Picture a pest control company spending $5,000 a month on ads: $3,000 on Google and $2,000 on Meta. Google brings in leads at about $45 each. Meta brings them in at about $110. The owner looks at those two numbers side by side, does the obvious math, and switches Meta off. That frees $2,000 a month, and Google is clearly the better channel anyway.
For two weeks, nothing happens. Then the phone gets a little quieter. Fewer people search for the company by name. Google's cost per lead creeps from $45 to the high $50s, and by the end of the second month the business is booking about 30 fewer leads a month to save $2,000. Nobody connects it to the Meta decision, because by then it looks like a slow market.
That company is an illustration, not a client, but the pattern is one we see constantly. A channel gets cut because its own report looks weak, and the bill shows up in a different channel's report a month later.
So here is the short version of this article. Before you cut an ad channel, find out what it feeds. Turn it down by half first, watch your total sales rather than the channel's own numbers, and only switch it off when the rest of the business does not notice. Sometimes the cut is exactly right. You just want to know that before you make it, not after.
Each Channel Gets Graded Alone. Your Customers Use Them All.
Every ad platform reports on itself. Google tells you what Google did. Meta tells you what Meta did. Neither one can see the other, and neither one is built to tell you what it handed to somebody else.
That matters because customers do not stay inside one channel. Two of the most common paths look like this.
A homeowner scrolls past your Meta ad, the one with your truck in a neighbor's driveway, and keeps scrolling. A week and a half later there is a trail of ants along the kitchen counter, and your name is the one they half remember. They type it into Google, click your ad or your listing, and book. Google records a lead. Meta records nothing, or at best a "view" it cannot prove mattered. On paper Meta looks expensive and Google looks brilliant, and part of what made Google brilliant was Meta.
The second path runs the other way. Someone clicks your Google ad from a desk computer, looks around, and leaves. Later, on the couch with their phone, they look you up by name, tap the free listing, and call. Your analytics calls that an organic lead. The paid click that started it gets no credit at all.
This is not just theory. One agency that runs Google Ads for a US law firm paused a Performance Max campaign whose own numbers looked poor. For about two weeks, nothing changed. Then cost per click in the firm's main Search campaign climbed back up from under $10, and its cost per lead pushed past $200. When they restarted Performance Max, cost per click dropped back under $10 within days. Same account, same budget on the Search campaign, and the only change was switching off the campaign that "wasn't working."
A smaller agency saw a starker version by accident. It turned off all paid search for a soccer business for a single day. Paid clicks went to zero, as expected. Organic clicks and impressions dropped sharply with them, then jumped back the day after the ads came on again. One day and one account is not proof of anything, and the agency itself pointed out that a Google algorithm update had been lifting the site's rankings that summer. But it is the same shape, and it is why that agency ramps clients' ads down instead of stopping them cold.
We have written before about the version of this that happens inside a single account, where cutting the video ads quietly drains the retargeting ads that close the sale. That piece, You Will Never Know Exactly Which Ad Made the Sale, covers why the reports never add up. This article is about the bigger decision: pulling a whole channel.
Some Channels Really Are Dead Weight. Their Own Report Can't Tell You Which.
If everything above were the whole story, the lesson would be "never cut anything," and that would be wrong. Plenty of ad spend does nothing at all. The trouble is that a channel's own report cannot tell you which kind you have.
The best-known proof comes from eBay. In 2012 its researchers stopped buying ads on searches for the eBay name on Yahoo and Bing. Almost all of those clicks simply came through the free listing instead, and the paper found no measurable short-term benefit from the brand ads at all. Then eBay switched off all its other search ads for 60 days, in regions covering about 30% of its US traffic. Overall sales barely moved. The ads did pull in new and occasional shoppers, but regular customers made up most of the spend, and they were coming anyway. On average, the researchers concluded, that spending lost money.
A marketer who used Haus told a similar story in a case study for that testing company. A two-week test that switched one channel off in some regions showed that the channel was not adding sales. They cut its spend by about half, saw no change in revenue, and moved the money to a channel that tested better.
Now the other side. Google's own researchers studied what happened when advertisers paused their search ads, and found that on average 89% of the paid clicks were not made up by organic clicks. Google is grading its own homework there, and you should read it that way. But it lines up with common sense: eBay is a name everyone already knows. A pest control company in one town is not. The less known your business is, the more of your ad traffic is people who would never have found you otherwise.
So the question is never "does this channel look cheap." It is "does this channel bring people who would not have come anyway." Searches for your own name are the classic case of spend that can look excellent and add little, because those people were already looking for you. Ads that reach strangers often look expensive and quietly do most of the work. It is the same split we described in Google Ads Catches Demand. Facebook Ads Creates It.: the channel that starts a customer's decision rarely gets the credit when they finally pick up the phone.
There is a second question underneath, which is whether the channel has had a fair chance yet. A campaign that is three weeks old is not a verdict. We cover how to tell an unfinished campaign from a failed one in How Long Before Your Ads Actually Work?. Everything below assumes the channel has been running long enough to judge.
A Channel Cut Without a Baseline Is Just a Guess.
The mistake most owners make is not the cut itself. It is cutting without a "before" picture, so that when something changes six weeks later there is nothing to compare it to.
Start by listing what the channel could be feeding. For most small businesses that list is short:
- Searches for your name. If a channel builds awareness, people search for you afterward.
- Organic and direct leads. Calls and form fills that your analytics files under organic search or "direct."
- Retargeting on other platforms. If your Meta ads send people to your website, your Google remarketing lists fill up from those visits too.
- Other channels' cost per lead. As the law firm found, one campaign can hold another's costs down.
Then pull four to eight weeks of baseline numbers, and pull them from your own records, not just the ad platforms:
- Total leads per week, from every source.
- Booked jobs or closed sales per week.
