Why is it a false narrative? The basic premise is simple: company X spends money on ads, then turns those ads off and finds that the revenue does not change. Then the conclusion is made that ads don't work. On the surface, it seems like sound logic, and that's why articles like that do find their way into reputable publishers.
But allow me to make an analogy that will hopefully highlight how absurdly oversimplified that takeaway is:
Company X decides to switch from PHP to Go, and in the process ends up with all sorts of bugs due to how the new garbage collector works. This costs the company a lot of money, and the conclusion is made that Go is a shitty programming language.
What both of these cases have in common is that neither takes into the account the proficiency level of the operator. It's somehow intuitively obvious that as you trade programming abstractions for performance, you will require a higher level of developer proficiency to extract that performance. But nobody makes that connection in the marketing world. The often quoted examples of marketing calamities, Ebay and Uber, have marketing budgets in the 100s of millions, which can be only deployed using multi-channel strategies. In other words, you might give 10% of that to Pinterest, and might be delighted to see that Pinterest reports a 300% ROAS. But it turns out that Pinterest will also report on a much longer lookback window than other platforms, and will use a different attribution model than, say, Facebook or Google. If you turn off your Pinterest spend and discover that the reported 300% ROAS was not accurate, it doesn't mean that ads don't work - it means that Pinterest had bad incrementality, and that it was some other platform that deserves all the credit. In short, in the days, weeks and sometimes months that it takes for a customer to convert, Pinterest had some impact, but much less than its in-platform figures indicated. You can measure incrementality using 3rd party vendors like Measured.com. The fact that Ebay's and Uber's marketing teams were apparently not doing that is inexcusable.
In some cases, the lack of ad performance is not just channel-specific, but true across most channels and sometimes even across the board. For example, some purchases are triggered by emergencies, and that's where search channels outperform TV and social. And in other cases, your customers just cannot be reached by ads at all - eg: military contracts. I am sure we can come up with many more examples that seem to show that ads don't work. But what those examples really show is that there's a certain amount of minimum qualification to make ads work, and a lot of people simply lack those qualifications.
That's alright, a lot of people also lack experience in Golang, but at least they think twice before making all sorts of assertions about the viability of that programming language. Not so much when it comes to marketing topics - somehow there's this belief that if a certain ad channel didn't work for you with your specific value prop and your specific expertise in this field, this is somehow relevant to everyone else. I can see how newbies can make that mistake. But for Rand Fishkin or the CMOs of Ebay and Uber to fall into that trap... blows my mind.
[0] https://www.forbes.com/sites/augustinefou/2021/01/02/when-bi... [1] https://sparktoro.com/blog/something-is-rotten-in-online-adv...