Matching Duplicate Records
Matching duplicate records is the process of comparing records field by field to decide which ones describe the same real-world person, company or deal. It's the first stage of deduplication: matching finds the duplicate sets, and merging then turns each set into one record.
Two records for the same customer rarely look identical. One has a typo, another has a phone number in a different format, a nickname or a work email instead of a personal one. Record matching uses a family of methods to see past those differences, from exact comparison to similar, phonetic, word-based, domain-based and AI duplicate matching. How strictly each field is compared, and how the fields are combined, decides which duplicates are found and which records are paired by mistake.
See how Dedupely matches duplicate records in your CRM
What Is Matching Duplicate Records?
Matching duplicate records, often called record matching or data matching, is the step in deduplication that answers one question: do these records describe the same thing? It applies to contacts and companies, and also to deals, leads and tickets in HubSpot; accounts and leads in Salesforce; people and organizations in Pipedrive; and leads and opportunities in Close. Each object needs its own logic, because two companies are the same for different reasons than two people are.
The matching methods form a family, and each one is built for a different kind of difference:
- Exact pairs values that are identical, often ignoring capitalization. It suits unique fields like email addresses.
- Similar allows small typos and missing letters, and can compare phone numbers without their formatting.
- Phonetic, or fuzzy, pairs names that sound alike but are spelled differently, such as Sean and Shawn.
- Shared words in any order compares the words inside a value, such as "Acme Logistics" and "Logistics by Acme."
- Domain root compares the core domain of an email or website, so sales@acme.com and jane@acme.co.uk can match.
- AI duplicate matching compares many fields and field combinations at once, catching duplicates no single rule describes.
Most matching combines several methods. Take "Jon Smith, jon@acme.com, (555) 010-2000" and "Jonathan Smith, j.smith@acme.com, 555-010-2000." An exact rule on email misses the pair. Exact on last name, similar on phone and domain root on email find it.
How Matching Duplicate Records Works
Matching turns a definition of "the same" into matching rules a system can run. It works in a few steps:
- Choose the fields. Strong identifiers, such as email, phone number or a tax ID, point to one person or company. Weaker fields, such as first name or city, are shared by many records and need support from other fields.
- Pick a method for each field. Each field gets the method that fits how its values vary: exact for emails, similar for phone numbers, phonetic for names, shared words for company names, domain root for websites.
- Combine the fields. In rule-based matching, every field in a rule has to match, so each field added makes the rule stricter. Different field combinations, such as name + email or name + phone, often get their own rules. Some tools instead give each field a weight and pair records whose total score passes a threshold.
- Group the matches. Records that meet a rule are grouped into duplicate sets, which can hold two records or many.
- Review and adjust. A sample of the sets is checked, pairs that aren't duplicates are rejected, and fields or methods are tightened or loosened.
Every choice moves the same balance. Stricter methods and more fields mean fewer false matches but more missed duplicates. Looser methods and fewer fields find more duplicates but pair more records that aren't the same. Good matching sits in the middle: strict on weak fields, more lenient where a strong field backs the match up.
Why Matching Duplicate Records Matters for CRM Teams
Matching decides what deduplication can fix. A duplicate that no rule finds stays in the CRM: it keeps inflating contact counts and pipeline totals, splits activity history across records, and gets counted again in CRMs that bill by record. Two reps can still call the same lead, and a customer can keep getting emails after unsubscribing on the other record.
A false match does the opposite kind of damage. Two different customers who share a name or a generic company domain end up in one duplicate set, and if that set is merged, their histories, owners and contact details are mixed into one record. In many systems a merge is hard or impossible to reverse, so the cost of a loose rule shows up after the fact, often in billing, routing or reporting.
Matching quality also decides how much of deduplication can run on its own. A rule that has proven clean on several scans can be trusted to run on a schedule as new records come in from forms, imports and syncs. A rule that sometimes pairs the wrong records needs a person to check every set. The clearer and better tested the matching rules, the less day-to-day review the CRM needs.
Getting Started With Matching Duplicate Records
Start strict and loosen step by step. Match on strong fields first, such as email or a customer ID, and check what each rule finds before you merge anything. A sample of the results shows quickly whether a rule pairs the right records.
Widen a rule only where real duplicates are being missed. Switch the field that breaks the match to a looser method, such as similar matching for phone numbers or phonetic matching for names, and keep at least one strong field strict. Pairs that turn out to be two different records should be rejected, so they don't come back in the next run.
Change one field at a time. When several fields change at once, it's hard to tell which change found the new duplicates and which one let the false matches in. Keep a note of what each change found, so the rules that work can be reused for the next object.
