 |
      |
 |
of Your Ad or Website... 100% to 1,000%
If you have ever lost money from advertising, then you must read this step-by-step recipe for "split testing," which can help improve the response you get from advertising... 100% to 1,000%!
And, you must read about why you should be delighted with break-even profits (and how break-even profits can make you very, very wealthy).
Please read this entire article. I consider it to be one of the most important things on this website.
So, without further ado, here's...
A "Paint by the Numbers" Guide
to
Split Testing for Lawyers
Instead of defining the term "split testing," I'm going to just tell you how to do it. The reason is because you will more fully understand the significance of "split testing" if you just read how to do it. So, let's get started...
Here's how to split test:
Step 1: Create two versions of an advertisement (or direct mail letter, or Pay-Per-Click ad, or website, etc.). For simplicity's sake, let's name them Ad #1 and Ad #2.
Step 2: Insert different contact information in each advertisement. For example, you can display different telephone numbers in each advertisement, or you can display different telephone extension numbers in each advertisement.
By assigning a different method of contact for each of the two advertisements, you are creating "tracking codes" that allow you to track which advertisement performs better.
Step 3: Make sure that both Ad #1 and Ad #2 are each shown to an equal number of people. Here's how:
If you are using a newspaper or magazine display ad, you can ask the publisher to create what's called an "A/B" split; this means that you will give the publisher both advertisements... and half of the newspapers or magazines will be printed with Ad #1, and the other half will be printed with Ad #2.
Direct mail letters can be split tested by mailing two versions of a letter... and mailing an equal number of each letter.
Google Adwords can be split tested by using a built-in feature that lets you automatically rotate two different versions of an advertisement -- and each version can be shown an equal number of times.
Websites can be split tested using various commercially available software.
I hope you get the point. Almost anything can be split tested.
Step 4: Keep the winner!
The benefit of showing two different advertisements... is that you can "test" two ads against each other... and see which ad gets a better response.
After you discover the "winner" of your split test... discard the ad with the lower response, and keep the winner. The winner is what Madison Avenue advertising agencies call the "control" ad.
Step 5: Play "Beat the Control." Here's how:
Create a new ad to split test against the "control."
The reason why you want to create a new split test is because you want to try and see if you can get even better results.
Keep split testing until you discover a new "control."
And, after that, keep trying to play "beat the control."
By constantly split testing in this way, you will be constantly improving your response rate.
IMPORTANT NOTES
(And how to really profit from
advertising...)
Important Note #1: Statistically speaking, you will need to have at least 20 responses from each ad, before you can be fairly certain that you have a winner or loser.
Here's an example that shows why:
You may find that early on... Ad #1 receives 4 responses, while Ad #2 receives 7 responses. If you were an impatient tester, you might think that Ad #2 is the better ad. However, if you wait until the end of the test, you might discover that Ad #1 received 30 responses, while Ad #2 only received 20 responses.
If you didn't wait for statistically valid results, you would have made the "loser" ad the control ad!
The moral of the story is to wait until you get at least 20 responses from each ad, before you declare a "winner."
Important Note #2: You must test only one thing at a time. You cannot test two completely different ads. For example, you can test one headline against another, or one photograph against another, but not both at the same time.
The reason is because if you test 22 different things all at once, and you discover that Ad #2 performs better than Ad #1... you won't know WHY you got the improvement.
Stated another way: Your improvement will be "because of" some changes, and "in spite of" others. And normally, you can't tell which changes improve the ad, unless you change just one thing at a time.
Why is this important? Here's an example that shows why:
Let's say you ignored my advice and made two ads, each with a different headline, and each with a different photograph. Let's say that Ad #1 has the better headline of the two, and Ad #2 has the better photograph of the two. So, the best ad would be a combination of Ad #1's headline, and Ad #2's photograph.
However, let's say that you did not know this. You only know that Ad #2 outperformed Ad #1.
In other words, Ad #1's photograph was so
bad, that it negated the effect of its good headline... and Ad #2's
photograph was so good, it negated the effect of its bad headline...
In this scenario, you would have never guessed that you should combine Ad #1's headline with Ad #2's photograph to create... a super-effective ad!
(Actually, you can test more than one thing at a time, if you use "multi-variate" testing, but that is an advanced concept... for now, just keep things simple and limit yourself to testing only one thing at a time.)
Important Note #3: Have realistic expectations.
Some people think that they can get a 10% to 15% response rate the first time they advertise. This is nonsense.
It is not uncommon to receive only a 0.05% response rate (yes, that is one-half of one percent) the first time you split test... although you can improve your response rate to 4% or 5% -- sometimes even 19% or 20% -- if you continually split test.
Important Note #4: Most advertisements produce "break even" results, which means that you make about as much money as you spend on advertising. However, you must realize that most of the profits in any business (including the "law business")... is made on the "back end."
Here's a story of how one man -- named Vincent James -- made over $100 million in a few short years... by initially generating "break even" results... and then profiting from the "back end":
Mr. James sold vitamins and herbs. He mailed out a letter that sold a bottle of his magic herbs for the price of $59.95. His profit? He broke even.
So, he mailed out the letter again, this time adding a paragraph that stated... if people bought his product, then they would be automatically enrolled in his "auto-ship" club... and the customers would receive a new bottle each month... and they would be automatically billed for $59.95.
