The Profit DNA: How To Unlock The Code To Maximum Profits Through Self-Evolving Websites
Key Terms:
Advertising, Testing, Conversion Optimization, Branding
Advertising, Testing, Conversion Optimization, Branding
The Self-Evolving Website
Did you know that there’s one change you can make to your website right now that’ll take you only minutes to do, but could increase your sales and conversions by 300% or more?
In fact, there are likely a lot of little tweaks here and there that could have a profound positive impact on your profits. In this case small hinges do swing big doors.
But knowing what to change, as well as what to change it to, is often the challenge. For instance, as a direct response copywriter who has written sales letters producing millions of dollars for top marketers, both online and off, I know that changing your headline has the greatest potential for boosting your sales, because many more people read the headline over the body sales copy. In fact, legendary adman David Olgilvy has said that five times as many people read the headline as they do the sales copy on average.
Your headline is quite literally an ad for your ad, so it had better do its job, or your sales message won’t even be read. And if they don’t read it, they certainly won’t buy your product or service.
You don’t want to be clever in your headline, and in most cases you don’t want to try to sell them on your product or service at that point. Instead, you want your headline to “sell” them on reading your sales letter, where it will do the selling.
And what about your sales letter itself? You probably know that a targeted message is more likely to be read and acted upon than a blanketed generic one. With all the different traffic sources sending visitors to your site, how can you make sure your sales message is relevant for each of them?
Wouldn’t it be easy and more profitable if your website was able to figure out all on its own the most targeted effective sales message, customized for each traffic source? And all based on the criteria you specify: more sales, more leads, whatever your goals may be.
That’s the basis of this report. I’m going to show you just how to set up your own self-evolving website on autopilot that’s designed to be as hands-off and automatic as possible. You simply define how you want it to evolve, then let it do its thing.
And best of all, it’ll be your visitors and your customers that’ll decide what’s most important to them. They’ll vote with their credit cards, and your website will optimize itself, evolving over time to be the perfect selling machine. No more guessing.
Along the way, you’ll also find out how much each visitor, lead, and customer is worth to you, taking the guesswork out of how much your business can afford to spend acquiring fresh leads.
And before you’re done reading this report, you’re going to seriously start thinking: “How fast can I spend a million dollars?” I’m not kidding. You’ll see what I mean shortly.
Now let’s talk a bit about what I mean by self-evolving.
I’m sure you’re familiar with Charles Darwin’s book, On the Origin of Species, where he introduced his theory of evolution. Originally published in 1859, his controversial work told us that species on earth have evolved over the course of generations through a process of natural selection. That is, the species that adapt and survive continue to spawn new generations, while those that fail to adapt die out of existence. When a male and female produce offspring, their DNA, the building blocks of life, are combined to form an entirely new sequence of genes.
We’ve come to understand genes as the “hidden code” to all of our hereditary traits, from the color of our eyes and hair, to how predisposed we are towards certain diseases and even abilities. In other words, the science of genetics.
Next, the new offspring combine their DNA with their mates’, producing the next generation of offspring, and the process continues.
Of course, that’s a simplified version of what happens, according to Darwin. I’m not here to address any theological considerations. For the sake of this report, I’m only concerned with how it’s supposed to work scientifically.
(Perhaps you’re thinking, what the heck does this have to do with marketing? Hang on, I’m getting to that…)
Now, much of what I just discussed is undoubtedly nothing new to you. However, what you may or may not know is that the concept of evolution can apply to other things as well. For example, with the emergence of faster and more powerful supercomputers, mathematicians have begun to experiment with genetic algorithms to solve complex problems. Speculators have used them to try to predict future market indicators based on past results, by
trying to filter out the “noise.” Genetic algorithms are designed to mimic evolution. That is, mathematical equations are paired with one-another to form new equations, and the ones that come closer to solving the problem at hand are allowed to “survive,” while the dead-end equations die out. The process is repeated through each “generation,” producing more offspring equations, which then are combined at random to produce the next generation. Occasionally, as in real life, random mutations are introduced into the equations to allow for
new variations to spawn. Given enough time and generations, the equations evolve into the most ideal solution to the problem. That’s because only the fittest survive.
Perhaps you’re beginning to see where I’m going with all of this. Yes, it is possible to take a similar approach with a sales letter on a website.
Here it is in a nutshell:
1) The problem to solve is identified (in our case more sales).
2) A mechanism is introduced to randomly combine elements on a sales page and serve them to visitors.
3) That mechanism lets the least profitable combinations die out, while the fittest (i.e. the
ones generating the most sales, or whatever our goal is) survive.
4) The process is repeated as often as necessary to produce the most profitable combination of elements.
5) New variations (i.e. new copy, layout changes, etc.) are introduced periodically to allow fresh ideas into the mix.