In Dedupely, each field in a Duplicate Search gets its own Match Option, from Exact to Similar, Fuzzy, Similar Words and Domain Root, so you can loosen one field without touching the others. Rejected Matches keeps the pairs that aren't duplicates out of future results.
Key Takeaways: Matching Duplicate Records
Matching duplicate records is the first stage of deduplication: it compares records field by field and groups the ones that describe the same person, company or deal into duplicate sets. A family of methods covers different kinds of differences, from exact and similar to phonetic, shared-word, domain root and AI duplicate matching. How strictly each field is compared, and how fields are combined, sets the balance between missed duplicates and false matches. In Dedupely, Duplicate Searches hold the rules, Match Options set how each field is compared, and Rejected Matches keep false pairs from coming back. Dedupely Intelligence finds what the searches miss. For the searches you trust, Auto Match notifies you when new matches reach a count you set, and Auto Merge merges new duplicates on a schedule.
Frequently Asked Questions About Matching Duplicate Records
Which match method should you use for names, emails, phone numbers and company names?
Use the strictest method each field can handle without missing real duplicates. Email addresses and IDs are unique, so exact matching suits them. Phone numbers often differ only in formatting or country codes, so a method that ignores formatting works better than exact. Personal names vary through typos, nicknames and alternate spellings, so similar or phonetic matching helps, as long as it's paired with a stronger field like email, phone or company. Company names often carry suffixes like Inc. or Ltd., or list their words in a different order, so word-based matching with a list of ignored terms fits them. Websites and email domains compare best on the root domain. In Dedupely, each field in a Duplicate Search gets its own Match Option: Exact, Similar, Fuzzy, Similar Words, Domain Root or None. The Match Options guide suggests starting with Exact on high-impact fields like name, email or phone.
Why does a matching rule miss duplicates you can see in the CRM?
Often, one field in the rule doesn't meet its criteria. In rule-based matching, every field in a rule has to match, so a single blank value, a different phone format, a nickname or hidden formatting can break the pair even when the other fields agree. Rules with many fields miss more for the same reason. To find the cause, compare the two records field by field and look for the one that differs. Then loosen that field's method or take it out of the rule. In Dedupely, every field in a Duplicate Search must meet its criteria, and hidden formatting can stop an Exact match. Switch that field to Similar or Fuzzy, or set it to None, then click Scan to update the results. Clone the search first to test the change without touching the active one. Dedupely Intelligence also checks contacts, companies and leads for field combinations your searches don't cover.
How do you stop matching rules from pairing records that aren't duplicates?
Back every loose comparison with a strong field. A rule on first name + last name alone pairs people who just share a name, while name + email or name + company + phone points to one person. Leave common words like "Inc." or "Group" out of company name comparisons, keep lenient methods off fields that many records share, and check a sample of each rule's results before merging anything. When a pair turns out to be two different records, reject it so it stays out of future results. In Dedupely, Safety Settings are on by default and hold back matches with more than 20 records, as well as matches built on poor or placeholder values such as "John Doe," in an Excluded tab for a closer look. Similar Words can skip up to 50 Ignored Terms. Rejected Matches keep false pairs out of current and future Duplicate Searches, and Manual Review holds uncertain sets until a person merges or removes them.
How do you match company names written in different ways, such as with and without "Inc."?
Compare the words in the name instead of the whole value, and ignore the words that don't identify the company. Legal suffixes like Inc., LLC and Ltd., and generic words like "Group" or "Club," make two names look different even when they belong to one company. Word-based matching that skips those terms and allows the words in any order pairs "Acme Logistics Inc." with "Acme Logistics." Matching on the first word only goes further, but it can also pair different companies that share that word, so combine it with a second field like city, phone or website domain. In Dedupely, the Similar Words Match Option compares shared words in Any Order, or the First or Last words, and lets you set how many words must match. It skips up to 50 Ignored Terms, such as Inc., LLC and Ltd. Domain Root on the website or email field then confirms the match with the company's core domain.
How do you automate matching duplicate records without merging the wrong ones?
Automate only the rules that have proven clean. Run each rule by hand first, check what it finds over a few scans, and reject the pairs that aren't duplicates. Rules built on several strong fields, such as email or name + company + phone, tend to give the cleanest results, and those are the ones worth putting on a schedule. Rules built on looser methods or weaker fields are better kept for review. In Dedupely, automation is set per Duplicate Search with Auto Match and Auto Merge. Auto Match checks for new duplicates on a schedule and notifies you when matches reach a count you set; it never merges. Auto Merge merges new duplicates on the schedule you choose. Because merges in Dedupely can't be undone, keep uncertain sets in Manual Review. Auto Match and Auto Merge are turned off in Dedupely Intelligence, so its matches are always reviewed and merged by hand.