- Searches for your business name. Google Ads shows these in the search terms report if you run a brand campaign, and Google Search Console shows them for free either way.
- What customers say when you ask how they heard about you. If you do not ask yet, start now. One question on your thank-you page picks up sales your ad reports never see.
- Each remaining channel's cost per lead.
Pick a normal stretch of the calendar to test in. A test that runs into your busy season, a holiday or a promotion will tell you about the holiday, not the channel.
Last, write down what you expect to happen. Something like: "If Meta is feeding Google, searches for our name will drop and Google's cost per lead will rise within a month. If it is not, nothing will move." It takes two minutes, and it stops you from fitting a story to whatever happens afterward.
Halving a Channel's Budget Shows Which of Your Other Sales It Was Feeding.
Here is the test itself, and it is deliberately boring.
Cut the channel's budget by half, not to zero. Half a budget still keeps the channel running, so if it turns out to be feeding something, you lose less while you find out, and you can put it back in an afternoon. Cutting to zero also does things that are slow to undo, which we get to in the last section.
Then hold it there for three to four weeks, plus one full sales cycle if yours is longer than a couple of weeks. The effect of a cut shows up late. In the law firm's case nothing moved for about two weeks, because people who had already seen the ads were still searching and booking. If you judge at week one, everything will look fine, and you will cut the rest of the budget right before the drop arrives.
Each week, look at the same handful of numbers you wrote down:
- Total leads and booked jobs. This is the number that matters most. Not the channel's own report.
- The other channels' cost per lead and cost per click. A feeder shows up here first.
- Searches for your name. Awareness channels show up here.
- "How did you hear about us" answers. Watch for the channel you cut fading out of the answers.
Then read the result.
If nothing moved, the channel was not adding much. Total leads held, the other channels' costs held, name searches held. Cut it further or switch it off, and put the money somewhere that tests better.
If something dropped, you found a feeder. Put the budget back, and from now on judge that channel by what happens to the whole business, not by its own cost per lead. You have just learned something no report in your account could have told you.
One caution. The test needs enough money moving to show up. If a channel spends $300 a month and brings two leads, halving it may not change anything you can measure, and that is an answer of sorts: the channel is too small to matter either way. The smaller the spend, the longer you need to watch before you trust what you see.
Most Wasted Ad Spend Hides Inside Channels You Plan to Keep.
When an owner says "we need to cut ad spend," the instinct is to drop a whole channel. Usually the better first move is smaller and safer: cut the money each channel is wasting, which helps no one.
Work in this order.
First, pure waste. This is spend that cannot be feeding anything because it never reaches a real customer. Searches for things you do not sell. Clicks from outside the area you serve, which Google's default location setting quietly allows. Placements on outside websites and apps that send cheap clicks and rarely real leads, which is why we tell LinkedIn advertisers to turn off LinkedIn Audience Network. Ads running at hours when nobody can answer the lead, which is the one good reason to turn your Meta ads off at night. Cutting this costs you nothing, and it often frees more money than owners expect.
Second, spend that props up other channels. These are the campaigns and channels that look weak on their own but might be feeding the ones that close: prospecting campaigns, video, awareness work, the second channel next to your main one. Test these with the half-budget method above. Some will turn out to be dead weight. Some will turn out to be holding your best channel up.
Last, the channels that drive sales. Every business has one or two. One agency that runs seven, eight or nine channels for some of its clients says only one to three of them ever bring in most of the customers, with the rest supporting them, and that cutting one of the main ones drags the supporting ones down with it. For most small businesses, that main channel is Google Search or Meta. Touch it last, and touch it gently.
This order also tends to be the honest answer to "we need to spend less this quarter." Most accounts we audit have enough waste in the first group to cover a real cut without touching anything that works.
A Switched-Off Channel Costs More to Restart Than to Keep Warm.
Sometimes the test says the channel is not pulling its weight, and switching it off is the right call. Fine. Just know that coming back later costs more than staying, so make the way back as cheap as you can.
The platforms have to relearn when a campaign restarts, and that relearning period costs money. We covered it in detail in Your Slow Season Is the Wrong Time to Stop Advertising, including Meta's seven-day line. Beyond that, a few cheap habits keep your options open:
- Leave your tracking installed. Keep the Meta pixel and the Google tag on your site even with ads off. Website audiences only fill while the tracking runs, and Meta's website audiences look back at most 180 days. Pull the pixel, and when you come back your retargeting audience starts from nothing.
- Keep your customer lists uploaded. They cost nothing to keep and they are the first thing you will want when you restart.
- Keep a small brand campaign if competitors bid on your name. This is the one place where spend that "adds little" still earns its keep: it stops a competitor's ad sitting above your own name.
- Keep the account alive. Google treats an account with no spend for more than 15 months as inactive and cancels it automatically. You can reactivate it, but there is no reason to give yourself the chore.
- Write down what you saw. Note why you cut the channel and what the numbers did. A year from now, when someone pitches you that same channel again, you will want the real answer, not a memory of one.
Total Sales Are the Only Report That Sees Every Channel.
Every ad report is honest about its own slice and silent about everyone else's. That is not a flaw anybody is going to fix. It is how the platforms are built, and it is why the most expensive cuts are the ones that look obvious on a single screen.
The good news is that you do not need special software to protect yourself. You need a before picture, a half-budget test, a few weeks of patience, and the discipline to judge by total leads and booked jobs instead of by the channel's own grade. Do that, and you will cut the dead weight with confidence and keep the quiet channel that was carrying more than its share.
This is a good part of what we do day to day: working out which spend in an account is waste, which is support and which is doing the heavy lifting, before anyone starts switching things off. If you would rather have a second set of eyes on that, we are happy to help. And if you would rather run the test yourself, everything you need is above.




