Deciding whether to continue, change or stop a channel

The review date arrives, and your channel has produced the sort of result that refuses to be shoehorned into a case study. You published the posts, sent the emails or joined the conversations. A few suitable people responded - perhaps one person even paid (perhaps nobody did). Meanwhile, somebody on LinkedIn has announced that their new content strategy generated $80,000 in thirty seconds.

This is when channels get abandoned through irritation or kept alive through stubbornness.

You need to make two decisions, in order. First, did you run the test you meant to run? Then, what does the evidence support doing next? The answer should be one of four things: continue unchanged, continue with one named change, pause until a stated condition exists, or stop.

Return to the purpose you gave the channel

Bring back your distribution brief, activity log and measurement plan. At the top should be the channel's one job, its main measure and the point at which you agreed to review it.

The purpose decides what counts as progress. If you used Reddit to learn how suitable customers describe a problem, a relevant conversation can make the test worthwhile even when nobody buys. If you used cold outreach to create qualified sales conversations, 100 delivered emails do not count as success by themselves. Delivery describes your activity; the replies and conversations describe what customers did afterwards.

Your review can lead to four decisions. This table compares what each one means and what you need to record before moving on.

“Not enough evidence yet” is a fine diagnosis, although it is not a fifth place to live forever. It usually means you need to complete the planned test or change its design so that the next review can produce a responsible verdict.

Check that the test actually happened

Before judging the channel, compare the work you planned with the work you completed.

• Did you reach the intended customer, rather than whoever happened to be easiest to find?
• Was the tactic permitted and visible in that setting?
• Did the message, offer and requested next step stay consistent enough to compare?
• Did the test run for the agreed period or number of customer opportunities?
• Did the tracking work?
• Did the planned follow-up happen?

An imperfect test can still teach you something; for example, missing one post from a four-week plan is unlikely to ruin the whole record. Reaching an entirely different audience, changing the offer halfway through or losing every source tag creates a much more serious gap.

When the test is materially incomplete, judge the test rather than the entire channel. Complete it if the missing work is still timely and it makes sense, or redesign it if the original plan could never have answered the question. “LinkedIn doesn't work” is quite a leap from “I posted twice, changed the subject, got bored, and became busy.”

Find the first place customers stopped moving

Once you know what happened, follow the customer route in order and find the first weak link. This tells you where to begin your investigation rather than proving the cause.

The distinction protects useful channels from problems elsewhere. If several customers buy through a partnership and then fail during setup, changing the partnership may remove a source of suitable customers while leaving the real problem untouched. The evidence has moved the investigation into activation; customer conversations and product behavior can help explain what happened next.

Continue when the intended movement is happening

Continue unchanged when suitable customers are taking the action you chose and you can sustain the work until the next review. That action might be a qualified conversation, an activated trial or a purchase.

A test doesn't always need a sale to continue. An early founder-led channel may be doing its job by creating conversations with people who have the problem and can buy. Likes and impressions cannot replace those conversations unless attention was genuinely the channel's agreed purpose.

Change the smallest thing that could answer the question

Choose change when the intended customer is still there, and the record points to one possible weakness in your first attempts. The change could sit in the audience, message, offer, proof, next action or tactic.

Name one primary change to try. A substantial offer or product change will require related tweaks, but the record should still say which idea you are testing. Change the customer, price, message and channel together, and you won't know which part helped.

There's no universal number of attempts that proves a channel works or fails. A practical brake is useful, though: after two completed tests produce no relevant customer movementAn observable action a suitable customer takes after your activity, such as replying, booking a call, starting a trial, paying or returning to the product., do not roll straight into a third by changing another subject line or post format. Pause long enough to collect fresh customer evidence, or write down the materially better reason that makes another test worth the work.

Pause when the route is premature

Some channels are plausible and poorly timed. For example, paid acquisition may need reliable conversion data, and a partnership may need a working integration. Search content may need publishing capacity that you simply don't have during a delivery-heavy month.

A pause needs a return conditionThe specific event that brings a paused channel back into consideration, such as completing a case study, reaching three retained customers or gaining the product capability the channel requires.. “Try again later” will sit in a planning document until your hair turns grey. A useful condition might be three retained customers, a completed case study, a fixed test budget or a named product capability. Once it exists, you can review the channel again.

Stop when your effort is better used elsewhere

Stopping is reasonable when you can't reach suitable customers under the channel's rules, or you find that you're attracting the wrong people. You may also want to stop when the product or customer has changed, or you simply don't enjoy using the channel itself.

Opportunity costWhat you give up by spending limited time, cash or capacity on one channel instead of another tested option. belongs in the decision, too. If two completed tests compete for the same four weekly hours and one creates materially stronger customer movement, continuing both may weaken the better one.

