Should I kill or keep this marketing channel?
Stop a marketing channel once it has had enough volume to judge and the early signal is still under its kill line. Kill or Keep, a free tool from InstinctGTM, checks your numbers against published rules and says keep it, kill it, or give it 14 more days, with the number that would change the verdict.
No signup. The result is yours before we ask for anything.
Run it on your own domain
How this works
- 1Too early comes first. Every channel has a minimum volume and a minimum number of days, and below either one there is no verdict, only how much more is needed. Cold email needs 100 emails sent and 14 days.
- 2A booked call, a customer or a sale keeps a channel. Paid ads are judged on cost per result instead: at or under $25 a result keeps them.
- 3Under the kill line, the channel is killed. For cold email that is 0.5% replies or fewer; for posts, an average of 1 reaction or fewer per post; for paid ads, $50 or more a result after $100.
- 4Anything between gets 14 more days with one thing changed, so the next test shows what made the difference.
- 5With a price entered, a customer who cost more than a third of their first year adds a warning to Keep it.
- 6Every result prints the number that would flip it, and the rules version it was judged on.
The rules, in full
Rules version 1, published 2026-10-28. Every number is a InstinctGTM starting rule, never an industry benchmark.
| Channel | Minimum volume | Minimum days | Kill line | Keep line |
|---|---|---|---|---|
| Cold email | 100 emails sent | 14 days | 0.5% positive replies or fewer | 1 booked call, customer or sale |
| LinkedIn posts | 8 posts published | 30 days | 1 or fewer reactions, comments and reposts per item, on average | 1 booked call, customer or sale |
| X posts | 8 posts published | 30 days | 1 or fewer likes, replies and reposts per item, on average | 1 booked call, customer or sale |
| Instagram or TikTok | 8 posts published | 30 days | 1 or fewer likes, comments, shares and saves per item, on average | 1 booked call, customer or sale |
| YouTube | 8 videos published | 30 days | 1 or fewer likes and comments per item, on average | 1 booked call, customer or sale |
| SEO or blog content | 6 articles published | 90 days | 50% visits from search or fewer | 1 booked call, customer or sale |
| Paid ads (Meta) | $100 | 14 days | $50 or more a result, after $100 | $25 or less a result |
| Paid ads (Google) | $100 | 21 days | $50 or more a result, after $100 | $25 or less a result |
| Paid ads (LinkedIn) | $100 | 21 days | $50 or more a result, after $100 | $25 or less a result |
| Communities | 6 helpful replies posted | 30 days | No replies or messages back after 10 | 1 booked call, customer or sale |
| Partnerships or referrals | 6 partners asked | 90 days | No introductions made after 10 | 1 booked call, customer or sale |
| Events | 6 events attended | 90 days | No follow-up conversations after 10 | 1 booked call, customer or sale |
| Something else | 10 things put out | 30 days | No responses after 20 | 1 booked call, customer or sale |
Changes to the rules
- v1, 28 Oct 2026: First rules. Channels with a plan engine lane use that lane’s starting criteria.
A worked example
Example numbers
Give it 14 more days
400 emails sent brought 6 positive replies, which is above the kill line, but no booked call, customer or sale yet. Run one more test and change one thing only, so you know what made the difference. Change one thing: the offer: what you ask for in the email.
What it cannot tell you
- Channels with long sales cycles, like paid ads for a large contract, can need longer than the minimum days before a booked call shows up.
- It reads only the numbers you enter. A reply that was hostile counts the same as a warm one, so count replies honestly.
- It judges one channel at a time and cannot see a channel helping another, like posts that make cold email land.
- The lines are starting rules. Your own history beats them once you have it.
Questions people ask
How long should you test a marketing channel before stopping?
Until it has had enough volume and enough days to judge, decided before you start. The starting rules here are 14 days and 100 emails for cold email, 30 days and 8 posts for LinkedIn posts, and 90 days for articles, because search is slow to answer.
When should you give up on a paid ad channel?
When you have spent enough to judge and a result still costs too much. The starting rule here is to stop once $100 is spent and a result costs $50 or more, and to keep going while it costs $25 or less.
What is a kill criterion in marketing?
A written rule, set before a test starts, that ends the test at a set point unless a set result appears. Writing it first stops the decision being made on mood, and it is the difference between a test and a habit.
Is my cold email working?
Count replies against emails sent once you have sent at least 100. At 0.5% replies or fewer, the starting rule here says stop. Any booked call says keep, and anything between says change one thing and run 14 more days.
Why change only one thing at a time?
Because a result after two changes cannot say which change did it. One change per test is slower, and it is the only way the next decision rests on something you know.
Are these numbers industry benchmarks?
No. They are InstinctGTM starting rules, the same ones a plan starts each channel with. Your own numbers replace them once you have a few tests behind you.
The short version, if you would rather not run it
Most channels die of neglect or live on hope, and both come from the same gap: nobody wrote down, before the first post or email, what result would mean stop. Without that line, every week brings one more reason to keep going or one bad day that ends a channel which was working.
The fix is boring. Pick a minimum volume and a minimum time, so nothing is judged on a handful of sends. Pick a kill line for the early signal, since replies and reactions arrive long before customers. Then treat any real result, a booked call or a sale, as the thing that keeps a channel alive.
When the answer is in between, the honest move is one more test with one variable changed: the opening line, the offer, the audience. Change two and the result tells you nothing.
The other tools
- How do I get the transcript of a YouTube video?
- How do I get the text of a TikTok video?
- How do I get the transcript of an Instagram Reel?
- How do I turn a post into a shareable image?
- How do I make a YouTube link open in the app?
- How many characters is my post, the way X counts?
- What is my engagement rate, and is it any good?
- Will email from my domain land in the inbox?
- Do AI assistants recommend my company?
- Which directories should my SaaS be listed in?
- What is wrong with my cold email?
- Does my writing sound like AI?
- Where do my buyers actually hang out?
- What does my website actually say I do?
- Which companies are buying what I sell right now?
- How do I make bold or italic text in a LinkedIn post?
- Is my homepage headline any good?
- Is the tracking on my website working?
- Which marketing channel should I start with?
- How many sales calls do I need to hit my revenue goal?