How do I set up hundreds of email addresses at once?

Before you build the file

An import is fast, which is exactly why the preparation matters. Four hundred addresses created correctly in one minute is excellent; four hundred created with the wrong destinations in one minute is a much bigger problem than doing it by hand would have been.

1. 1

   Decide the address format first, and write it down

   firstname.lastname@, f.lastname@, firstname@ — pick one and apply it everywhere. Changing the convention later means every person tells every contact a new address, which is the one part of this that cannot be automated.

2. 2

   Agree what happens to duplicates

   Two people named Sam Jones is not an edge case, it is a certainty above about a hundred people. Decide the rule now — a middle initial, a number, the department — rather than during the import.

3. 3

   Confirm the destinations are current

   A list exported from a directory six months ago contains people who have left and addresses that have changed. This is the most common source of failed imports, and it is fixed by asking, not by importing.

4. 4

   Check the domain is verified for sending first

   Import into a domain that is not yet verified and every address is created but nothing works, which looks like the import failed when it did not.

Building the spreadsheet

Any spreadsheet program will do. Save as CSV, which is the plain-text format every importer reads.

• One row per address-and-destination pair. For an address reaching three people, use three rows with the same address. It reads clearly and it checks easily.
• A header row with the exact column names the importer expects. Header names are matched literally; Address and address may not be the same thing.
• No blank rows and no notes at the bottom. A note under the data becomes a row with an empty address, and an empty address is an error to investigate.
• No formatting. Colour, bold and merged cells do not survive CSV. If you have used colour to mean something, add a column that says it in words.
• Watch for autocorrect. Spreadsheet programs turn addresses into links and sometimes alter characters. Check a few rows in a plain text editor before importing.

The dry run, which is the actual work

A dry run validates every row and reports what it would do, without creating anything. Run it, fix what it reports, and run it again until it is clean. This is not a preliminary step before the real one — it is the step, and the import that follows is the trivial part.

The part that does not scale: people

Creating addresses is instant, and so is delivery: our own mail server forwards to a destination the moment its address exists, and nobody has to click anything first. What still takes time at four hundred rows is the people on the other end — the first message they receive from an address they did not know existed, and the strict filters some of their inboxes sit behind.

Plan it as a communications exercise rather than a technical one.

1. 1

   Tell everyone before you import

   One message: you are getting an address on our domain, mail to it will land in your usual inbox, here is what the first test message will look like. This single step prevents most of the confusion that follows.

2. 2

   Import in batches you can check

   Fifty at a time is manageable; four hundred at once produces a list nobody can work through if something is off. Batches also mean a mistake affects fifty rows, not four hundred.

3. 3

   Do the difficult inboxes first

   Strict corporate filters and shared inboxes are where a forwarded message is most likely to land in spam. Finding that out on day one is much better than on the day you expected to finish.

4. 4

   Check and re-sync, do not re-import

   An address that is not receiving yet is waiting on the domain's DNS, not on a person. Re-importing creates nothing new and tells you nothing; Check and re-sync rewrites the route and says what it found.

5. 5

   Decide in advance what happens to leavers

   Some of the people in the file will leave. Decide now whether their address is removed, pointed at a manager, or switched off — before the question arrives one at a time.

Checking the result

• Count the rows against the addresses created. They should match exactly. A difference means rows were skipped, and the report says which.
• Spot-check ten at random. Not the first ten — the first ten are the ones you already looked at while building the file.
• Send a real message to three of them from outside. Different people, different destination providers.
• Review the destination column once more, sorted. It is the last chance to catch a valid address belonging to the wrong person before mail starts flowing to it.

An import is only as good as the spreadsheet. The minute the import takes is never the risk; the hour spent reading the destination column is what makes it safe.

The half nobody plans: taking addresses away

Every organisation that can create four hundred addresses in a minute discovers, a year later, that it removes them one at a time. The result is predictable and universal: a domain carrying addresses for people who left, departments that merged and products that were retired, none of which anybody dares delete because nobody can say what still uses them.

The fix is to treat removal as part of the same exercise rather than as a tidy-up later. Two habits do most of the work.

• Record why each address exists, at the moment you create it. A column in the import file naming the owner or the reason costs nothing now and is the only thing that makes a review possible in two years.
• Disable before you delete. A disabled address stops delivering but still exists, so if somebody notices, you re-enable it in seconds. Delete only after a quiet month — at which point nobody is guessing.