• 228 destinations
  • No KYC, ever
  • Bitcoin, Monero, USDT and 4 more

CampaignsBeginner8 min readUpdated 18 October 2026

Preparing a contact list: E.164, deduplication and cleaning

E.164 with worked examples by country, spreadsheet traps, deduplication, consent columns and landlines: how to prepare a list that imports and sends clean.

Part 3 of 4 in the path Your first week

Files the importer reads

CSV, TXT and XLSX, up to 50 MB per file and up to 1,000,000 numbers per campaign, with or without a header row. The first column that looks like phone numbers is used as the destination; every other column becomes a variable you can use in the text, such as {first_name}. Paste a list directly if it is short.

The layout that causes the fewest surprises is one row per person, the phone number first, then only the columns the message will use, with short lower-case names that are easy to type in braces:

CSV
phone,first_name,store
+447400123456,Amelia,Leeds
+33612345678,Louis,Lyon
+4915123456789,Jonas,Hamburg

Everything else you know about a person, including their consent record, belongs in your master list rather than in the file you upload.

E.164 in one minute

E.164 is the international numbering format defined by the ITU: a plus sign, the country code, the national number without the trunk prefix, no spaces. +44 7700 900123 becomes +447700900123. It is the only format carriers understand, and the only one that tells us which country's rate and rules apply.

Two terms do all the work. The country code has one to three digits: 1 for the United States and Canada, 44 for the United Kingdom, 234 for Nigeria. The trunk prefix is what callers dial before a national number from inside the country: 0 in most countries, 1 in the United States and Canada, 8 in Russia, and nothing at all in a few, such as Spain. It belongs to domestic dialling, not to the number, so it disappears in E.164. A complete number never has more than 15 digits after the plus. The glossary has a short entry on E.164.

Worked examples by country

The method is the same everywhere: strip the formatting, drop the trunk prefix, add the plus and the country code.

CountryWritten locallyCountry codeE.164
United Kingdom07400 12345644+447400123456
France06 12 34 56 7833+33612345678
Germany01512 345678949+4915123456789
Spain612 34 56 7834+34612345678
United States(201) 555-01231+12015550123
Brazil(11) 91234-567855+5511912345678
Nigeria0803 123 4567234+2348031234567
India98765 4321091+919876543210

A few cases need more than the rule:

  • United States and Canada. A number written 1-201-555-0123 already includes the country code; do not add a second 1. A bare ten-digit number needs the 1 added.
  • Brazil. Numbers dialled long-distance within the country can carry a 0 and a two-digit carrier code before the area code. Both are dialling prefixes, not part of the number.
  • India. Mobile numbers are usually written as ten digits; a leading 0, when present, is the trunk prefix.
  • Italy. The exception that breaks blanket rules: landline numbers keep their leading 0 in international format (+39 06 for Rome), because there the 0 is part of the number. Italian mobile numbers start with 3 and have no 0 to drop.
  • 00 and 011. These are international access codes: 00 from most countries, 011 from the United States and Canada. 0044 7400 123456 and 011 44 7400 123456 are both +447400123456.

This is why a spreadsheet formula that strips every leading zero is a bad idea, and why the importer's default country is the better tool: it applies the rule of the country you name.

Spreadsheet traps

Many broken numbers are broken by a spreadsheet before they ever reach the importer. A spreadsheet that reads a phone number as a quantity does three things to it:

  • It drops the leading 0: 07400123456 becomes 7400123456.
  • It drops the plus sign: +447400123456 becomes 447400123456.
  • It may display long numbers in scientific notation, such as 4.47E+11, and a CSV saved from that view can keep only the digits shown on screen. Those digits are then gone from the file for good.

The fix is to format the phone column as text before typing or pasting into it, or to import the source data as text in the first place. When you save a CSV, choose UTF-8, so that names such as Zoë, Łukasz or Siobhán survive intact for personalisation. If a file arrives already damaged, go back to the source export rather than repairing digits by guesswork.

What normalisation does

Numbers in local format are converted when the country is known: set a default country for the file, or add a country column. Spaces, dots, dashes and parentheses are stripped. A leading 00 is treated as the plus sign. Numbers that still cannot be resolved, too short, too long, unknown prefix, are excluded and listed, never sent, never billed.

