How to clean a prospect list before the first email goes out

Clean a prospect list by reading its rows against each company's own website before any of them is sent: check that the name belongs to the company you mean, merge branches of one brand into a single row, and set aside any row whose research says nothing. Then take out everyone on your do-not-contact list, by person and by company, and re-read each row on the day it goes out.

The count hides the errors

A list arrives with a number on it, and the number is what gets checked. One sales automation tool emails new users when their first list is ready, with five of the leads previewed to show what the list holds. Two of those emails, read closely, carried two different kinds of error in their five rows.

In a list of actors and talent, one actor ran a company under their own surname. The tool had matched that name to a fintech of the same name, so the description under a working actor was about digital banking, and it ended in an internal placeholder line saying no further details were available. In a list of more than ten thousand leads at recruiting firms, three of the five previewed leads were owners of franchises of the same staffing brand in three different towns, with near-identical descriptions.

Those were the five rows picked to show the list at its best, and the headline number gave no sign of either problem. Both surface only after the mail has gone.

Customers of another prospecting tool describe the same failures in its user community: people matched to the wrong profile, the wrong companies targeted, leads paid for that turned out to be duplicates or impossible to contact, and contacts who had left their company by the time the campaign reached them.

The list is what the cold email sequence calls the biggest difference between a sequence that works and one that does not. These are the checks that tell you whether yours deserves the copy you are about to write for it.

Four questions for every row

Open the company's own site, and ask of the row:

• Is this the company I mean? A company named after its founder, or one that shares its name with another business, is where a name match goes wrong, as it did for the actor. Check that what the site says they sell matches the line in your list.
• Is this one company or several? Franchises, regional offices and subsidiaries of one brand belong in one row. Decide who at the brand you want to reach, and write to them once.
• Does the research say anything? A placeholder, a boilerplate description or a line that could sit under any company in the list is an empty cell. The paste test from personalisation that works applies here too: if the sentence would still make sense moved to another row, it tells you nothing about this one.
• Is it still true today? A list ages in hours. One list of fourteen agencies, re-read later on the day it was compiled, already had two rows that were wrong: one agency had turned into a data vendor, and another was trading under a different name on its own domain.

You do not need to read every row to learn what a list is worth. Read a sample first. If the sample turns up any of the four, read the rest before any of it sends.

Let the list come back shorter

The honest thing to do with a row you cannot check is to set it aside. One research service that finds companies by recent events says so in its own words: the number you ask for is a ceiling, and it returns fewer companies when public evidence cannot confirm both the date of the event and the identity of the company.

The same rule works by hand. When twenty outbound agencies were ranked from their own pages, five more could not be read: a domain with nothing on it, a server that failed three times in a row, a page that rendered no text, a parked domain, and one whose site could not be found at all. They went at the bottom as unread, each with its reason, and none of them was ranked on a guess.

That is the same move as the activation column in the only prospects worth emailing this quarter: a row with an empty cell gets watched, and it does not get emailed.

Build the do-not-contact list first

A different kind of row does its damage with people who already know you: the right company at the wrong moment, such as a customer, someone already in a conversation with you, or a person who said no last month. In the same user community, people asked how to upload the contacts they were already talking to so that the tool would leave them alone. One described a tool messaging people again and again until they worried for their professional credibility on their own LinkedIn account. Another had to apologise to someone they were already in a thread with, after an automated sequence reached that person as well.

So the do-not-contact list comes before the prospect list, and it holds:

• your customers, and anyone you are already talking to;
• anyone who has said no, or asked not to hear from you;
• companies as well as people, by web domain, so that a new contact at a customer is caught too.

One sending tool's suppression list carries the detail a simple filter misses: adding someone to it cancels mail already queued to them, as well as anything sent later. If your list only filters the next import, a sequence that started last week keeps going to the person who asked you to stop.

And make sending wait for a person. The activation email that previewed five leads ended by saying that outreach to the whole list would start automatically in two hours unless the user reviewed it first. With that default, every error above reaches a stranger before anyone has read a row.

Five things

1. Read a sample of rows against each company's own site before any of them sends.
2. Merge franchises, offices and subsidiaries of one brand into one row.
3. Set aside the rows you cannot verify, with the reason, and leave them out of the send.
4. Build the do-not-contact list first, by person and by domain, and make it stop mail already queued.
5. Re-read each row on the day it goes out.