Cold Email Not Working? A 30-Point Audit to Find the Cause

When cold email has never booked a single call, the cause is rarely the subject line. In a campaign that never worked, something was usually broken before the first email left the outbox: no clear entry offer, no proof a stranger can check, a list of the wrong people, or a sending domain the inbox does not trust.

This audit finds which one. It is 30 yes-or-no checks in 6 layers: foundations and proof, list and targeting, offer and copy, sender and deliverability, follow-up, and test size. Every "no" is a red flag. The layer with the biggest share of red flags is where you start.

Give it 20 minutes with 3 tabs open: your sending tool, your LinkedIn profile, your website.

One rule sits under the whole list. If a campaign gets nothing back from a few hundred of the right people, sending it to 20,000 will not change the answer. It only takes longer to hear it.

Did it ever work, even for one week?

Ask this before anything else, because it splits the diagnosis in two.

A campaign that produced replies and calls for a while and then went quiet has a decay problem. The list ran out, the domains lost reputation, or the market got tired of you. That has its own playbook: check list freshness, domain reputation and message fatigue, in that order, before you touch anything else.

A campaign that never produced anything worth a calendar slot is different. Nothing decayed, because nothing was ever there. That is a foundations problem far more often than a sending problem, and this checklist is built for it.

Your reply pattern already points at the likely layer:

Still run all 30. The table only tells you where to look first.

How do you score it?

Answer every check with yes or no. "Sort of" is a no. "We are working on it" is a no.

Count the no answers per layer and write the number next to the layer name. At the end you will have 6 numbers. The biggest one, relative to the size of the layer, is your problem. If two layers are close, the higher one comes first.

Be harder on yourself than feels fair. Nobody else will score this for you, and a generous self-audit just sends you back to rewriting subject lines.

Layer 1: Is there anything worth replying to?

Layer 1 is about what a prospect sees when they look you up, and they do look. The buying decision is made online, before any call is booked, on your profile, your website and your proof. If that check fails, the reply never comes, and you never find out why.

The order below is the foundations sequence I use before any list gets built: profile, one entry offer, a video sales letter, written proof with real numbers in it, a home page that matches, and only then the list and the outreach. It is the order I have ended up trusting over 12 years of building outbound for B2B companies.

1. The sender's LinkedIn profile says which buyer you serve and which problem you remove. The headline and the first lines of the About section say it plainly. A list of past job titles does not count.
2. You sell one entry offer. One clear first thing a buyer can say yes to, instead of a menu of 7 services.
3. A video sales letter exists and is easy to find. In 5 to 10 minutes you explain the problem and how you solve it. It sits in the featured section of the profile or on the home page.
4. At least 2 case studies show a measured result and how long it took. "Great partner, highly recommend" is a testimonial. A case study says what changed and how fast.
5. The first screen of your home page repeats what the profile promises. Same offer, same buyer, same words.
6. A stranger could explain what makes you different after 2 minutes of looking. Different from the next vendor in their inbox, and in words they would use themselves.

Red flags here explain the most painful pattern in outbound: the email gets read, the prospect clicks through, finds nothing that earns trust, and leaves. A sender can push 5,000 "personalized" messages a week into that gap and hear nothing back.

It shows up in reply rates too. The same wording got 2% replies from one sender and 7.5% from another, and nobody touched the copy. The difference was what sat behind the sender.

Layer 2: Are you writing to people who can buy?

A rewritten email cannot save a list full of people who will never buy, and it cannot save a sender nobody trusts either. That is why this layer comes before the copy.

1. Your ideal customer fits in one sentence. It names the industry, the headcount range, the countries, and the title of whoever owns the problem.
2. Every address was verified before sending, and bounces stay under 2%. Instantly, which sells cold email sending software, sets the cold email target below 2% in a guide updated September 2026 and warns that anything over 5% can damage your sending reputation. That is vendor guidance, so treat 2% as a ceiling to stay well away from.
3. The list is built on named people. Role inboxes like info@ or sales@ are not your main contacts.
4. The list was built for this offer and this campaign. Not bought in bulk, not inherited from an old campaign, not pulled so long ago that half the people have changed jobs.
5. The person you email can sign, or can walk the deal to whoever does. An intern cannot move the deal. Neither can the founder of a company too small to afford you.

If checks 8 to 10 failed, stop here and fix the data first. I go through that line by line, filter by filter, in how my team takes a B2B lead list apart.

Layer 3: Does the email ask for something small and clear?

When a campaign goes silent, the first thing most founders do is rewrite the copy. Sometimes that is right. Usually it is the third or fourth thing that needed fixing, which is why this layer sits in the middle.

1. The first email sells a conversation, not your whole service. It gives one reason to talk. The full offer can wait for the call.
2. There is one ask, and it is easy to say yes to. One question or one short call. Not a demo, a proposal and a calendar link in the same email.
3. The first 2 lines name a problem the reader would recognize in their own words. If you swapped in a different company name, the email should stop making sense.
4. It reads on a phone without scrolling. Send it to yourself and open it on your own phone. If you have to scroll, it is too long.
5. One piece of proof is in the email or one click away. A number from a case study, a named result, a short video. Something checkable.
6. Read out loud, it sounds like one person writing to another. No merge fields showing, no template phrases, nothing a prospect has already seen 40 times this week.

A campaign can pass layers 1 and 2 and still fail here. That is the only case where rewriting the email is the main fix.

