How to merge two business lists into one without losing records
After an acquisition, a team merger or a new CRM, two lists of accounts must become one. The order to match records in, which values win, and how to keep the result clean.
Part of the guide Location intelligence automation
To merge two business lists into one, put both into the same columns, match records in order of how reliable the evidence is (same phone number first, then same website, then the same name at the same spot), decide field by field which list's value wins, and keep a column that says where each row came from. Rows that match on nothing are added as new; rows that match on weak evidence go to a person to confirm. Done this way, nobody loses an account and nobody gets two calls.
This is for operations, sales operations and data teams joining lists after one company buys another, after two regional teams combine, or when moving from spreadsheets into a CRM. It walks through preparing the columns, the matching order, the rules for which values survive, a made-up merger in Mexico City, the mistakes that cause lost or doubled accounts, and how an AI employee keeps the merged list tidy afterwards.
Prepare both lists first
Most of the work happens before any matching. Give both lists the same column names: name, address, city, phone, website, latitude, longitude, and each list's own account ID. Then tidy the values that matching depends on:
- Phones: keep digits only, and keep the country code. Matching on the last nine digits ignores the different ways people type the start.
- Websites: strip https://, www. and anything after the domain, so a menu page and a home page count as the same site.
- Names: leave them as they are. Rewriting names by hand creates differences instead of removing them.
- Positions: add them where missing by looking each business up again by its name near its address. A list without positions can only match on text.
Add a source column to each list (A or B) before you start. It is the cheapest insurance you will buy: every question later about where a row came from has an answer.
Match in order of how sure you can be
- Same phone number. The strongest common key for small businesses. Two records with the same last nine digits are almost always the same business.
- Same website domain. Strong for companies, weaker for branches of a chain, which share one domain. For chains, add the city or position to the test.
- Same name at the same spot. Names compared without capitals or accents, and positions within a short distance, such as 60 metres.
- Similar name only. Never merge on this alone. Send these pairs to a person with both records side by side.
Decide which values win
When two records are the same business, their fields can still disagree. Set the rules before you start, not row by row:
- Account IDs: keep both. The merged row carries list A's ID and list B's ID, so either system can find it.
- Contact details: the most recently checked value wins. If neither has a date, look the business up again and take what it shows now.
- Commercial fields (terms, owner, segment): the list of the company that will own the account wins, and the other value is kept in a note column.
- Status: if either list says closed, check before keeping the account active.
A made-up merger in Mexico City
All the numbers here are invented. Say a food wholesaler buys a smaller rival, and both sell to restaurants and grocers across Mexico City. List A has 900 accounts; list B has 600.
| Step | Pairs found (made up) | Left unmatched in B |
|---|---|---|
| Same phone | 210 | 390 |
| Same website | 35 | 355 |
| Same name within 60 m | 48 | 307 |
| Similar name only, sent for review | 22 | 285 added as new |
Invented figures for illustration only.
The merged list ends with 900 plus 285 rows, plus whichever of the 22 reviewed pairs turn out to be different businesses. Each of the 293 merged rows keeps both IDs. The 22 reviewed pairs took one person an afternoon, which is far less than sorting out the doubled calls later.
Mistakes that lose or double accounts
- Merging on name alone. Two pharmacies called the same thing in different districts become one record, and one customer disappears.
- Merging all branches of a chain into one. Same website, same name, different addresses: they are separate accounts for a field team.
- Dropping the losing list's ID. Invoices and history in the old system then point to nothing.
- Deleting instead of flagging. Mark the duplicate and keep it for a while; deleting it makes mistakes impossible to undo.
- Stopping after the merge. New rows keep arriving from both teams. Without a check on every addition, the duplicates come back within months.
Keeping the merged list clean with an AI employee
The one-off merge is a project; keeping it clean is a routine an AI employee can run. In xMap Flows, upload the merged list to My data, then build:
- New rows: starts whenever a team adds rows to the shared table or sheet.
- Map business scan: looks each new row up by its name near its address, adding the current phone, website, position and status.
- Match with my table: checks each new row against the merged list by phone, then a second match by website, keeping all and adding a match column.
- Remove duplicates: one of each, by the same place, keeping the first.
- Branch: matched rows go to a review email, unmatched ones to the list.
- Email: a short daily note of possible duplicates for a person to confirm.
What lands in the inbox is a list of new rows that look like existing accounts, each with both records and the evidence ("same phone" or "same name, 40 m apart"). The Place classifier can also give both lists one set of categories, so "Grocery" in one list and "Abarrotes" in the other read the same. The Monthly outlet census of my city template rebuilds a full outlet list each month to compare against.
Questions
What is the best field to match two business lists on?
The phone number, compared by its last nine digits. Then the website domain, then the name together with the position. Never the name on its own.
Should I delete duplicates after merging?
Flag them and keep both IDs on the surviving record for a while. Deleting makes a wrong merge impossible to undo.
How do I handle branches of the same chain?
Treat each address as its own account. Branches share a website and often a name, so add the position or address to the test.
How long does a merge stay clean?
Only as long as new rows are checked. A routine that checks every addition against the merged list stops duplicates coming back.
Templates that do this
Read next
More on business and places data · Every article · Templates