reading a field that was never standardized.
That pattern is behind a lot of automation that “doesn’t work”. Segments miss people, lead scoring counts the same person twice, a renewal sequence goes to a contact whose company churned last year. Cleaning the data is usually less exciting than building a new workflow, and it’s often the thing that makes the existing ones finally behave.
Why does CRM data get messy?
CRM data gets messy because many people and systems write to it, each with slightly different rules, and nothing enforces one standard. Forms, imports, integrations, workflows and manual edits all set values. Over a few years, the same field ends up holding five spellings, two formats and a lot of blanks.
Some decay is unavoidable. People change jobs, email addresses stop working, companies get acquired. The rest comes from setup: free-text fields where a dropdown was needed, imports that were never mapped carefully, integrations that create instead of updating, and no one owning a field’s definition. A cleanup fixes the current state. Fixing the causes is what keeps it clean.
What we clean
How a cleanup runs
Profile the data
We measure the problem before changing anything: how many duplicates by which match rule, which values exist in each important field, how many records are missing key data, which states look wrong. You see the size of each issue before agreeing what to fix.
Agree the rules
For each issue we write a rule with you. Which record wins in a merge. What the standard country list is. When a contact counts as stale. What lifecycle stage a contact with a closed-won deal should be in. These decisions belong to the business, and we bring a recommendation for each.
Back up
The affected records and properties are exported before any change, so every fix can be checked and reversed if needed.
Fix in batches
Changes run in controlled batches, starting with a small sample you can check. Merges, bulk updates and archiving each follow the agreed rule. Anything ambiguous goes to a review list rather than being guessed.
Prevent recurrence
We find the sources of the mess and close them: dropdowns instead of free text, form and import validation, integration matching rules, workflows that keep lifecycle stages current, and clear ownership for key fields.
Is bad data making your automation misfire?
We'll profile your database, show you the size of each data problem and recommend what to clean first.
Merging duplicates without losing history
Merging is where cleanups go wrong most often, because many platforms make merges hard or impossible to undo. Before merging, we decide which record survives and how conflicting values are resolved. The most recently updated email address may be right, while the oldest create date and original source should be kept. Ownership, lifecycle stage and consent each need their own rule.
Consent deserves special attention. If one duplicate is unsubscribed and the other isn’t, the merged record must stay unsubscribed. We check how your platform handles this in a merge rather than assuming, and fix the result where it doesn’t.
Stale contacts: suppress, archive or delete?
There’s no single answer, and it depends on the platform and your use for the data. Hard bounces and spam complaints should be suppressed from sending in every case. Long-term non-engagers can usually be moved out of active marketing segments, given a short re-engagement attempt, then suppressed or archived. Deletion makes sense for test records, junk and data you have no reason to keep, and it can matter for privacy obligations. Our guide to email list hygiene covers the sending side in more depth.
On platforms that bill by contact count, such as HubSpot marketing contacts or Klaviyo active profiles, cleanup can lower cost too. We’ll show you which records are counted before deciding what to change.
Keeping it clean afterward
A one-off cleanup without prevention lasts a few months. The most useful long-term step is clear property governance: each important field has a definition, an owner, a list of allowed values and a record of which systems write to it. A data dictionary holds all of that in one place.
From there, most of the prevention is configuration. Validation on forms. Matching rules on integrations so they update instead of create. Import templates with fixed column mappings. A small set of workflows that keep lifecycle stages and associations current. And a scheduled check, monthly or quarterly, on the handful of fields your automation depends on most. If you’d like us to run those checks, they fit naturally into CRM optimization.
Cleanup is quoted after profiling, because the effort depends on database size, the number of issues and how much needs a human decision. Related starting prices are on the pricing page.
How long will a cleanup take for us?
It depends on the size and state of the database. A focused cleanup of one CRM usually takes a few weeks, including profiling, agreeing rules and running the fixes in batches. The rule-setting step often takes longer than the fixes themselves.
Can’t we just use a deduplication tool?
Tools help, and we use your platform’s own duplicate management and specialist apps where they fit. A tool still needs rules for which record wins and which values to keep, and it won’t fix inconsistent values, wrong lifecycle states or broken associations. The rules are where the work is.
Will cleanup break our existing workflows or reports?
It can change what they do, which is often the point. A workflow filtering on “United States” will suddenly catch everyone it was missing. We check which workflows, segments and reports read each field before changing it and tell you what will behave differently.
Do you delete contacts during a cleanup?
Only where you’ve agreed a rule for it, and only after the affected records are backed up. Test records and obvious junk usually go. Old but legitimate contacts are more often suppressed or archived, so history and consent records stay intact. If you’re about to change platforms, cleanup tends to run as the first phase of a CRM migration, and deciding what to leave behind becomes part of the mapping.
Can you clean data in Klaviyo or another email platform as well as the CRM?
Yes. Ecommerce and email platforms collect duplicates, stale profiles and inconsistent properties too, usually from store syncs, sign-up forms and imports. The method is the same; the platform specifics differ. See our Klaviyo and HubSpot pages for how we work in each.
Is there a guide if we want to do it ourselves?
Our CRM data cleanup guide walks through the process step by step. If you’d rather know where you stand first, the CRM Health Check gives a quick read on your data and other areas.