Deduptio

How to normalize phone numbers in Attio to find duplicates

The short answer

Attio normalizes a phone number for you, but only if the number reaches a phone number attribute with its country code. It stores the result in E.164, such as +14155550132. So the work is in three places: convert numbers to E.164 before an import, because Attio skips the ones it cannot validate; move numbers out of text fields, where nothing normalizes them; then look for shared numbers yourself. Attio does not flag two people who hold the same phone number.

This page is about the phone values themselves and about matching on them. The full match rule for people, including email, LinkedIn and name, is in finding duplicate people in Attio.

What Attio already does to a phone number

Five facts from Attio's documentation decide how much of the work is left to you.

The standard Companies object has no phone attribute at all. A company switchboard number in your workspace sits in a custom attribute, and its type decides whether Attio normalized it.

Per Attio's phone number attribute type, People object, create and manage attributes, CSV import formatting guide and merge and delete records pages, checked November 14, 2026.

Where unnormalized numbers still hide

Inside a phone number attribute, (415) 555-0132 and +1 415-555-0132 are already one value. The numbers that break matching are the ones Attio never normalized.

WhereExampleWhat resolves it
The source file, before import020 7946 0958, no country codeConvert to E.164 before importing; Attio skips the value otherwise
A text attributeA custom Mobile text field holding 415.555.0132Convert, then move the value into a phone number attribute
A number with an extension+1 415 555 0132 x204Split the extension into its own attribute
A number under the wrong countryAn Australian mobile imported with +1 in frontFix the country code from the record's location
Notes and email signatures“Call me on 0412 345 678”Nothing automatic; add the number to the record by hand

The first row costs the most. In the import preview a value in the wrong format shows an Invalid format error. If you do not correct it, the import skips that value and creates the record without it. A file of 5,000 people with national-format numbers arrives as 5,000 people with no phone, and a phone match has nothing to compare.

Convert numbers to E.164 before they reach Attio

Work in a copy of the source file. Keep the original column, and write the result into a new one.

  1. Split off extensions. Move anything after x, ext or #into a separate column. E.164 has no place for an extension, and Attio's documentation does not mention one.
  2. Strip the formatting. Remove spaces, brackets, dots and hyphens. Keep the digits and a leading +. In Google Sheets: =REGEXREPLACE(A2,"[^0-9+]","").
  3. Turn an international prefix into +. 0044 20 7946 0958 is how the number is dialled from many countries. Replace the leading 00 with +. From the US and Canada the prefix is 011.
  4. Add the country code to national numbers. A number with no +needs a country. Take it from the record: the person's location, the company's country, or the source system's region setting. A 10-digit US number 4155550132 becomes +14155550132.
  5. Drop the trunk zero where the country uses one. A UK number written 020 7946 0958 is +442079460958: the leading zero goes. This rule is per country. Italian landlines keep their zero after +39.
  6. Import, and read the preview. Map the new column to Phone numbers. Correct every Invalid format cell in the preview by clicking into it, or the value is skipped.

Steps 4 and 5 are where a spreadsheet formula stops being enough. For a file that mixes countries, run the column through Google's open-source libphonenumber library, which knows each country's rules and takes a default region per row. Do not default every row to +1. Attio's validation rejects some wrong guesses, because it checks area codes, but a wrong guess that happens to form a valid number is stored as a different person's phone.

A phone number does not stop an import from creating duplicates. Attio matches imported people to existing records on email address, the unique attribute. How to map it is in duplicates from a CSV import.

Move numbers out of text fields

A Mobile or Direct linecolumn mapped to a text attribute never passed Attio's validation. Every tool that compares text compares it as typed, so 415.555.0132 and +14155550132 are two values.

  1. Export the people to CSV with the text field and each record's Record ID.
  2. Convert the text column to E.164 with the six steps above.
  3. Re-import the file with the record ID mapped, and the converted column mapped to Phone numbers. With a record ID, Attio updates the existing record and skips rows whose ID matches nothing.
  4. Keep the text field as a record of what the source system said, or delete it.

Test the re-import on ten records first. Attio documents that an import adds multi-select values to the existing ones, but it does not say this about phone numbers specifically. Check that a person who already had a number now has both.

Find the people who share a number

Once the numbers sit in a phone number attribute in E.164, equal numbers are equal strings. Attio will not point them out, so you look for them one of three ways.

{ "filter": { "phone_numbers": "+14155550132" } }

A shared number is evidence, not proof

A phone number identifies a line, and a line is not always one person. Four cases produce a correct match on the number and a wrong merge.

The rule that follows: treat a shared mobile number plus the same name as a duplicate. Treat a shared number with different names as colleagues until you have looked. A number that sits on five people at one company is almost certainly a switchboard; leave those records alone.

To merge a confirmed pair, open one record, click the ⋮ icon in the top right, choose Merge records, add the other record and put the one whose values you trust on the right. Attio merges people and companies, one pair at a time, and documents no undo. Its merge page does not say how the phone numbers of the two records combine, so open the result and check that both numbers are there. The steps in full are in how to merge records in Attio.

What Deduptio normalizes, and what it does not

Deduptio compares phone numbers by their digits. When a match rule keys on an attribute of Attio's phone number type, every value is reduced to digits, and a number longer than 10 digits is compared on its last 10. Every value of a multi-value attribute is compared, so two people match if they share any one number. Nothing is written back to Attio; only the comparison value is normalized.

The limits, stated plainly:

You decide per group before anything merges, and merges run through Attio's native merge. Details are in match rules and plans.

Find the people who share a phone number

Connect your Attio workspace, write one match rule, and run a scan. Scans are read-only — you see every group and the reason it matched before anything is merged.

Start a free Attio duplicate scan

Checklist

  1. List every attribute that holds a phone number, and note which are text.
  2. Before each import, convert the phone column to E.164 and split off extensions.
  3. In the import preview, correct every Invalid format cell; uncorrected values are skipped.
  4. Move numbers from text attributes into Phone numbers, testing on ten records first.
  5. Group people by number; require the same name before calling a pair a duplicate.
  6. After each merge, confirm both records' numbers are on the result.

Related