When you think about it, I’m not introducing anything really new here. Smart marketers have already been using these principles for quite some time now, even though they may not have thought about their websites as being self-evolving.
But the big picture I want you to see is: simple A/B testing, or even multivariate testing (both which I’ll cover ahead) are good starting points, but there are additional strategies you can implement to really take your business to the next level.
I want you to think about your website as an opportunity. An opportunity for you to do a little bit of forward thinking and upfront setup, in which your website will ultimately take the ball and run with your vision. In short, your website will learn how to sell most effectively, with a bit of input and prodding by you. And the best part about this approach is it doesn’t require any software that doesn’t already exist (in fact, you may be using what you’ll need already). If it all sounds complicated and like too much work, relax. It’ll become clearer to you as I run through it step-by-step right here in this report. And your testing software will be doing the brunt of the work. You just need to aim it in the right direction and let it do its thing.
Now let’s dive right in and turn your website into an intelligent evolving profit engine.
Taming the Money Pit
Advertising is expensive. On top of the expense, it’s often a big waste if it doesn’t pay for itself. How many times have you spent hundreds or thousands of dollars to advertise your business, only to see that money flushed down the toilet? Maybe you’ve spent a mint on an ad in the hopes that you’ll see a flood of new business? And when only two or three people respond, it sure seems like a waste.
More than 89% of all businesses believe advertising is important. Yet 75% of them also admit they are unsure how to make their ads work consistently or how to raise their response rates.
And to make matters worse, often the big advertising agencies are the worst culprits. They often make decisions about the content of ads for their clients based on their desire to win advertising awards. Why? Because it positions them as the “premier” agency for big firms to seek out.
In other words, it’s good for them, not necessarily their clients. I’m not saying all advertising agencies fall into this trap, just a lot of them. Want proof?
Remember those Superbowl ads just before the dot com bust? Were they entertaining? You bet! Did people everywhere stop what they were doing to watch them? Sure! Tons of eyeballs on those ads. Were they effective? Nope. Not by a long shot. Just look at the results. Look, those commercials are one example, but I've seen it countless times. Remember in the 90's AT&T had their million dollar slogan: "The Right Choice." Would you or I buy from AT&T because they told us THEY were the right choice? Not me. Not you. Not anyone. However, when we learned they had great prices and great value, and their unique selling proposition set them apart from their
competitors, well then, now they are onto something. Friendly operator assistance. Money back guarantee. No charge to switch. Fast service. Clear calls. Low prices. You get the idea.
In fact, the US Army just paid $200 million to a marketing firm to come up with their new slogan: Army Strong. That’s $100 million a word! Will it be worth it? Time will tell, depending on what they do with it. But I’m willing to bet if they took that $200 million and invested it in direct response advertising, they would have gotten a LOT more bang for their buck.
Can you imagine if you dumped a ton of money into an ad agency and they produced results for you like the AT&T slogan, the Army slogan, or those ineffective Superbowl ads? Well, hopefully you didn't do that, but if you did, there is still hope. There is a completely effective way to market your business and get real results.
Direct Response Versus Branding
There are primarily two schools of thought on advertising: direct response and branding/brand-building.
Direct response ads are designed to get prospects to take some specific action right there and then. Because if they put aside the ad and do something else, the chances are much smaller that they’ll come back later and buy. With direct response advertising, you’ll frequently see scarcity selling tactics used to get them to act immediately.
“Call within the next 10 minutes…”
“Only the next 200 people to sign-up will qualify…”
“The price goes up this weekend…”
You get the idea. You’ll also find a clear call to action in direct response ads.
“Register now to start your membership immediately.”
“Simply check off which report you’d like and return the postage-paid postcard today.”
The branding (also called the “open house,” or brand-building) model of advertising, on the other hand, is generally used by large corporations to create an awareness of their brand, so when the time comes to buy that type of product or service, the prospect will think of them first.
Slogans, jingles, and the like are frequently used with branding ads for specifically that purpose. Branding, however, is expensive, time-consuming, and requires a lot of brand equity and trust over time before people make decisions to buy from them. Large corporations, for example, typically spend millions of dollars in advertising costs in order to blanket their target audience with as much exposure as possible.
In general, advertisers that focus on branding are never certain how much return on
investment they get with their ads. There are too many factors to consider, and there’s no
reliable tracking mechanism in place.
Direct response advertisers, on the other hand, can usually track sales from any given ad to the
penny. Whether it’s a coupon to be clipped, a special phone number to call, or a business reply
card, everything is coded and tracked.
Brand advertising typically uses 30-second television commercials. Direct response uses two minutes or longer (and with infomercials, 30 minutes or longer).
Brand advertisers typically broadcast their commercials during expensive prime time, where
they can reach the most people. Direct response advertisers have learned through testing that
they make more sales late at night.
Brand advertisers tend to use shorter sales copy, while direct response companies use long
copy.
