CSV Formatting Mistakes That Break Cold Campaigns

This is the least interesting article in this library and probably the one that will save you the most embarrassment. Spam filters get the attention, but a large share of cold emails that fail do so before deliverability is involved at all: they arrive perfectly, in the inbox, reading Hi , — and the recipient learns everything they need to know about how the message was produced.

A broken merge field is worse than no personalisation. A message that opens Hi there, is neutral. A message that opens Hi , is a confession.

Whitespace is the main offender

The single most common cause of a failed merge is a space you cannot see.

Spreadsheet exports routinely carry trailing whitespace in header cells, so a column that looks like First Name is actually First Name . Merge fields match exactly, the lookup fails, and the field resolves to empty for every row in the file. Nothing in the spreadsheet indicates a problem, and the template looks correct, because the template is correct.

The same applies to values. A contact whose first name is Sarah will merge as Sarah — usually harmless, occasionally not, and always a sign the file was not cleaned. Non-breaking spaces, which arrive when data is pasted from a web page, are worse: they look identical to a normal space and match nothing.

Trim every header and every value on import. If your spreadsheet tool has a TRIM function, run it across the whole sheet before exporting; it costs one step and removes an entire category of failure.

Names are data, not text

The second most common problem is treating a name column as though it contains what its label says.

A first_name column populated from a data provider frequently contains full names, and the resulting greeting — Hi Sarah Chen, — is subtly wrong in a way that reads as automated, because no human writes that. All-caps values from CRM exports produce Hi SARAH,. Titles produce Hi Dr. Chen, in a message otherwise written to be informal.

Company names have the same problem in a more visible place. Providers return legal entity names, so a sentence written for Acme becomes I noticed Acme Corporation Private Limited is hiring. Keep two columns if you need the legal name: one for records, one for use in sentences.

The general rule is that any value you drop into the middle of a sentence must already read like something a person would type there. If it needs cleaning to read naturally, clean it in the file, not in the copy.

Encoding, and why it fails late

File encoding is the mistake most likely to reach real recipients, because it usually survives testing.

Default spreadsheet exports use a regional encoding rather than UTF-8. Plain ASCII names pass through unharmed, which is why a test row called John Smith proves nothing. Names containing accented or non-Latin characters — José, Müller, Łukasz, any name in a non-Latin script — are mangled into replacement characters that arrive in the delivered message.

Save as CSV UTF-8 explicitly, every time. Then include a name with a non-ASCII character in your test send, so that the check actually exercises the failure mode.

Smart quotes and em dashes belong in the same category. Copied from a document, they render inconsistently across mail clients and occasionally as literal escape sequences. If any part of your file passed through a word processor, paste it as plain text before saving.

Decide about blanks before you import

Missing values are inevitable in any list of real size. What matters is that you decide what happens to them deliberately.

There are three defensible options. Exclude contacts missing a field the campaign depends on, and send them a different campaign. Write the sentence so it reads correctly when the value is empty — which usually means moving the merge field out of the opening line. Or use a fallback, accepting that Hi there, is widely recognised as a fallback and buys you very little.

What is not defensible is finding out at send time. A blank cell in a {{company}} field does not produce a sentence with a gap; it produces a sentence that is grammatically broken, and it goes to everyone in the file who is missing that value.

Always read the rendered message

Every one of these failures is caught by the same thirty-second check: send the campaign to yourself, with a handful of real rows including at least one awkward one, and read the message that arrives.

Not the template. The rendered message, in a real inbox, as the recipient will see it. Every problem described in this article is obvious at that moment and invisible before it.

RepMail validates placeholders at generation and again at send: a template containing a merge field the list cannot fill is blocked rather than delivered with a gap, so Hi , does not reach a recipient. Uploads are checked per address with duplicate detection inside the file, and the campaign confirmation screen shows the effective From address and the rendered result before anything is sent. The file hygiene above is still worth doing — the platform can stop a broken merge, but it cannot tell that SARAH should have been Sarah.