Our reply rate went up 317% in a week on the exact same audience. Three changes. Not one of them was rewriting the email.
That is the useful part, so I will start there and then give you the order I work in.
The three changes
We retired the infrastructure instead of repairing it. The inboxes were cooked. We did not try to nurse them back. Fresh domains, fresh inboxes, nine days of warmup, then back to sending at low volume.
That is the part people resist, because new infrastructure costs money and repairing feels free. It is not free. Most people are too cheap here and leave real pipeline on the table because they will not renew inboxes.
We changed the data source, not the filters. Same ideal customer profile, different place to get it from. We used a lookalike filter seeded from our most recent closed won customers, which surfaced companies we had never contacted. Same definition of a good account, a different set of accounts.
We stopped asking for the meeting. The old ask was a meeting. The new one was a report from a client who had compared us against their other vendors. Their data, not ours. We ran three angles off that, including one built around a meme, and let people reply to the thing rather than to the calendar.
Same list. Same product. Same person writing.
Vibes GTM, which is a real method
Before the order, the thing underneath it.
I don't run placement tests. I don't grade every metric on a dashboard. I look at the overall picture instead of over analysing every detail, and I would rather spend that hour on the list.
This sounds sloppy and it is the opposite. Most outbound dashboards tell you something you would have learned from a week of sending anyway, and the time you spend reading them is time the campaign is not improving. Watch the big signal, act on it, move on.
The first 24 hours tell you most of it
One thing before the order itself.
You can usually tell within a day of launching whether a campaign is going to work. Most people miss that window because they are waiting politely for a week of data before they look, and by then they have sent the whole list.
I teach this as the hangover: the morning after launch, when the numbers are ugly and you have to decide whether you are adjusting or starting again. The decision is not which of six possible causes is to blame. It is a much simpler one. Are you tweaking or rebuilding?
Tweak when the list is right and the copy is soft. That is a line level fix and you should expect to iterate two or three times before it lands.
Rebuild when the list is wrong. Copy cannot rescue a list, and every hour spent rewriting for the wrong audience produces a better email that nobody needed.
Go back further than either when you have run three genuinely different angles at a good list on healthy infrastructure and none of them moved. At that point the email is not the problem and neither is the list. What you are offering is not interesting to the people you picked, which is a harder answer and usually the right one.
Knowing which of those three you are in, on day one rather than day eight, is most of the value in this whole page.
The order
1. The copy
Start here, because it is free to check and it is where the problem usually started.
Read the email as the person receiving it. You have about three seconds in the preview pane. If you want a second opinion on it before you start rewriting, the cold email grader will score it and tell you the one thing to fix first.
The mistakes I see constantly, after reviewing thousands of sequences:
- It is written for everyone. "I help B2B companies book more meetings" could go to anyone, which means it lands with no one.
- It opens with your company. "At [Company], we help..." Nobody cares yet. Use those three seconds to say something about them.
- It is four paragraphs. I still see 300 word cold emails. One problem, one outcome, one ask. Everything else is noise.
- There is a link in it. Links hammer deliverability. Great copy that lands in spam is still dead copy. Move it to a follow up. This includes your signature.
- Every follow up is a check in. If each step is "just circling back" with no new angle, you have automated being annoying. New hook, new proof point, new angle each time.
2. The lead list
If the copy is sound, the list is next, and it is the bigger lever of the two.
The honest test of a list is what happens to sends per positive reply. My best week was 738 sends per positive reply. My worst was 6,355. Same product, same person writing the copy. The only variable was how tight the targeting was.
That is an eight times difference from the list alone, which is more than any rewrite has ever given me.
Check that the list was verified before it went out. Bad data means bounces, bounces mean a destroyed sender reputation, and at that point you are not diagnosing a reply rate problem any more, you are diagnosing a deliverability problem you caused.
And resist the instinct to fix a weak list by sending it to more people. In 8 million emails of platform data, tripling send volume produced fewer total replies, not more. More sends is a lazy substitute for better targeting.
There is a longer version of this argument in cold email personalisation: the list does the work.
3. Buy more infrastructure
If the copy is right and the list is right and replies are still flat, the inboxes are the problem. Buy more.
Note the verb. Not repair, not diagnose, not test. The 317% week started with retiring inboxes rather than rescuing them, because the cost of nursing a burned domain is weeks of your campaign being dead while you find out it did not work.
The maths for how many to buy, and the per provider sending limits, are in how many inboxes for cold email.
Here is the rule of thumb worth holding on to: bad copy plus great infrastructure is fixable. Great copy plus bad infrastructure equals burned domains.
While you are in there, check the follow ups
Not a cause of a dead campaign so much as a cap on a working one.
Across the same platform data, the first email plus three follow ups captures almost everything. After that you are not following up, you are annoying people. If your sequence runs to eight steps, the last four are costing you more in reputation than they return in replies.
And answer your replies
The fastest fix on this page and the one most teams ignore.
When a positive reply comes in, answer it within three minutes. Every reply is the highest intent moment in the whole funnel and it decays fast. A campaign that books nothing is sometimes not a campaign problem at all, it is four warm replies that went cold in an afternoon.
The short version
Copy first, because it is free to check. List second, because it is the biggest lever. Infrastructure third, and when it is infrastructure, buy new rather than repair.
Do not run placement tests. Do not build a dashboard. Look at the big picture, fix the thing that is obviously wrong, and send again.