They both have their place. That is, neither one of them is right or wrong. They’re just different
in their approach.
Neither are they mutually exclusive. I personally believe that any company advertising to build
their brand should also find a way to do direct response advertising as well.
This white paper won’t address branding-type advertising, as I’m assuming you don’t have
millions of dollars lying around to spend on media costs and expensive ad agencies.
In fact, even if you did, you would likely get a bigger piece of the pie if you focused on direct
response marketing and built your brand solely as a “by-product.”
Enough said about this topic. I only wanted to distinguish between the two, because we’ll be
talking about the latter of them, direct response.
Brand-builders hope you’ll remember them when you need what they sell.
Why Testing Is Important To Your Business
Imagine you had two different ads created for you. How do you know which one is likely to
produce the better response?
In 1886, retail businessman John Wanamaker famously quipped, "I know that 50% of my
advertising is wasted. I just don't know which half." But the truth is much worse. In fact, most
advertising is wasted, because most advertising is not based on what’s been proven to work.
So to go back to my earlier question: how do you know which ad will perform better?
To answer that question, we must first have a history of test results. That is, we must have a
pretty good idea as to what types of elements in an ad produce the best response, based on
real-world tests. Not what you think or I think what will do best. Not your mother or next-door
neighbor. Not even the million dollar ad agencies. In fact, nobody would really know what to
focus on, UNLESS they had access to these proven test results.
And in order to get these test results, there would have to be some way to specifically measure
the results of any given ad. We would need to know which types of elements performed well
and which did not. Then we could analyze the ads and determine the common techniques used
among the successful ones.
Well, fortunately much of this work has been already done for you. There are over a hundred
years of history on advertising test results in which you can base your assumptions. So if you
knew which kinds of ads performed better, then you might be able to take an educated guess
as to which of the two ads would perform better than the other.
Of course, that kind of information is just a “starting point.” You still need to test your own ads.
I’m going to show you how to test your own sales letter the right way, so you can cut the dogs
loose and go with the winners every time. And ultimately we’re going to look at how to make
your website “learn” to show the most profitable combination of elements that make up a sales
letter.
No matter how successful your sales letter is, I can’t stress the importance enough of testing it
to improve it even further. Sometimes the tiniest tweak can be responsible for boosting a sales
letter’s conversion rate considerably. But how will you ever find out which tweaks those are
unless you test them?
On the Copywriting Board and other forums that discuss copywriting, people are always posting
their websites, asking for critiques. Sometimes it’s fairly obvious what they could do to improve
their response. Other times what we may think will work better can actually make things worse.
In the end it doesn’t matter what you or I or other copywriters think. It only matters what your
prospects think, and they weigh in on how much they like your headline or offer by purchasing
and becoming a customer.
Remember, the name of the game is to always try to beat your control (your best-performing
sales letter). Once you have a proven sales letter, it truly is money in the bank. When you know
your average response rate, you can predict what getting more of the same kinds of prospects
to your website will do to your bottom line. It’s the equivalent of printing money on demand.
When you get right down to it, who wouldn’t want their average conversion rate to be as high
as possible? I don’t know of anyone who wants a lower conversion rate, but I do know plenty of
people who don’t test. Whether they realize it or not, they are settling for less than they’re
capable of achieving.
How often have you gone to a marketing or copywriting forum and heard
someone talk about testing?
How does my headline look? Is this a good price to sell my widget? What kind of font should I
use? What’s the best ending digit for a price? Should I send them to an opt-in page first, or send
them straight to my sales letter? Will a picture of me help to add more credibility and lead to
more sales? Should I use a header graphic?
And what’s the answer that marketing veterans and novices alike always give in response?
You guessed right! “Test it.”
And while that answer is probably the right one for those types of questions, it’s amazing to me
how many people know about testing (in theory anyway) and still don’t practice what they
preach. Still others know they should test but have no idea how to do it properly, even if they
think they know.
The nice thing about testing is that it doesn’t require a lot of marketing know-how to do it.
Anyone with a few basic tools and a solid understanding of the basics can run circles around
some so-called marketing “gurus” when it comes to testing. I hate to use the ‘ol cliché, but it
definitely ain’t rocket science.
So let’s start by going over some of the basics.
The two main goals of testing are to…
1) Spend less money to acquire customers
2) Make more money from the customers that you do get
Seems straightforward, right?
It can be summarized even further, by stating that the
primary objective of testing is to learn how to maximize
your return on investment (ROI).
Think about it. ROI doesn't just measure your sales
copy's effectiveness. It encapsulates your market, your
offer, and your copy.
As a meaningful statistic, conversion rate by itself
means very little. For example, to say that my sales
letter converts at 8% means nothing by itself, unless
you’re comparing it to your old conversion rate. If it’s
gone up, that’s good news, but it could double and that
still may only mean an extra $5 a day (assuming you
were making only $5/day to begin with).
