#TBT: Experts Aren’t Always What They Seem and Other Lessons Theaters Need to Learn
Happy #TBT. To celebrate, we’re sharing an oldie-but-goodie post from Jim: Unforced Error of the Almost Unbelievable Kind.
First, I’d like to thank Ethan Siegel for telling this story. It takes a big person to cop to something like this, and I suspect he’s doing it because he’d like others to avoid such a terrible mistake. I don’t want anything I say to come across as a personal slam because, Lord knows, I’ve made my share of mistakes as an entrepreneur.
Just none of quite this nature or sudden impact.
In 2007, Ethan let Google ‘optimize’ his Ad Words campaign, which, according to him, makes up “90% of [their] ad budget and generates about 90% of [their] sales.”
Let me start by breaking out the yellow highlighter and calling attention to that number: 90%. Ninety fargin’ percent.
As you’ll read, not only did Google’s changes not help, they crashed sales 30%, overnight, while driving up costs 30% at the same time.
Bummer. Those are big numbers. In fact, they’re kill-a-company kind of numbers. But it got worse because there was no way to automatically roll back, and after Siegel painstakingly re-created his previous campaigns, he learned that he lost all the ‘juice’ his ads had earned over the years when it was making 90% of his company’s revenue.
So where do I start with what to make of this? Don’t do any of these things:
• Rely solely on an asset you don’t control and can’t even influence, for whom you are a drop in the bucket. Remember that 90% sales number? Too much for one source. That’s why investment bankers diversify portfolios, because what if your Enron stock suddenly stops doubling every year?
• Fail to understand the tools well enough to know how they really work. This is truly an unforced error. As a marketer (or really, anyone in business), before undertaking something like this, you should ask a lot of questions about the mechanics. I used to click links in test versions of email campaigns I was doing as a young marketer, almost obsessively. Does it work? Does it still work? Actually, I still do that sometimes.
• Not test with a small subset of audience before rolling out to everyone. This was a test, but you don’t test to your whole audience. Why not break off 5 or 10% of your ads and try it there? Then, whatever happens, the worst-case scenario is that you lose 5 or 10%. If it works, you’ve lost absolutely nothing and can roll out everyone immediately after.
• Trust that some ‘expert’ knows better than you with no information to demonstrate that. I don’t want to embarrass anyone, but we recently did a test where a recognized expert in a certain kind of web design gave us an ‘optimized’ version of something that we tested against both our current version and another version that a marketing director whipped up in PowerPoint in about an hour. (And he’s no designer.) The expert version, in a scientifically sound test, got whipped by both. Hey, not their fault, but just because someone is an expert doesn’t mean they know what to do with your business. In my experience, 4 out of 5 ‘experts’ are what I would call well-read laymen who can’t point to a single demonstrable relevant result they’ve ever achieved.
• Expose your company’s single point of failure to risk with no back-up plan. It’s bad enough to have a single point of failure. It’s Evel Knievel-crazy to expose that single point of failure to wild risks. Astronauts don’t test the durability of their spacewalk lifelines by sawing on them with a box cutter while floating around the International Space Station. At least, not without a second lifeline in place!
• Change any major process without the records and ability to roll back to the previous one. The Army had horses AND tanks for a while for a reason. Tally ho!
and on a slightly more topical note …
• Fall prey to the misconception that Google Ad Words equals marketing. The world’s a big place. Marketers who’ve grown up in the last decade sometimes get a little Google-blind. Don’t.
Again, I think Ethan Siegel is a saint for publishing this, and my purpose here is not to denigrate, but hopefully, to help achieve what I assume is his purpose, which is to teach others and help them avoid this kind of mistake. If we ever bump into each other, Ethan, the first drink is on me!