The 2027 Marketing Plan Starts With What You Stop Paying For
If you can’t prove which spend is driving revenue, every new pitch for your 2027 marketing plan will sound like a good idea. Most reports show clicks and impressions, which reveal very little about what deserves to stay, and a stop list fixes that in five steps. What Goes on the Stop List Before Anything Gets Added to Your 2027 Marketing Plan? Every channel and tactic you pay for goes on it, ranked from best supported to least supported, before you approve anything new. Our guide to how much a small business should spend on marketing in 2027 sized the total, and the stop list ranks what’s inside it. Most pitches ask you to spend more, and that includes the ones agencies like ours send, but few say what the new spend should replace. A plan built that way only ever expands. Cutting on reflex is the other default. It’s easy to see why when profits come in short and someone needs an answer fast. The CMO Survey from Duke’s Fuqua School of Business asked U.S. marketing leaders how often executives cut marketing before other areas once profits fall short. The average answer was about 45% of the time. In the minds of the 160 leaders surveyed, marketing often gets cut first. That’s a perception, not a tally of budgets cut, and many of those leaders run companies far larger than yours. A cut made under that pressure follows the deadline instead of the evidence, and a stop list gives you the evidence first. The obvious way to rank your spending is to sort last year’s return on investment (ROI) reports by result. Those reports come with a catch. Why Can’t Your Marketing ROI Reports Rank Your Spend? Dashboard ROI compares people who saw an ad with people who didn’t. Since the people who see ads are often the ones already likely to buy, the ads can take credit for sales they didn’t cause. Two studies show how far that goes, and a third explains why small campaigns can’t easily fix it. Reports Tend to Overstate What Ads Produced Economists Tom Blake, Chris Nosko, and Steven Tadelis ran field experiments on eBay’s paid search ads. Ads bought on searches that included eBay’s own name showed no measurable short-term benefit. Ads on other keywords did bring in new and infrequent customers. Frequent customers, whose buying the ads didn’t change, made up most of the spending, so average returns came out negative. Standard estimates built from existing data had shown far better results, because clicks and purchases tend to move together. Researchers at Northwestern’s Kellogg School of Management and Facebook compared 15 ad experiments against the observational methods many advertisers use. Those methods read results from existing data with no control group. They mostly overstated the effect of the ads. In half of the studies, the estimated lift in purchases was off by a factor of three. Some estimates came in too low, so reports can miss in both directions, though they usually miss high. Direction is one problem, and scale is another that affects smaller companies most. Small Campaigns Rarely Produce a Clear Return Economists Randall Lewis and Justin Rao studied 25 large field experiments with major U.S. retailers and brokerages. Most reached millions of customers, yet the typical range of plausible returns was over 100 percentage points wide. With a range that wide, a reported gain can easily hide a loss. Sales per customer vary so much from one buyer to the next that an informative test can need more than 10 million person-weeks. That equals a million customers watched for ten weeks. A campaign that reaches thousands of people has far less data to work with. The single ROI number on its report is one point inside a very wide range, which makes reported return an unreliable way to rank spend. A better ranking comes from evidence you can verify yourself. How Do You Grade Each Line Item by Its Evidence? Sort every channel and tactic into one of three grades, proven, plausible, or unknown. The grade depends on how directly you can trace customers or revenue back to that spend. Start with a list from your invoices and ad accounts. Give each recurring charge a row, with its cost, the result it’s supposed to produce, and who owns it. The grades work like this: Grade What the evidence looks like What it earns Proven Booked calls name it, a coded offer tracks it, or a test confirmed it First claim on freed budget Plausible A sound reason it works, but other activity could explain results A place for now, plus a pause test Unknown Reports show clicks or impressions, and no one can say what it produced A pause test comes first Two questions do most of the sorting. Can you name customers or sales this spend produced, and did anything change the last time you paused or shifted it? Yes to both makes it proven, yes to one makes it plausible, and no to both makes it unknown. Grade by the evidence you hold today. A platform’s own conversion report can support plausible but not proven, since it has the same overstatement problem. Unknown only means the evidence is missing, and a short pause can supply it. How Do You Test a Cut Before You Commit to It? Pause the channel for a window you set ahead of time, and bring it back if leads or sales drop. Test one unknown or plausible channel at a time, in four steps: Write down what you expect to happen. Name the calls, leads, or sales this spend should produce and where you’ll see them. Set the window and the end date. Cover at least one full sales cycle, and put the end date on the calendar before you pause anything. Change nothing else. When two things shift at once, you can’t tell which one moved the results. Decide what counts as a drop. Set the threshold before you