Reading the Numbers and Diagnosing the Funnel
What it does: tells you which stage of the funnel is actually broken, so the fix in SOP 18 lands on the right thing.
Run it when: weekly, and always before changing anything in a campaign.
Time: 15 minutes.
Depends on: a campaign that has been running long enough to have data. See step 2.
Before you start
At least 50 sends per message variant. Below that there is nothing to read
The campaign's own stats and funnel view open
Last week's numbers, so you can see direction as well as level
The steps
1. Read the funnel in order, not the total.
Sent, accepted, replied, booked. The number that matters is the first stage where the drop-off is worse than it should be. A booking problem three stages downstream is usually an acceptance problem wearing a disguise.
2. Do not judge anything before 50 sends per message.
Rewriting copy after fifteen sends and silence is noise-chasing. The funnel is a lagging indicator: a lead has to see the request, accept, receive the message and then reply, and each of those takes days. Early numbers always look worse than the campaign is.
3. Compare against the healthy bands.
Acceptance rate: 20 to 40 percent is typical and healthy, and 40 to 60 percent is achievable with a tight ICP and a good profile. Reply rate: 10 to 25 percent is typical and healthy, 25 to 40 percent is possible. These are diagnostic bands, never promises. Use them to locate the problem: "32 percent acceptance is normal, so the lever here is reply rate."
4. Prefer your own history over any band.
Once the workspace has run a few campaigns, its own numbers beat every generic benchmark. The question stops being "is 24 percent good" and becomes "why is this campaign at 24 when the last three were at 35".
5. Low acceptance points at targeting or profile, not copy.
If people are not accepting, they are not reading anything you wrote in the message steps. Look at the ICP, whether the list is active people, and what the sender's profile looks like to a stranger. A vague headline with nothing in Featured is a profile problem, not a copy problem.
6. Good acceptance with low replies points at the first message.
They said yes and then said nothing. That is the message: too long, too pitchy, a link in it, a CTA that asks for too much, or an opener that reads automated.
7. Good replies with no bookings points at the inbox, not the campaign.
Check reply speed first. Threads that sit for days die, and no campaign setting will fix that. After speed, look at the motion from SOP 16 and whether the ask is too big for the temperature of the thread.
8. Sanity-check the boring explanations before the interesting ones.
A disconnected sender, an account at its cap, working hours that never opened, a list that turned out to be 80 percent the wrong seniority. These explain more flat campaigns than copy ever does.
9. Write the diagnosis down as one sentence.
"Acceptance is fine at 34 percent, replies are 6 percent, so the first message is the problem." That sentence is the input to SOP 18. Without it, the fix is guesswork.
Do this / Not this
Do this | Not this |
|---|---|
Read the funnel stage by stage | Stare at the booked-meetings number |
Wait for 50 sends per message | Rewrite after 15 sends of silence |
Bands as diagnosis | Bands as promises to a client |
Your own history over generic numbers | Benchmarks from a blog post |
Check senders and caps first | Assume the copy is the problem |
One-sentence diagnosis before any change | Change three things and see |
Done when
Every stage of the funnel has a number next to it
The weakest stage is identified
The boring explanations have been ruled out
The diagnosis is written in one sentence
The comparison is against your own history where it exists
