How to Clean a Cold Email List Before Sending

Start with the format, not the mailbox

Most list problems begin before verification. Duplicate rows, trailing spaces, mixed case, "Name <email>" strings, multiple addresses in one cell and spreadsheet columns shifted by a comma all produce avoidable failures. Normalise every address to lowercase, trim whitespace, split combined fields and remove rows with more than one address.

De-duplicate on the address, not the name. The same person can appear twice with different capitalisation or a different company name, and both copies will receive the campaign. Remove the duplicates before verification so you are not paying to check the same mailbox twice.

Verify in the right order

Syntax checking is instant and catches typos. Domain and MX checks confirm the domain can receive mail. Mailbox verification goes further by asking the receiving server whether the address exists, which is more expensive and less certain because some servers refuse to answer.

Run the cheap checks first, then mailbox verification, then review the uncertain results. A verification tool will usually return valid, invalid, risky or unknown. Treat unknown as a decision rather than a pass: catch-all domains, greylisting and servers that hide answers all produce unknown results.

Use list cleaning and verification on the import, and read role addresses and catch-all domains before deciding what to do with info@ and sales@ contacts.

Deal with role and disposable addresses deliberately

Role addresses such as info@, support@ and sales@ reach a shared inbox, not a person. They are sometimes useful for vendor research and rarely useful for a personal cold email. Keep them in a separate segment and contact them only when the offer genuinely belongs to that team.

Disposable domains exist to be abandoned. Remove them, along with obviously fake domains and addresses that belong to your own company or your competitors' test domains. A suppression list from previous campaigns is the last filter, and it should include hard bounces, complaints and every opt-out.

Segment before the first send

Clean lists still fail when relevance is ignored. Split by industry, company size, job title or the trigger that made you contact them, then write a first line that matches the segment. A verified list of 800 relevant contacts will outperform a verified list of 8,000 contacts who cannot use the offer.

Mail Pro supports unlimited lists and contacts, so segmentation costs nothing except the discipline to maintain it. Import the clean segment, keep the original source and verification report, and note when it was last verified.

A cleaning routine you can repeat

1. Export the raw list and keep a copy.
2. Normalise addresses and remove duplicates.
3. Reject invalid syntax and malformed domains.
4. Run MX and mailbox verification.
5. Remove invalid and disposable addresses.
6. Move role addresses into their own segment.
7. Apply the account-wide suppression list.
8. Assign each remaining contact to one relevant segment.
9. Record the verification date and source.
10. Re-verify before reusing the list later.

Worked example: an old conference list

A team has 6,000 addresses from a conference three years ago. They de-duplicate to 5,400, verification removes a further share, and role addresses move to a small vendor segment. The remaining contacts are split by company size and the offer is written twice. The first send goes to the smallest segment, bounces stay low and replies arrive. Only then does the rest of the list go out, in batches.

Without the cleaning pass they would have sent one large, partly dead list from a warm domain and spent the following month repairing the reputation. The cleaning took a day and protected the domain.

Cleaning is maintenance, not a one-off

Set a re-verification date for every list. Addresses change jobs, companies close mailboxes and domains lapse. A list verified last year is a list of guesses this year. The suppression list is the only part that should never need re-checking, because it records what already went wrong.