Three numbers that tell you why a campaign stalled
When outreach stops working, most people change everything at once. Three numbers identify which part broke.
By The Coaching World
Published November 6, 2026 · 6 min read
Photo by Brian Wangenheim on Unsplash
Three numbers, in order
When a campaign stops producing, the instinct is to change the message. Usually the message was not the problem, and changing it destroys the only evidence you had.
Three numbers identify which part broke, and they have to be read in order, because a failure in the first makes the other two meaningless.
This article is for anyone whose outreach worked and then did not. It covers the three numbers, what each failure looks like, what to do for each, and the trap of changing several things at once.
Number one: are messages arriving
Delivery or acceptance rate. Of the people you attempted to reach, how many actually received anything.
This is mechanical rather than persuasive, and it is the most common cause of a sudden stall. Accounts get restricted. Domains get flagged. A list of addresses degrades. A platform changes how it handles volume.
If this number dropped, nothing downstream can be interpreted. Your reply rate fell because fewer people saw anything, and rewriting the message would be fixing a part that still works.
The tell is suddenness. Delivery problems usually appear as a sharp fall rather than a gradual decline, and they often coincide with a change in volume or a new tool.
Number two: are the right people receiving them
Of those who received a message, how many were plausibly the right person. The clearest proxy is the proportion of replies saying wrong area, not my remit, or try someone else.
This number degrades slowly and invisibly. Lists age. People change roles. A source that was accurate a year ago now contains a growing share of people who have moved on.
It also degrades when you expand. Running out of prospects in a tight segment leads to widening the criteria, and the widened list performs worse while looking similar in a spreadsheet.
If this is the problem, the fix is list work, and no amount of writing improves it.
Number three: is the message landing
Reply rate, read only after the first two are confirmed healthy.
This is the number everyone looks at first, and it is the one most likely to be a symptom of the other two.
A genuine message problem usually appears as a gradual decline rather than a cliff, and it is often seasonal or saturation-related. The same audience contacted repeatedly responds less each time. A market that has been approached by many similar businesses develops resistance to the approach rather than to you.
If delivery and targeting are intact and replies have fallen, then the message is the work.
Reading them in order
The sequence matters because each number constrains the next.
Check delivery first, and if it is down, fix only that and wait a fortnight before looking at anything else.
Check targeting second. If wrong-person replies have risen, rebuild the list and wait again.
Only then consider the message. By that point you have eliminated the two explanations that are cheaper to fix and more likely to be the cause.
Most people run this in reverse, which is why campaigns get rewritten repeatedly while the actual fault persists.
The trap
Changing several variables at once is the most common and most expensive error.
A stalled campaign feels urgent, so people rebuild the list, rewrite the sequence and change the targeting in the same week. If results improve, nothing has been learned about which change mattered. If they do not, the position is worse than before, because the previous baseline no longer exists.
One change, then enough volume to read the result, then the next. It feels slow and it is the only approach that accumulates knowledge.
What a healthy campaign looks like over time
Worth knowing, because gradual decline is normal rather than a failure.
Any finite audience saturates. Response rates to a given segment fall as you work through it, and that is the channel behaving correctly rather than breaking.
The response is new segments rather than new copy. A message that worked well against an exhausted audience will work again against a fresh one.
Businesses that interpret saturation as a message problem end up with excellent copy and no audience left to send it to.
Recording the baseline before you need it
The diagnosis only works if you know what normal looked like, and most people discover this at the moment they need the comparison.
Once a week, write down the three numbers. Five minutes. Delivery, wrong-person rate, reply rate, with the date and the volume sent.
A year of those lines turns a vague sense that things got worse into a specific week where something changed, and specific weeks can usually be matched to something that happened. A new tool, a change in volume, a different list source.
Without the baseline, every stall produces the same conversation, which is a debate about whether the message was better last time based on nobody's recollection.
Add a fourth column for anything you changed that week. That column is what makes the record diagnostic rather than merely descriptive.
When all three look fine
Occasionally the numbers are healthy and nothing is closing, which points further down the process.
That is not a campaign problem. Messages are arriving, reaching the right people, and producing replies. What follows the reply is where the loss sits, and the diagnosis moves to the conversation and the offer.
Worth separating clearly, because the response is different. A campaign problem is fixed with lists and copy. A conversion problem is fixed in how meetings are run and what is being sold, and no amount of outreach work touches it.
The sequence for that is its own exercise, but it starts the same way: establish which stage is losing people before changing anything.
How much data is enough
The numbers mislead badly at small volumes, and this causes more bad decisions than any other factor.
A reply rate calculated on thirty messages can swing by half on two replies. Treating that swing as a signal leads to changes that are responding to noise, and the next thirty messages produce a different random answer.
A few hundred messages is the point at which a reply rate starts meaning something. Below that, watch delivery and wrong-person replies, which stabilise faster, and hold off on judging the message.
The practical rule is to let a change run until the sample is large enough to distinguish it from chance. For most small campaigns that is a fortnight or more, which feels unbearably slow when results are poor and is the only way to learn anything.
Our articles are researched and reviewed by our editors. Read our editorial standards.
Related articles
Why your outreach starts to feel spammy, and what changes it
The discomfort is usually accurate. It comes from specific properties of the message, and each of them is fixable.
The question worth asking after every sales call
Most call reviews ask what went well. One better question turns every conversation into usable information about your offer.
When a coaching client needs a therapist, and what ICF says to do about it
ICF publishes a free 24 page guide on referring coaching clients to therapy. What it says to watch for, how to raise it, and what to do when a client says no.