Your cold emails get opened and nobody replies. It is the list, not the inbox.
Your cold emails are getting opened. Half of them, sometimes more. And the reply count is four, maybe five, none of them going anywhere.
If that is you, your first instinct was probably to blame the inbox. Deliverability. Spam filters. Subject lines. Maybe you have already rewritten the copy twice. I have been inside exactly this problem with a B2B services firm I work with, and the inbox was not the issue. The list was.
Opens tell you about the inbox, replies tell you about the list
Here is what the campaign looked like. Over two weeks it contacted roughly 1,900 people. About half of them opened. Four real humans replied. Zero meetings. And 6.6% of the emails bounced.
Everyone's first guess was deliverability. It is always the first guess, because it feels technical and comfortably out of your hands. But an open rate near half tells you the emails were landing. People were seeing them. The inbox was doing its job.
The bounce rate was a real problem, just a small and mechanical one. I will come back to it.
When I diagnose a campaign, I work in a fixed order:
- Inbox: opens and bounces.
- List: persona, qualifier, timing, overlap.
- Offer.
- Copy.
Each step only gets attention once the one before it is clean. Copy is the last thing I touch. Most founders start there, because rewriting an email feels like progress. With opens near half and replies near zero, rewriting copy is polishing a letter that is addressed to the wrong person.
The list had the right companies and the wrong doors
The list had been built as "C-level in operations and finance". Sensible sounding. Also wrong.
When I read the firm's own case studies, the people who had actually bought in those industries were the CIO, the Head of Digital or the Head of IT. Same companies. Different door.
It was a reasonable guess about who buys. That is the problem: it was a guess, and nobody had checked it against the firm's own history.
This is why the opens were so misleading. The readers were real. They were just the wrong readers. Operations and finance people can open all day. In this firm's history, they were not the ones who signed. An open from the wrong person looks identical in your dashboard to an open from the right one.
The filter would have rejected their best customers
The second problem was the qualifier. The list filter asked one question: "is this a software company?"
Tidy question. It would also have rejected several of the firm's best existing customers, including a bank and a manufacturer.
The better question was: "does this company build or run a significant software surface?" A bank's mobile app qualifies. A manufacturer's dealer portal qualifies. A retailer's e-commerce site qualifies. The first filter asks what a company is. The second asks what it runs, and what it runs is what predicts the need.
Where do you find the right question? Not in a brainstorm. The firm's own case studies turned out to be the best ICP document it had. They show which industries bought, which job titles signed and what triggered the project. Nobody had read them with that question in mind.
Your closed deals are a list-building tool. Most firms file them under marketing and never read them again.
Hiring posts are a free signal, and a noisy one
The third fix was timing. Companies that are currently hiring for the role the service supports are in-market. A job post is a company telling the market what it needs, in public, for free.
But raw signal is noisy. When we cleaned the raw hiring-signal pool, only about 30% of the companies in it were true fits, once direct competitors and large captive centres were removed.
Read that both ways. The signal was good enough to build the list on. It was also dirty enough that most of the raw pool, roughly two companies in three, would have been the wrong send. Most teams either skip timing or treat the raw pool as the finished list.
A third of the new list was not new
Last two checks, and both are unglamorous.
Suppression. When the fresh list was checked against everyone already emailed in older campaigns, 36% overlapped. Without that check, a third of the "new" campaign would have been repeat sends to people who had already ignored it once. Sending again to someone who ignored you is not persistence. It is noise with your name on it.
Verification. The 6.6% bounce rate came from skipping verification. Every address now gets verified before it enters a sequence. It is boring and mechanical, and it was the only inbox-side fix this campaign actually needed.
Now the honest part. The rebuilt campaign has not run long enough for me to report results, and I am not going to pretend otherwise. This article is about how to find the problem. That is the part most teams skip, and it was the part that was broken here.
What you can run this week
If your opens are healthy and your replies are near zero, run this before you touch the copy:
- Pull your last campaign's numbers: opens, bounces, replies. Decide which problem you actually have.
- Read your own case studies or closed deals. Write down the job titles that signed. Compare them with who is on your list.
- Test your qualifier against your best existing customers. If the filter would have rejected them, the filter is wrong.
- Add one timing signal, like relevant job posts, and clean it before you trust it.
- Suppress everyone you have already emailed in older campaigns.
- Verify every address before it enters a sequence.
That is the whole method. If you would rather have someone run it for you, Outbound Engine is the program where I do this for clients.
Opens are the inbox talking. Replies are the list talking. Learn to tell them apart before you rewrite a single word.