The default country applies only to numbers in local format. A number that already starts with a plus or with 00 carries its own country code, so +33 6 12 34 56 78 stays French in a file whose default country is the United Kingdom. A file that mixes countries in local format is the one case a single default cannot solve: add the country column, or convert the foreign numbers at the source.

Deduplication

The same number written two ways is one recipient. After normalisation, duplicates are merged and the first row's variables are kept. The import report says how many were merged so you can check the source when the count surprises you.

With the United Kingdom as default country, 07400 123456, +44 7400 123456 and 0044 7400 123456 all become +447400123456: one recipient, one message, one charge. Three habits follow:

  • Sort before you export. Because the first row wins, put the row with the best data first, usually the most recently updated record.
  • Merge lists before you import them. Two campaigns to two overlapping lists send two messages to everyone on both; one file built from both lists sends one.
  • Read a high merge count as a symptom. A form that stores every submission, or a shop export with one row per order instead of one per customer, inflates the file.

Deduplication works on numbers, not on people. A customer with a personal and a work phone receives two messages unless you choose one at the source.

A campaign file needs numbers and variables; the proof that each person agreed belongs in the master list it is exported from, next to every number. The question "when did this person agree, and to what?" always arrives one number at a time, and these columns answer it:

ColumnWhat it holdsExample
consent_sourceWhere the number was collectedCheckout form, in-store sign-up
consent_atDate and time of the opt-inAn ISO 8601 timestamp
consent_scopeWhat the person agreed toMarketing SMS, or service messages only
consent_wordingThe wording or version of the form they sawsignup-form-v3
opted_out_atWhen they left, if they didEmpty for active subscribers

Before each campaign, export only the rows whose scope covers the message, leave out everyone with an opt-out date, and drop the consent columns from the upload itself. What counts as proof differs between countries; the guide to marketing SMS rules covers the common core. Harvested lists are refused under the acceptable use policy, and bought lists fail the consent test almost everywhere.

Landlines and other dead ends

Most fixed lines cannot receive SMS, and a correctly formatted landline is still a landline. Where mobile and fixed numbers use different prefixes, filter at the source: UK mobile numbers start with 07, French mobiles with 06 or 07, German mobiles with 015, 016 or 017. In the United States and Canada, mobile and fixed numbers share area codes and cannot be told apart from the number alone, so the safest North American list is one collected as mobile numbers in the first place. A message to a fixed line that reaches a carrier and fails is billed like any other attempt, which makes the source the cheaper place to remove them.

The same goes for numbers that were never personal: switchboards, fax lines, a shared reception desk. A generic phone field in a CRM holds whichever number the customer typed; if you also hold a field labelled mobile, use that one.

The import report

Rows read, rows ready, invalid rows with the reason, duplicates merged, destinations found with a count per country and the cost split at current rates. Read it before composing: a list that turns out to be 40% Brazil needs a numeric sender strategy that a European list does not, as the sender ID rules explain.

Three questions make the reading quick. Does the gap between rows read and rows ready make sense? Are the invalid reasons ones you recognise, such as a column of numbers that lost their zeros, or something new? And are the countries the ones you collected from? A country you did not expect, or too few numbers in one you did, usually means numbers were written without their country code and read as local.

Keeping a list healthy

A list decays on its own: people change numbers, operators reassign numbers that have been disconnected, and consent given long ago fades from memory. Keeping it healthy is a routine, not a project:

  • After each campaign, remove numbers reported as unknown. The report has a one-click action for it.
  • Keep absent subscribers for a later send, but drop numbers that are absent campaign after campaign.
  • Keep every opt-out in a suppression list of your own and strip those numbers from every file before import, whichever tool or team sends next.
  • Store the consent source with each number.
  • Re-check a dormant list with a small send to a random sample before a large campaign; the unknown share in the sample tells you what the full send would waste.
  • Give people a way to update their number, such as a field in their account, so the list follows them.

Per-message records are kept for 90 days and campaign summaries indefinitely, long enough to compare two campaigns to the same list and see whether it is ageing: a rising share of unknown numbers is the clearest sign. Export per-message results if you want a longer history. The guide to why SMS fail explains each failure reason, and your first campaign, step by step puts the list to work.

Put it into practice.An email address opens the account. Rates are public, payment is crypto, and the test key is free.
Start sending

Something unclear or out of date? Tell us on the support page; corrections are folded back into the guide and the update date changes.