Guides · Marketing
How to Test a New Marketing Channel: Write the Stop Rule First
By Josh Croom · October 27, 2026 · 10 min read
To test a new marketing channel, run it as a small test with a fixed budget, a two-week window and one number that decides it. Write down the rule for when you'll stop before you spend the first dollar, so the test ends with an answer instead of a charge that runs for months.
The ad nobody can explain
Month seven. A shop owner agrees to try a new kind of social media advertising in March, and since nobody sets an end date, it just keeps running. By September the post has a pile of likes and the card has a steady charge, and nobody in the office can say whether one new customer came from it.
The ad started without a finish line, so no one ever had a day set aside to look at the number and decide. A good test puts the dates, the budget, the one number that matters and the stop rule in writing before any money moves.
Why most tests should be small
In Harvard Business Review, two test experts reported how often new ideas helped once someone tested them. The third number comes from the paid ad accounts we run.
- 10% to 20%of online tests at Google and Bing came out aheadthe share that helped
- 1 in 3tests across Microsoft made things worseabout a third helped, and a third made no difference
- 4%of full weeks on the paid accounts we run brought zero conversionsJuly to September 2026
Even experts guess wrong
In 2012, a Microsoft employee working on Bing had an idea to change how the search engine showed ad headlines. The idea sat untouched for more than six months, because the program managers ranked it a low priority. When an engineer finally ran a simple experiment on it, revenue went up 12%. Kohavi and Thomke wrote that the change was worth more than $100 million a year in the United States alone.
A small idea can pay off in a big way, and a big idea that everyone loves can do nothing at all. You can't tell ahead of time, and neither can the person pitching you, so find out with a small budget and a clear rule.
Why a two-week stop rule is fair
A tight stop rule can feel harsh, as if you're not giving the channel a real chance, but our own numbers point the other way. Only 4% of full weeks on the paid ad accounts we run brought in no conversion at all, so advertising that works rarely goes a whole week with nothing to show.
So when a new channel spends half its budget and nobody signs up, treat that as your answer. Two weeks gives a new channel room to show whether people respond, and it keeps a dud to one small budget instead of a year of charges.
How to test a new marketing channel, step by step
These steps work for a mailer, a social ad, a booth at an event or a new kind of post. Do them in order, because each one leans on the one before it.
Step 1
Start with the cheapest test
Point the test at a page you already have before you build anything new. If the test brings people in, build the new page then, because a page built for a channel that flops wastes money on top of the ad spend.
Step 2
Pick the one number that decides
Count the people who showed up: they booked, came in, called and became a job, or bought. A signup who never shows is worth nothing to you, so count at the door. Treat visits and cost per signup as helper numbers, because they explain the result without deciding it.
Step 3
Give the test its own way to count
Use a link only this test uses, and add the channel to your "How did you hear about us?" question as one of a few set answers. An online ad can also get its own tracking line that rings your main number. Never print a tracking line on a mailer or a sign, because the print lasts longer than the phone line.
Step 4
Set the budget and the dates
Pick a small, fixed budget for two weeks, an amount you could lose without it hurting. Put the start day, the stop day and the review day on the calendar before the test begins, and name one person who owns it.
Step 5
Write the stop rule down
Decide what ends the test early, and write it where everyone can see it. For example, if half the budget is spent and nobody has booked, you stop. Nobody changes the rule once the test starts, because the urge to give it one more week is strongest after the money is already spent.
Step 6
Check the rules before you launch
On social sites, ads about credit, housing, jobs and some money offers fall into special groups with narrower targeting, so check before you plan who sees them. Never hint that a school, a venue or a government office backs you, and name a venue only as the place. Get written permission before you reuse words someone else owns.
Step 7
Pick one of three on review day
On review day you have three choices: keep the channel, tune it or stop it. Stopping is a good result when a number backs it, because you learned the answer for one small budget. A tune means you change one thing and run another two weeks under the same rule.
Step 8
Grow only what paid
Give a channel more money only when it brings in people who show up at a cost you'd gladly pay again, and keep changing one thing at a time as you grow it. Try a video only after a plain image works, and add a second audience only after the first one holds its cost week after week.
What counts and what doesn't
Most owners judge a test on the numbers that are easiest to see. Those show up first, but they rarely tell you whether the channel paid for itself.
