Decide what counts as a duplicate
Picture a small service business the week before a promotional email. Marketing pulls a list of “active customers.” The CRM shows three records that look like the same person: one from a website form last year, one from a trade-show badge scan, and one entered by hand when they called the office. One record has a do-not-contact flag. Another has the email they use today. If someone merges the wrong way—or skips the merge and exports all three—the same person can get three messages, or the do-not-contact flag can disappear into a deleted record.
That scene is an illustration, not a TwoQM client. Many Rio Grande Valley businesses already collect leads into a CRM and still treat duplicates as a last-minute cleanup. This article is about the rules that decide which records are duplicates, who may merge them, which fields win, and what must pass before any marketing send. It is not open-house lead capture into a CRM, which covers how a new inquiry lands in the system. This piece starts after those records already exist and someone is about to email, text, or build an ads audience from them.
Software can suggest matches. People confirm them. Write down the match keys your business will trust, in order:
Exact email (normalized: lowercase, trimmed spaces).
Exact phone (digits only, same country format).
Company name + person name when both are present and email or phone is missing.
Near-matches (similar names, shared company, close phone numbers) stay in a review queue. They are not merges until a named person says so. If two records share an email but different phones, treat that as a conflict for a person to resolve, not an automatic merge.
Who may merge, and what fields win
Merges change history. Only roles you name may confirm them—often an operations lead or a CRM owner, not every salesperson. When two records become one, write field-win rules ahead of time:
Contact methods: keep the email and phone marked “preferred,” or the ones used most recently in a logged conversation.
Do-not-contact and suppress flags: never lose them. If either record is suppressed, the survivor stays suppressed until a person lifts the flag on purpose.
Source and first-touch date: keep the earliest first-touch and both source values in an audit note when both matter.
Owner and open tasks: keep the current owner unless your written rule says otherwise; open tasks move with the survivor.
Every merge writes an audit line: who confirmed it, when, which record IDs were joined, and which fields were kept. Corrections are new lines; the original merge stays visible.
Checklist before any marketing send
Here is an original example, not a measured result. A twelve-person Valley firm freezes exports the night before a newsletter and runs this list:
Open duplicate suggestions older than the firm’s review window are cleared or parked with a named owner.
The send candidate list is built from the survivor records only—no raw import staging tables.
Suppress and do-not-contact flags are checked on the survivor after recent merges.
Bounce and invalid-address flags from the last campaign are still honored.
A named human reviews a sample of the list (or the full list when it is small) and records approval with a date and their name.
Only then does marketing export or sync the audience.
That approval step is not optional theater. It is the line between “software prepared a list” and “a person authorized this send.” Business automation can stage the candidate list and alert the reviewer when items are waiting. It should not send the campaign on its own.
Where AI helps, and where it stops
AI can draft a short summary of why two records might match, suggest which email looks preferred, or prepare a Spanish note for bilingual staff who review merges. A person still confirms every merge. AI never merges contacts alone, never clears a suppress flag, never builds a final audience without the named approve step, and never sends marketing email, SMS, or ads.
Connect it to the systems you already run
Your CRM stays the system of record for contacts. Email and SMS tools should pull from survivor records after hygiene rules run, not from ad-hoc spreadsheets. Ads audiences built from CRM exports follow the same checklist. TwoQM’s CRM work and connected systems work are about recording those rules and handoffs around tools you already use—not promising open rates, deliverability, or spam outcomes. Nothing here promises campaign results.
Where to start this week
Print your match keys and field-win rules on one page. Name who may merge. Name who approves send candidate lists. Run the checklist on the next campaign before anyone exports. Bring that one-pager when you want tooling that suggests duplicates, keeps an audit trail, and stages lists for human approval.
If you want those merge rules and pre-send checks recorded in one place, TwoQM can scope CRM hygiene and automation around your existing stack. Pricing explains how business apps, systems, and custom software are scoped. TwoQM builds tooling; it does not send your marketing for you, and no software makes a list “safe” without a named person reviewing it.