The Farmer with the Magic Harvester
A farmer once poured every dollar he had into building the most efficient corn harvester ever made. It was almost magic, picking each ear at exactly the right moment, faster than any of his neighbors could manage. To fund it, he stopped fertilizing his fields. He stopped sowing new seeds.
The first spring, he had a bumper crop and beat every farmer in the county. The second spring, his crop was a little smaller, but the harvester still won the day. By the third spring, he barely matched the previous year. By the fourth, his magnificent machine was scraping a meager harvest off thinning fields.
In nearly every marketing exec ed program we’re part of, the same question surfaces: brand marketing or performance marketing?
"Brand marketing" is upper-funnel work - growing awareness of your offering and building positive, persuasive associations. The idea is that when those folks enter the market for their next purchase in the category, you have an advantage. Think old-school TV, YouTube pre-roll, billboards, white papers, conference talks — things that reach people when they aren't actively looking. For farmers, this is fertilizing the fields and sowing new seeds.
"Performance marketing" focuses on people who are already in the market, usually based on data we have about them. Google search results, social ads, even ads served inside LLM results that are all aimed at a click and a conversion. For farmers, this is harvesting the ears that have already grown on their stalks.
People ask us, "Which one do we invest in?" Just to cut right to it: you need both, and figuring out the proper mix requires a commitment to test and learn. The shorthand we sometimes use is that one half generates demand and the other captures it.
Let's be honest. Performance ads have some pretty compelling benefits. They're easy to measure. Someone clicks or they don't. And that binary response means we can automate it. Google has been personalizing performance content in search ads since the mid-2010s, and now they (and others) offer solutions that automatically find the right people, figure out the optimal message, and use GenAI to generate a custom ad across the network. So — wow. A) I can measure it and show my boss how well it worked yesterday, B) I don't even have to pick targets, and C) maybe I don't even need to make the ads. Pretty compelling.*
Brand marketing, by contrast, is notoriously harder to measure and slower to pay off. There are testing services that'll tell you how a piece of creative performs against category norms and prior work. We can track brand health metrics: awareness, consideration, intent to buy, the positive and negative associations different groups have with the brand. But performance tells me sales, and brand tells me some metric that might lead to sales down the road. The uncertainty makes people uncomfortable.
Most folks intuitively get that you should be forming positive associations before a customer is in the buying phase. Once they're in market, people tend to stick with the brands they're familiar with and don't bother to stray without a compelling reason which is usually a lower price or a strong brand message.
So how do you justify brand marketing internally? One answer is a market test. Pick two markets, lean into upper-funnel work there, and track brand health metrics against control markets. You should expect the test markets to outperform over time. It might take a month. It might take a year. And while you're tracking brand metrics, track performance metrics too because when brand marketing works, performance metrics lift as well. Lower-funnel demand capture improves because more demand is being generated upstream.
The magic harvester is great if you only need to harvest one crop. If you want a farm that lasts, you have to seed and fertilize. When your team has doubts, pick a couple of fields and show them. Just make sure you give it the time to grow.
The incrementality of performance marketing is a story for another time. The short version: performance ads tend to work best on people who were going to buy anyway — meaning you didn't actually need to pay to reach them, and you're probably overestimating the tool's contribution.