Helper numbers
- Likes, shares and comments.
- Clicks and website visits.
- Signups from people who never show.
- How many people saw the ad.
The number that decides
- Appointments people kept.
- People who came in the door.
- Calls that turned into booked jobs.
- Orders that shipped and stayed paid.
Give the test its own name
When each test has its own link or tracking line, its people land in your records under its own name, so you can count them apart from everyone who would have found you anyway.
Calls and forms
Your test plan, on one page
Fill this in before the test starts, and put it somewhere the whole team will see it. Leave no line blank, because each blank is a question nobody answered before the money went out.
| Line | What to write |
|---|---|
| Channel | One new channel, named plainly, such as a mailer to two ZIP codes or one kind of social ad. |
| Budget | A fixed amount for the whole test that you could lose without it hurting. |
| Start day and stop day | Two weeks apart, both on the calendar. |
| The one number that decides | People who booked, came in or bought, counted at the door. |
| How we count it | Its own link, its own tracking line or its own answer to "How did you hear about us?" |
| When we stop | Your stop rule, such as half the budget spent and nobody booked. |
| Review day | The day after the stop day, with the owner in the room. |
| Owner | One named person who runs the test and reports the number. |
Mistakes that spoil a test
These four habits keep a test from giving an answer, even when the channel itself was good.
- Judging the test on clicks or likes, which tell you people saw it but not that anyone came in.
- Running two new channels at once on one phone number, so you can't tell which one made the phone ring.
- Changing the stop rule after the test starts, usually to give the channel one more week.
- Adding a second channel before the first one has paid for itself, which splits your budget and your attention.
Is your next test ready?
Answer five questions about the test you're planning. Your answers stay on your own screen.
Are the start day, stop day and review day on a calendar?
Did you pick one number that counts people at the door?
Does the test have its own link, tracking line or answer choice?
Is the stop rule written down where the team can see it?
Is this the only new channel you're starting right now?
Your test is ready, so launch it, leave the rule alone and read the number on review day.
You're close. Fix each question you answered No before launch, because each one is a way the test can end without an answer.
Hold off for a few days and fill in the test plan above first, so the money you spend buys you a clear answer.
Before you launch
Tick each one off as you go. Your ticks stay on this device.
0 of 7 done
When the channel is Google Search
If the channel you want to test is Google Search, check the demand first. Our search tools count the Google searches for your trade in your town, so you can see whether enough people are looking before you spend anything. A search ad can only reach people who are already searching for what you sell.
Then make sure someone answers, because a test that brings in calls nobody returns will look like a failure even when the channel did its job. Pair it with the callback routine in our speed to lead guide.
Our plans for local businesses come with a stop rule of their own. Stop anytime in your first 30 days, and the fee never starts. The 13-cycle plan has no 30-Day Stop, and our pricing page shows both ways to start.
Talk it through with us
If you're weighing a new channel and want a second set of eyes on the plan, book a call with Miguel. He'll ask what you're thinking of trying, how you count new customers today and what a good result would look like for you.
Questions people ask
How long should you test a new marketing channel?
Two weeks is a good length for a brand-new channel with a small, fixed budget, followed by a review day. A running Google Ads account or a Google Ad Grant is different, so judge those over three months against their own past.
What should you measure in a marketing test?
Measure the people who showed up: they booked, came in or bought. Clicks, likes, visits and cost per signup help explain the result, but they don't decide it.
How much should you spend to test a marketing channel?
Spend a small, fixed amount you could lose without it hurting, and set it before the test starts. Write a stop rule too, such as stopping when half the budget is gone and nobody has booked.
Can you test two marketing channels at the same time?
Only if each one has its own link and its own tracking line. On one phone number, two new channels blur together, and you won't know which one made the phone ring.
See how we run Google Ads, with every call and form counted
Read next
- Call Log Template: What to Write Down From Every CallA ready-to-copy call log, the one question to ask every caller, and how to sort wrong numbers.
- ChatGPT Custom Instructions Examples: Make Each Fix Only OnceEight custom instructions examples to copy, and a four-line card that carries your fixes into every chat.
- ChatGPT Prompt Templates: Write the Weekly Ask Once and Save ItCopy-ready ChatGPT prompt templates for the jobs you run every week, and how to save each one.