There’s a lot of confidence in performance marketing about what you should never do. Never use lead gen forms. Never run multiple text options. Never touch interest targeting or lookalikes now that it’s all about going broad. Whatever the current orthodoxy is, someone will tell you it’s a hard rule.
I don’t deal in absolutes.
Best practice is right most of the time
Not because the best-practice advice is wrong. Most of it is right, most of the time. But “most of the time” is doing a lot of work in that sentence. Every so often, breaking the rule is exactly what works:
- The account that responds to interest targeting when everyone said go broad.
- The lead gen form that converts when the playbook said send them to the site.
- The aggressive ad text testing that pays off.
The only way you know, for your account, in your industry, is to try it. An absolute rule robs you of the test before you’ve run it. It decides the answer for an account it’s never seen.
Take ad copy. The rule everyone repeats is keep it short. In crowdfunding I do the opposite, because long copy beats short copy when people are deciding whether to invest, and switching to it helped halve Tanica’s cost per EOI.
Why break rules? Because performance marketing cares only about performance. Not rules. And if the performance is good when rules are broken, then the question isn’t “did you stick to the rules?” It’s “what results did you get?”
Why the “wrong” way sometimes wins
I’ve taken over accounts from people who were, by any textbook measure, just winging it. No rulebook, no orthodoxy. And plenty of the time it was going pretty well. Sometimes a campaign structure that “shouldn’t” work does. Sometimes breaking the rules is the whole reason the account performs.
A lot of the time the reason is simple: the unconventional approach helps create ads that don’t look like ads. The rules optimise for looking like a competent advertiser. The results come from finding what works.
What I’ll tell you instead
So no, I won’t tell you I absolutely never do something. I’ll tell you what usually works, why, and that I’m willing to test anything with a good hypothesis.