A dirty CRM silently taxes every forecast, every campaign, and every rep. This guide walks you through a repeatable cleanup playbook used by RevOps teams to audit the mess, dedupe without blowing up account history, fill the gaps, standardize the picklists, and lock in ongoing hygiene jobs so you never land back in the swamp. Follow the steps in order. Nothing here is optional if you want numbers you can trust.
Time: 1-2 weeks for first pass, ongoing after
You cannot clean what you cannot measure. Pull hard numbers before you touch a single record. Export counts for every object and compute completeness percentages for the fields reports and workflows actually rely on. Most teams discover that half their custom fields are below 20% populated and that one picklist has 14 variants of the same stage. Capture these metrics in a shared dashboard so you can show real cleanup progress to leadership, not vibes. This audit becomes your baseline and the yardstick you measure the cleanup against.
Dedupe fails when teams argue about matching logic mid-merge. Lock the rules before you run anything. Pick a primary key per object (normalized email for Contacts and Leads, normalized domain plus company name for Accounts), then layer secondary keys for near-matches. Decide the merge-winner precedence (newest activity, highest completeness, owner tenure) and write it down. Get sales and marketing to sign the document. The rule set should be short enough to fit on one page and specific enough that two admins would merge the same way.
Execute the merges in a sandbox first, then in production in waves (Leads, Contacts, Accounts, Opportunities). Always merge, never hard-delete, when the loser record carries history, activities, or an owner who is still an active user. Strkr treats active users as never-hard-deletable, and the same discipline applies to the records they own: merging preserves the history and reassigns relationships cleanly, while a delete orphans activities and breaks audit trails. Run merges in batches small enough to review and reversible via your backup.
With duplicates gone, focus on the fields your reports and automations actually read. Export records with empty required fields and route them to the right owner for enrichment, not to a central admin who does not know the account. For accounts, use enrichment providers or domain-based lookups to fill industry, employee count, and country. For contacts and leads, use the same enrichment plus owner-driven corrections. Set a deadline. Any record that cannot be enriched by the deadline moves to archive, not into reports.
Picklist drift is the silent killer of dashboards. Collapse variants of the same value into one, then lock the field so free-text entry is disabled. Industry, Country, Lead Source, Status, and Stage are the usual offenders. Build a one-time crosswalk from every historical value to the new canonical value, run a bulk update using the crosswalk, and only then publish the new picklist. Document the canonical list in your admin wiki so new teammates do not reinvent the mess six months later.
Not every old record deserves to clutter active views. Define an archive policy: Leads with no activity in 18 months, Opportunities closed-lost over 24 months old, Contacts whose owning Account is archived. Flip these to an Archived segment or inactive status rather than deleting them. Hard-delete is reserved for true noise (bot form fills, test rows, obvious junk) and is never applied to records owned by an active user without an explicit owner-change step first. Archiving preserves audit trails, keeps compliance happy, and keeps the active workspace focused.
A clean CRM dirties fast if the front door is wide open. Add validation at every point where records are created: web forms, import jobs, API endpoints, manual entry, integrations with marketing automation. Required fields on create, format validation on email and phone, domain-based company deduping on Lead create, and picklist-only values on categorical fields. Validation is cheap at the entry point and expensive after the fact. The cleanup you just finished is only worth it if you harden the pipes behind it.
Data hygiene is a role, not a project. Assign a data owner per object (Accounts, Contacts, Leads, Opportunities, Activities), publish a monthly hygiene dashboard, and schedule recurring automations that run the cleanup rules continuously: nightly duplicate scans, weekly fill-rate reports, monthly archive sweeps, quarterly picklist audits. The team that pays attention to hygiene for 30 minutes a week never has to run another two-week cleanup project. The team that ignores it ends up right back here in 18 months.
The first full pass typically takes one to two weeks of focused effort for a mid-sized CRM: a few days to audit and define rules, a few days to dedupe and enrich, and a few more to standardize picklists and stand up the ongoing hygiene jobs. After that, hygiene is a 30-minute weekly ritual, not a project. Larger datasets with heavy custom-object usage can extend to three or four weeks.
Merge, almost always. A merge preserves activities, notes, attachments, and ownership history on the surviving record and reassigns relationships cleanly. Hard-delete is reserved for true garbage like bot form submissions, test rows, and obvious junk leads. Never hard-delete a record owned by an active user without first reassigning ownership and reviewing the attached history.
For Contacts and Leads, the primary key is a normalized email address: lowercased, plus-aliases stripped, trimmed of whitespace. For Accounts, use normalized domain plus normalized company name together, because many enterprise accounts share domains and many small businesses share names. Layer secondary fuzzy keys like phone or address only to flag candidates for human review, not for automatic merges.
Harden the entry points and schedule the hygiene jobs. Validation on web forms, imports, APIs, and manual entry stops most new mess at the door. Nightly duplicate scans, weekly fill-rate reports, monthly archive sweeps, and a quarterly picklist audit catch the rest. Assign an owner per object, review a hygiene dashboard monthly with sales and marketing leadership, and the problem stays solved.
Archive them. Flip the status to Archived or move them into an Archived segment, exclude them from default views and routing automations, but keep them searchable for compliance and audit. Archiving is the right default for closed-lost opportunities over 24 months old, leads with no activity in 18 months, and contacts whose owning account is itself archived.
RevOps or a named data owner, with an owner assigned per object. The owner publishes a monthly hygiene dashboard, runs the recurring cleanup jobs, and brings deltas to a 30-minute monthly review with sales and marketing leadership. Without a named owner and a standing meeting, the dashboard turns green once and nobody looks at it again until the next crisis.
Strkr ships with dedupe-on-create, picklist locks, and built-in hygiene dashboards so your team spends time selling, not scrubbing. Start free or take the full tour.
No sales call, no migration consultant, no four-month implementation. Enter your card, get 14 days of the full Pro tier, cancel any time before day 14 with zero charge. Spin up a workspace, import your CSV, and have something useful before lunch.