Layer 4: Does it land in the inbox at all?

If layer 4 fails, every other number in your tool is fiction. A 0% reply rate from emails that went straight to spam tells you nothing about your offer.

Google and Microsoft publish rules that matter here. Checked September 2026:

Sources: the Gmail requirements page in Google's admin help and Microsoft's April 2025 post on Outlook authentication.

Both sets of rules are written for personal inboxes, and your B2B prospects mostly read mail on company domains. Most small cold campaigns also sit well under 5,000 emails a day, so it is tempting to skip this. Don't. Microsoft's own FAQ on that page says senders below the threshold also benefit from the same authentication, and Google asks every sender for SPF or DKIM regardless of volume.

1. You are not sending cold email from your main company domain. The domain your clients and invoices use stays out of it.
2. SPF, DKIM and DMARC all pass. Send a test to a Gmail address, open "Show original" and read the 3 results. Each one should say pass.
3. New domains and inboxes were warmed before real volume. Every new inbox I set up gets a 10 to 14 day warm-up. Sending hard from a domain registered last week is a red flag.
4. Your spam rate is under Google's 0.1% target. Check it in Google Postmaster Tools. A small campaign may find the dashboard empty, because Google leaves out days with too few messages. Empty means unknown, so keep checking as volume grows.
5. Your sending domains are not on a major blocklist. A free blocklist lookup takes a minute.
6. Daily volume per inbox is modest and steady. No bursts of hundreds from one mailbox on launch day.

The setup itself (domains, inboxes, DNS records) is covered in the boring part of cold email that decides whether anyone sees it. Provider rules change, so check both pages above before you rely on them. This is not legal advice.

Layer 5: Did you follow up enough to know?

A lot of campaigns that "never worked" were never finished. Most prospects saw 1 email, maybe 2, and the founder called it.

1. Every prospect received the full sequence before you judged the result. A first email plus follow-ups, with real days between them. My own cold email sequences have 4 steps. Replies tend to come late, most often on the third follow-up. Stopping after 1 or 2 messages leaves the test unfinished.
2. Each follow-up adds something new. A proof point, a different angle, a short question. "Just bumping this" adds nothing.
3. Someone answers interested replies the same working day. The prospect was interested when they wrote, and that interest does not sit and wait for you.
4. Replies are sorted into categories. Positive, not now, wrong person, not interested, remove me. Each category points to a different layer of this audit.

Founders rarely bother with check 27, yet it makes the next audit fast. "Wrong person" in bulk points at layer 2. "Not interested" in bulk points at layer 3. Silence points at layer 4 or layer 1.

Layer 6: Was the test big enough to mean anything?

The opposite mistake from volume-first is quitting on a sample too small to read.

1. Enough of the right people were contacted to expect a handful of replies. Take a normal reply rate, multiply it by the people you actually reached, and see whether "zero" was even a surprise. I keep the reply rates normal campaigns actually get in a separate roundup.
2. You tested one offer on one segment at a time. Sending 5 offers to 5 markets at 60 people each gives you 5 results that mean nothing.
3. The campaign ran long enough for most of the list to finish the sequence. Judging in week one judges the first email only.

This layer cuts both ways. It protects you from killing a campaign too early. It also stops you from treating volume as the cure. If 400 of the right people got the full sequence and nobody replied, the answer is in layers 1 to 4. Doubling the list only doubles the number of people who saw a version of you that was not ready.

What does your score tell you to fix first?

Start at the layer with the biggest share of red flags, with one exception.

The exception is authentication. If check 19 failed, fix it today. It is usually an hour of DNS work, and until it passes, every other number in this audit is unreliable.

After that, go to your heaviest layer. If two layers are close, fix the higher one first, because a problem higher up can hide whether the fixes below it worked.

Fix the funnel first. Then scale. I have said it before and I will keep saying it, because the pull to fix a quiet campaign with more sending is strong, and it almost never pays. Once the funnel converts, adding mailboxes without breaking what works is a separate job with its own checks.

So you know who is talking: Smirnov Consulting Group is a Prague-based B2B outbound lead generation agency that runs cold email and LinkedIn campaigns for founder-led B2B companies and books qualified sales calls. Outreach is the last thing we switch on for new clients, after the foundations and the list.

What does a filled-in score sheet look like?

Here is how to read one. The numbers below illustrate the logic. They do not come from a client.

• Layer 1, foundations: 4 of 6 red
• Layer 2, list: 2 of 5 red
• Layer 3, offer and copy: 3 of 6 red
• Layer 4, deliverability: 0 of 6 red
• Layer 5, follow-up: 1 of 4 red
• Layer 6, test size: 1 of 3 red

That is 11 red flags out of 30. Deliverability is clean, so the emails were seen. Layer 1 is the heaviest by share, so the people who looked found nothing to trust. Layer 3 is close behind, but a better email sent from the same empty profile would meet the same silence.

The tempting move is to rewrite the sequence and double the volume. The right move, in this order:

1. Rebuild the profile around one entry offer.
2. Record the video sales letter and write up 2 case studies with numbers and timeframes.
3. Make the home page say what the profile says.
4. Rebuild the list for that one offer and verify it.
5. Rewrite the first email so it asks for a conversation and carries one proof point.
6. Relaunch to a few hundred of the right people, run the full sequence, and score the audit again before adding volume.

That takes weeks. Most founders skip it for exactly that reason, and then run the same broken campaign at twice the size.