The Q4 Call
It’s mid-October. Your Q4 pacing is visible by now, 2027 planning starts in a few weeks, and somewhere on your media plan is a channel you’ve quietly decided to stop funding.
Before you cut it, one question: did you ever actually test it?
Earlier this year, we published four requirements that have to be in place before a dollar goes into a new platform. Read backwards, they tell you whether the thing you’re about to kill ever got a fair run.
Neutral attribution, in place before launch. Without a multi-touch source of truth, you were grading the new channel on its own self-reported numbers while Meta graded itself on different ones.
A success metric, agreed in writing, before the first impression. Not after.
Audience overlap, checked. A new channel is worth money because it reaches people your existing channels don’t. If it was largely hitting the same people already seeing your Instagram and Google ads, you bought frequency and called it reach.
A minimum learning period, committed to in advance. Most platforms need 30 to 60 days.
Which points at the trap underneath all of this. Benchmarking a discovery channel against your best bottom-funnel channel’s ROAS guarantees the discovery channel loses. You’ve designed the test so the answer was fixed before it started.
Here’s why this matters more than the money you’d save by cutting.
A channel you cut for real reasons is a decision. A channel you cut because the test was badly built becomes a belief, and beliefs are sticky. It gets written into next year’s plan as settled fact, repeated by people who weren’t there, and it keeps you out of that channel long after the conditions that made it fail have changed.
So run the audit before you make the cut. Twenty minutes per channel, using the four above. Some of your kills will hold up. The ones that don’t are channels you should re-test properly in Q1, with the requirements actually in place.
What’s the channel you’re about to cut? Reply and tell me which of the four was missing. I’d guess it’s the learning period.
Ask Your Team
“When we launched this channel, what did we agree success would look like, and where is that written down?”
Green flag: a number from before launch, in a doc, with a date on it.
Red flag: a ROAS comparison against Meta that got constructed afterward.
“How long did we run it before we judged it?”
Green flag: 30 to 60 days, and they can tell you that platform’s learning period.
Red flag: under 30 days. Whatever that produced, it wasn’t a result.
You’re invited to the world’s largest virtual email marketing conference.
Become an email marketing GURU for FREE.
GURU Conference, powered by Constant Contact, is back November 12–13. It’s 100% virtual and 100% free, and it’s packed with tactics you can use right away on newsletters, deliverability, design trends, AI, and what NOT to do with email.
What to expect:
-
Speakers including Molly Ringwald, Dan Levy, Amy Porterfield, Frank Vella + more
-
Sessions from world-class marketers at top brands
-
Networking every day
-
DJs, world-record breaking, and an alarming amount of rom-com fun!
Last year, 29,000+ marketers joined us. This year is even bigger, and spots are limited.
What Good Looks Like
|
Requirement |
If it was missing, what you actually learned |
|---|---|
|
Neutral attribution in place |
Nothing comparable. You read the channel’s own homework. |
|
Success metric agreed in writing, pre-launch |
Nothing durable. The goalposts moved after kickoff. |
|
Audience overlap checked |
Unknown. If overlap turned out high, cutting is correct. If nobody checked, you don’t have grounds either way. |
|
30 to 60 day learning period completed |
Nothing. The platform never finished optimizing. |
One of these failures justifies a cut. Three of them mean you’re about to turn a non-result into a company belief.
This Week
Runway shipped an AI agent that creates and measures ads. Genuinely interesting, genuinely early. Revisit when someone you trust has run real spend through it.
Agency execs have quietly stopped hiring entry-level talent. A real story about the industry’s future. Nothing for you to decide this quarter.
Not ignorable: if ChatGPT ads are on your 2027 test list, note that the channel currently fails requirement one. Measurement gaps are keeping most advertisers’ budgets at test level, which means you cannot yet compare it fairly against anything else you run.
That’s an argument for sizing it as a learning budget with a defined question attached, not for sitting out. Just don’t let it into a channel comparison it can’t honestly compete in.