Audit and clean CRM data quality

Audit CRM data quality, detect duplicates, flag stale records, score completeness

CRM admin or marketing ops cleaning up legacy data before a migration or merge campaign—measure health, detect garbage, plan sprints.

• · CSV export file path (contacts, companies, or deals)
• · Data type (contacts/companies/deals)
• · CRM system (HubSpot, Salesforce, Pipedrive, other)
• · Critical fields to measure (email, phone, company name, deal stage, last activity, owner)

• · Data profile: total records, columns, data types, fill rate per column, date range, unique vs. total
• · Duplicate groups with confidence levels (HIGH: exact email match; MEDIUM: fuzzy name match or cross-field)
• · Stale record flags (last activity threshold)
• · Row completeness percentage
• · CRM hygiene quality score (0-100)
• · Prioritized cleanup plan with merge/delete recommendations

• · Bash + Python one-liners for CSV parsing and fuzzy matching

• · CSV export exists at provided path
• · At least 10 rows of data (optionally samples first 10K rows for large CSVs)
• · Standard CRM column names recognizable

• · CSV encoding mismatches (UTF-8 vs. Latin-1) → garbled data
• · Column mapping wrong (email in 'Email Address', fuzzy names) → duplicates missed
• · No date columns → cannot detect stale records
• · Critical field list not provided → defaults to generic columns, missing business context

• · Four-step methodology with explicit skip warnings
• · Three types of duplicates with confidence levels and normalization rules
• · Fuzzy match criteria: company name normalization (strip legal suffixes, punctuation), name variants (Bob/Robert), edit distance ≤2
• · Large CSV sampling rule (first 10K rows for samples ≥ 10K, extrapolate results)
• · Recommendation per duplicate group with 'keep most complete + recent' rule