There are a ton of gurus, books, courses, seminars, coaching programs etc. for marketers, all of them claiming to make you a genius, to deliver the All Pervasive, Hidden Secrets of the Marketing Universe.
A few of them are actually good. And I only endorse people whom I can fully recommend as head and shoulders above the rest of the crowd.
But there’s the old saying “you can’t learn to ride a bicycle at a seminar.” And it really doesn’t matter what aspect of life we’re talking about, there’s an enormous, huge difference between learning from a guru and actually doing it.
The real, bottom line truth about really talented marketers and anyone else comes down to one word.
Just one.
The word is: TESTING.
If you don’t test – if you don’t develop an idea, run it up the flagpole and see who salutes, with hard numbers and percentages – you really don’t know anything.
If you DO test – you know everything. And nobody can argue with you. If they do argue with you, you don’t care anyway.
That’s all there is to it.
The questions one can ask, or debate, are endless. Does long copy work or is short copy better? Should I have a sales letter website or a traditional one? Should I price this at $500 or $1000? Should I use a soft-sell approach or is hard-sell better?
The answer to every question is, test.
One of the reasons I quickly became obsessed with Google AdWords is the fact that you can test things with such speedy precision. Split test two ads and measure the response to a tenth of a percent. Split test two landing pages on your website and see which one gets more orders or leads. Turn the test on, and turn it off – instantly.
Beware: When you test, you need statistically significant results. Now we could get a math professor to wax eloquent about this, but essentially you need enough trials to have reliable numbers. If you flip a coin four times and it comes up heads all four times, it doesn’t mean that there’s no tail.
There’s still a 50% probability of being heads or tails, you just happened to get four heads in a row.
A simplified but useful rule for testing is: You need to get 30+ responses, under a consistent set of conditions, in order for your numbers to be accurate within 5%.
So if you’re split testing two pay-per-click ads on Google, one got 5 clicks and the other got 10, you really don’t know which one is better yet. The luck of the draw could be favoring the one. But when one gets 30 and the other gets 50, you’re pretty certain. Same is true with direct mail, or responses to a web page, or email, or whatever. At least 30 responses for each thing you test.
Testing used to be a lot harder than it is now. Read Scientific Advertising, written in 1918 by Claude Hopkins. He talks about taking months to run ads in newspapers, collect coupons from retail stores, and assess the success of campaigns. Now you can send out a few thousand emails, split between one offer and another, and you know within a few hours which offer is better.
Testing used to be very expensive and take weeks or months. Now it’s cheap and it takes hours or days. There’s no reason why you can’t know anything you want to know with inexpensive testing! And no reason to spend millions of dollars developing an idea that nobody will buy. Sales people look at rejection as failure, but direct marketers see it as only a test. Which way do you want it?
Testing weeds out those bad ideas, and leads you to the good ones. The answers will always surprise you.
Photo by JD Hancock cc by-sa
Share This Post
About the Author
Perry Marshall has launched two revolutions in sales and marketing. In Pay-Per-Click advertising, he pioneered best practices and wrote the world's best selling book on Google advertising. And he's driven the 80/20 Principle deeper than any other author, creating a new movement in business.
He is referenced across the Internet and by Harvard Business Review, The New York Times, INC and Forbes Magazine.
7 Comments on “The One True Way to Become a Marketing Genius”
I am still amazed with how little testing goes on. I feel every time I run a new campaign and tell the higher ups I will have data in a day or two, they look at me like I am a genius (which I am OK with my boss thinking that). I still work with the marketing club I help restart at my alma mater, and besides pushing networking skills, showing them how to design and run test I feel is the greatest advice I can give them. Thanks for the article.
It’s a basic discipline which, based on my last 15 years of experience, only 5% will ever implement in a rigorous way. Automatic advantage for you.
I don’t care how much you claim to know (esp. in marketing), if you don’t test, you don’t know anything.
YEP.
You are very clever. Thanks for this post.
Perry, but what about keywords? When I start a new campaign I try few hundreds of KW just to choose few of them with the best CTR. I cannot wait until each of them gets 30 clicks, especially when I see there are keywords with hundreds of impressions without any clicks.
So whats the rule of tumb for keywords? Maybe 300 impressions to see the real CTR?
Also dont forget that KW alone means nothing, it works only together with ads. And ad alone means also nothing, it works with KW and landing page. And conversion rate for landing page means nothing unless you know with which ad you get this conversion, etc. Its not easy to isolate the variables.
BTW, do you know the “explore and exploit” algorithm? Explore means testing and finding new ways, exploit means profiting from the results of testing. How many resources should companies alocate for exploring and how many for exploiting? Maybe 10%/90%? Dont forget that only 20% of tests will work and 80% will fail, so the most time and money for testing will be lost and the results from the 20% must pay for it.
PPC is a special case, much more testing than anywhere else, but still if you test 2 ads, 2 landing pages, several keywords all the time, it means you split 50/50 for exploring (the new ad) and exploiting (the old best ad). 80% of the new ads will fail, they will just decrease your CTR, quality score, increase CPC, etc. Is there not a better way how to split the exploring and exploiting in PPC?
John,
You can figure out an ad is a bad loser within a click or two. But winners take longer to declare.
Ideally you should spend 3-10% of your budget on exploration.
When you’ve got a great control ad, it’s hard to beat, so I make multiple copies of the same winner and test them against one incumbent. That way a bad new test doesn’t tank my CTR.