Keep the test, evidence, reasoning and anything that could reopen a decision. A channel that's wrong at ten customers may become useful at 100; a channel full of the wrong buyers will remain full of the wrong buyers, however diligent you are.

Give the channel enough time to show the right signs

Cold outreach can show replies within days. SEO, GEO and Pinterest build over longer periods. Your review window should reflect the time in which the intended customer action could reasonably happen.

Long-term does not mean exempt from evidence. During a slower test, look for early movement that fits its purpose, such as relevant visits, saves, returning readers, citations or inquiries. A six-month commercial result may be uncertain while the first six weeks still show whether the work is reaching anybody useful.

Keep the counts beside the rates. “One payment” means something very different when it comes from two conversations instead of 200. Early results are useful and fragile; generic benchmarks rarely repair that problem because another business will have a different buyer, offer and buying process.

Count cash, time and capacity separately

A channel can produce customers and remain a poor use of your resources. Record cash, your hours and the effect on delivery separately. A $0 community test that takes eight hours has a different constraint from a $300 advertisement that takes one.

If a promotion fills the week with small, unsuitable projects, the acquisition number can look healthy while the business becomes harder to run. Judge the company you can deliver, not an imaginary one staffed by eight rested versions of you.

Watch the stories that arrive before the evidence

You may want to stop because you're bored, a competitor appears to be winning elsewhere, or the algorithm has become a convenient suspect. You may want to continue because one impressive person replied, or because stopping would make the previous month feel wasted.

That last pull has a name. Research on the sunk-cost effect found that people become more inclined to continue once they have invested time, effort or money. Agree on the purpose, main measure and review pointThe date, number of customer opportunities or other boundary agreed before the test, when you'll examine the evidence and decide what happens next. before the result arrives, then judge the next unit of effort. You can't recover a month already spent by spending another - that's the sunk costTime, effort or money you've already spent and can't recover. It shouldn't decide whether the next unit of effort deserves to be spent. fallacy.

Early channel decisions still involve judgment, small samples and incomplete attribution. Write down what happened, what remains unknown and what would change your mind.

RelayReport cannot name a winner yet

RelayReport's first scorecard shows four qualified trial starts. Two have LinkedIn as their first known source, one came from cold email, and one is unknown. Two trial users activated, and one paid $49. The test used $0 in cash and six founder hours.

It is tempting to write “LinkedIn wins.” The record doesn't show audience reach or the tactic for either channel; it also doesn't connect the activated and paid users to a source or tell us when each test began. The unknown source needs to stay unknown.

The current decision is to change the test. RelayReport can repair the activity and source record, then run one bounded LinkedIn test with a named objective, a consistent tactic and an agreed review point. The existing result remains productive: it simply cannot carry the verdict somebody wants to put on it.

Record the decision you are making now

Complete the channel decision record with the evidence you already have. It asks for the test you planned, the work you completed, the first weak or unknown step, your cash and time, one verdict and the evidence that would overturn it.

If you want AI to prepare the review, give it the records rather than access to your live campaigns. Use this prompt:

I am reviewing one go-to-market channel. Use only the distribution brief,
activity log, measurement plan, customer records and feedback I supply.

Do not invent missing figures or infer why a customer acted. Do not connect to
accounts, change a campaign, spend money, publish content, stop a channel or
contact anybody.

1. Restate the channel, tactic, intended customer, one job, main measure and
   agreed review point.
2. Compare the planned test with the test I actually completed. Check the
   intended buyer, permitted tactic, message and offer, duration, tracking and
   follow-up.
3. Separate my activity, suitable-customer movement and commercial results.
   Keep counts and denominators beside rates.
4. Mark every missing or unreliable input. If the test was materially
   incomplete, say that before evaluating the channel.
5. Locate the first weak or unknown step: suitable people reached, relevant
   response, buying progress, purchase, value received or sustainability.
6. Separate what the record shows from possible explanations that still need
   customer evidence.
7. Show cash spent, founder hours and delivery effect separately. Include any
   tested opportunity cost I supplied.
8. Draft the case for each verdict: continue unchanged, change one primary
   element, pause until a named condition or stop. Do not choose for me.
9. For the strongest supported verdict, draft one next action, the next review
   date or return condition, and the evidence that would overturn it.

Return standard Markdown in my first person. End with a decision record for my
review and approval.

You're ready to move on when the record contains a decision you can act on, a reason supported by the evidence and a point at which you will look again. The next chapter turns these individual reviews into a weekly go-to-market rhythm, so your channels do not need a small constitutional crisis each week.