The answer
Largely. Finding duplicates, merging exact matches and blocking new duplicates at the point of entry are highly automatable today. In our model about 75% of the work can go to software. What stays human is reviewing fuzzy matches, such as a shared office phone number, and settling arguments over who owns a merged contact.
The tasks, and how much of each is automatable today
Six tasks make up duplicate merging. Blocking duplicates at entry, finding them and merging them are 90 to 95% automatable. Choosing which field values survive is about 60% automatable with rules. Fixing attribution after a merge is about 60% automatable. Settling ownership disputes between reps stays human, because it is a fairness question, not a data one.
| Task | Min / week | Automatable |
|---|---|---|
| Find duplicate contactsExact matches on email and normalised phone are reliable. Name-only matches are not, and should never auto-merge. | 90 | |
| Decide which record and which values to keepRules like 'most recently updated wins, but keep the oldest source' cover most cases. Conflicting notes need a person. | 60 | |
| Merge the recordsCRM merge APIs move activities and deals onto the surviving record. Check that custom objects follow too. | 60 | |
| Stop duplicates being created at entryA lookup before create, on normalised email and phone, prevents most duplicates from ever existing. | 45 | |
| Fix lead source and ownership after a mergeMerges can overwrite the original lead source. Rules can preserve the first-touch value; odd cases need review. | 45 | |
| Settle disputes over who owns a contactSoftware can apply a rule, such as first owner wins. Reps who disagree need a manager, not a workflow. | 30 | |
| Whole process | 330 |
Hours saved per week
For a sales team with a 25,000-contact CRM adding about 300 contacts a week from forms, imports and calls, duplicate handling takes around 5.5 hours a week by hand. Automating the tasks above returns about 4 hours a week. The rest is reviewing uncertain matches and the occasional ownership dispute.
Modelled on: A sales team with a CRM of about 25,000 contacts, adding around 300 a week from web forms, list imports, calls and DMs.
- Around 5% of new contacts arrive as duplicates of an existing record.
- About 90 minutes a week spent searching for and identifying duplicates.
- About 2 minutes per merge, including choosing which values to keep.
- Around 30 minutes a week resolving which rep owns a merged contact.
Tools that do the job
You need matching rules, a merge step that keeps the right values, and a check before new contacts are created. HubSpot and Salesforce have built-in duplicate tools for exact matches. n8n or Make add custom rules, such as normalised phone numbers, and a lookup before every create. GoHighLevel can be set to dedupe on email or phone.
- HubSpot
- Built-in duplicate detection and merge, with an API for custom rules.
- Salesforce
- Matching and duplicate rules that can block or alert on new duplicates.
- GoHighLevel
- Contact deduplication setting on email or phone for inbound leads.
- n8n
- Normalises phone numbers and emails, runs the lookup before create and builds the weekly review queue.
- Google Sheets
- Review queue for fuzzy matches, showing both records side by side.
A simple workflow
The workflow has two parts. Prevention: before any contact is created, look up the normalised email and phone, and update the existing record instead. Clean-up: a weekly job finds duplicates, auto-merges exact email matches, and queues fuzzy matches for review with both records side by side.
- NormaliseLower-case emails, strip spaces, and convert phone numbers to international format before anything else.
- PreventBefore creating a contact, search on normalised email and phone. If found, update the existing record.
- FindWeekly job scans for duplicates created by imports or integrations that skipped the check.
- MergeExact email matches merge automatically, keeping the first-touch source and the current owner.
- ReviewPhone-only and name-based matches go to a review queue with both records shown.
Keep these human
- Reviewing phone-only or name-only matches. Families, offices and agencies share numbers; merging them joins different people.
- Ownership disputes. Who gets credit for a contact affects pay and morale.
Where it goes wrong
- Auto-merging on phone number alone. Two people at one office become one contact, and messages go to the wrong person.
- Losing attribution. A merge that keeps the newest lead source erases where the customer first came from. Protect first-touch fields.
- Cleaning without preventing. A one-off clean-up is undone within months if new duplicates keep arriving. Fix entry first.
What most guides miss
Most guides explain how to merge duplicates. The bigger risk is merging the wrong ones. Only auto-merge on exact, normalised email. Send phone-only and name matches to review, because shared numbers are common. And protect first-touch lead source during merges, or your marketing reports quietly lose their history.
The evidence
- Poor data quality costs organisations an average of $12.9 million a year.Gartner, How to Improve Your Data Quality (2021)
- Sales reps spend 70% of their time on non-selling tasks, including manually entering and fixing customer data.Salesforce, State of Sales, 6th edition (2024)
- Data processing has a technical automation potential of 69%.McKinsey Global Institute, A future that works: Automation, employment, and productivity (2017)