His response rate did not drop, nor did
it increase. He "broke even" on the initial sale... and then made
$59.95 every month for 4 months (most canceled the auto-shipment after 4
months).
From that break-even letter, he made $240 per person. It only cost him $2 per bottle to make the product, so his profits were $232 for each person who ordered.
This is how he made $100 million in a few short years.
(This is how West, Bancroft-Whitney, and other legal publishers really make their money... through pocket parts and CD-ROM updates. This is also how "case management software" companies also make their money -- through software updates and backend "technical support plans.")
The moral of the story? You will almost always get new clients at break-even... or even at a slight loss. (The Sharper Image Catalog loses $10 on each initial sale, because they know that they can make a huge profit on the "back end.")
Here are some ways to profit from the back end:
Your clients may need other legal services, or they might need non-legal services which you can sell (or get a referral fee on).
You can sell or rent your mailing list to others.
Future referrals from clients are a form of back end.
The back end can be a payment plan. For example, let's say you have a very complex estate planning case... and let's say you want to charge $50,000. If you structure a payment plan where the client pays over time (so that it will be easier for the client to pay... and so that the client can get increased tax savings), those payments over time are the back end.
You can put a client on a regular service plan -- for example, perhaps you can schedule quarterly "check ups" or "updates" for a client.
Use your imagination -- and try to get those back end profits.
If you can't think of any... and you're not making a good profit on the "front end," then perhaps you should avoid advertising -- and focus on other marketing activities instead.
Important Note #5: "Test small, and profit big." Here's what it means:
DO NOT spend large amounts of money on advertising until after you conduct one or more split tests... and get a control ad that you are happy with.
These split tests should be small tests, and they should be cheap to run.
After you find your control ad, THEN "roll out" with more frequent advertisements, larger advertisements, and/or media with larger circulations.
DO NOT test with a full page advertisement in the Wall Street Journal.
DO NOT test with a double-spread Yellow Pages ad.
DO NOT test your fancy brochure on millions of people.
Test small, then "roll out" big.
Important Note #6:
WARNING: "Roll outs" typically perform worse than the initial test. Here are some reasons why:
(a) Timing
People respond differently, depending on the time of the year, and the season. I don't know why this is -- it's just something that's been observed by large mail order catalog companies. Human beings act and think differently at different times.
(b) Non-Representative Tests
Your test may have been conducted on a group of people who were especially eager for your offer, or who were especially NOT eager for your offer.
For Example: Let's say you are testing direct mail, and you wanted to test a mailing list of buyers of exotic sports cars. (Let's pretend that you see an opportunity in "exotic sports car law.")
Let's assume that you wanted to test a list of 5,000 names, which is typically the minimum order. Being a smart tester, you ask the mailing list company for what's called an "nth name select" (for example, every 5th name on the list).
However, what the mailing list companies sometimes do... is give you "hotline names" instead of what you ordered. "Hotline names" in this example would probably be people who bought an exotic sports car within the last 90 days.
The reason why mailing list companies sometimes do this... is because they want you to believe that this is a very responsive list (hotline names are almost always the most responsive names)... and they want to entice you to buy more names when you "roll out."
And, you have no way of knowing whether or not the mailing list company is doing this.
In fact, in almost every test you do, you never know if your tests are being conducted on a representative sample.
(c) Changing Markets
Another thing to keep in mind is this: By the time you "roll out," the market could be different.
For Example: If your competitors saw your tests and "rolled out" copycat advertisements... then the market would be different.
Another Example: In California, when estate planning attorneys started giving free seminars on the benefits of Living Wills... it nearly wiped out the competition. They single-handedly changed the market for estate planning.
So, for all of these reasons (and more) -- the results of the "roll out" will almost always be different from the test... and the "roll out" will almost always get less of a response.
Statistics is fairly complicated, so I'm not going to get into a discussion of how to mathematically estimate the amount of difference you can expect on any given "roll out."
(However, if you are using direct mail, an easy way to estimate "differentials" is to use the free calculator here: http://edithroman.com/calculators/pieces.aspx If you play with the calculator for a little while, you'll very quickly understand how to use it.)
In any case, all you need to know about "differentials" is this: the response rate of the "roll out" will likely differ (for the worse) from the response rate of your tests.
So, the moral here is to "roll out" slowly... and to keep careful track of your response rates as you roll out.
Important Note #7: Split test many different things.
For example, if you are split testing direct mail...
Split test the letter, the envelope, the enclosed business cards, the mailing list -- and more! (Just remember -- don't split test everything at the same time -- see Important Note #2 above.)
There are many things that you can test in any given advertising campaign... make sure that you split test everything that is important.
(And when I say important, I mean just that -- don't waste time split testing things that are not important, such as the response rate of cream-colored paper vs. off-white paper.)
In Conclusion: If you want to make advertising work for you, then split test... and thoroughly understand the implications of both the "front end" and the "back end."
"If you are not split testing, then you are not advertising -- you are running a promotion."
Attorney at Law
Join
Our Fax List!
You'll receive our newsletter Rainmaker's Report, as well as
special announcements just for list members.
All faxes published by
Michael Lee.
|
|