Another figure tossed about is traffic. "I get 100,000
hits a month." My answer: "That's great, but what is
your ROI? Are you losing money or making money?"
You certainly can't tell from the number of hits.
In fact, expert copywriter Dr. Harlan Kilstein jokingly
says “hits” stands for How Idiots Track Success!
Kidding aside though, if I pay six figures to get traffic,
I'm going to get traffic. But does it pay for itself? Does it
sell? Is my ROI worth it? That's the key question.
So what does this all have to do with testing, you ask?
Plenty.
What if sales letter "A" gives a 5% response, meaning
5% of the people buy. And sales letter "B" gives a 9%
response. Comparing the two would indicate that letter
"B" wins, right?
But if you measure your lifetime customer value, you
may find that letter "A" yields a better ROI in the long
run than letter "B", when you factor in back-end offers and such. All things being equal, that's
unlikely to happen unless the offer differs between the two sales letters. But it's just a point to
illustrate the kind of thinking that needs to go into testing.
The 3 Factors Affecting
Response Rate
You can hire the best copywriter
in the world to write your sales
letter, but if your target audience
is chiropractors and you mail your
letter to lawyers, it’s going to
perform poorly, no matter how
brilliant the copy.
The three things, generally in order of importance, that contribute toward your response rate are:
1) Your market
2) Your offer
3) Your copy
So if you mail your letter to the
wrong list, it doesn’t matter how
great your offer or your copy.
If you mail your letter to the right
list, but the offer is weak, your
copy is only going to go so far.
But if your market and offer are
both right, then focus on the copy to boost response.
Too many people just split test the front-end offer (split testing is also known as A/B testing,
like in the example above). If I'm split testing the price (offer), I may find out that I sell less with
a higher price point on the front-end, but my back-end sales increase as a result, ultimately
giving me a higher lifetime ROI. In fact, some sellers purposely set the initial price higher to
weed out the tire-kickers.
Others purposely set their front-end price lower, because they want to build their list of leads
more. It all depends on your marketing plan, and your long-term ROI plays a huge factor in
determining whether it's successful or not.
And even if you’re just testing the front-end offer (or you only have one product to sell—in that
case, better get busy creating a back-end product!), a 5% conversion of test subject “A” netting
you $100 in profit per sale may be better than a 9% conversion of test subject “B”, which only
nets you $10 in profit per sale, even though test subject “B” has the higher conversion rate.
The point is you need to consider all of this. And you need to know your website’s “visitor
value,” which I’ll cover in just a moment.
Ok, so what about split-testing tools?
There are a number of good tools available, some more robust and feature-rich than others.
You can spend a lot on testing software, or in some cases you can even find it for free if you
know where to look.
In fact, my Copywriter’s Toolkit comes with free split testing software you can use “right out of
the box” to start boosting your sales and conversions, in many cases by huge margins.
That’s because a simple headline change can often double or triple your conversions alone. It
takes some time to write out a bunch of headlines to test (having a headline swipe file can
help), but it’s time well spent.
Once you have 2 or 3 good headlines to test, it only takes minutes to post it live and start
measuring their effectiveness with the right software. The testing software in the Copywriter’s
Toolkit is designed to be easy to install, easy to use, but powerful and precise.
After all, you can always add more bells and whistles later on. You probably already know the
importance of testing. Many people do. But how often do they actually do it?
Not nearly as often as they should. In fact, I suspect a big reason is because they just don’t have
their website set up correctly for testing. Well, here’s your chance to set it up once and be
ready to test profit-pulling headlines and other elements for life.
Multivariate Testing
e’ve talked about split testing, now we’re going to discuss multivariate testing,
sometimes known as Taguchi testing (named after Dr. Genichi Taguchi, who
pioneered it—interestingly enough, Dr. Taguchi applied these concepts to
engineering and manufacturing productivity, which were later applied to advertising).
In split testing, you have two or more different ads that are served to a segment of your
market, generally each with a single variance from one another. The testing of a single variance,
or factor, is also known as One Factor at a Time testing (OFAT), and has been around since
Claude Hopkins first drew attention to the idea in his classic book Scientific Advertising.
Sometimes, however, the ads are completely different from one another (for example, when
challenging a control).
With multivariate testing, you are testing several variables at once, and a mathematical
algorithm extrapolates the variables that are the most profitable (as you can see, we’re getting
closer to our goal of a self-evolving website).
In this manner, you can test multiple headlines, leads, offers, guarantees, and whatever you
want to test on your website simultaneously, and let the software determine which ones
provide the greatest visitor value. Then those winning elements can be shown more frequently
(i.e. the fittest survive) until the non-profitable ones eventually are phased out completely.
What you end up with is a sales letter that ultimately optimizes itself (evolves) based on the
actions your prospects take.
Before diving further into multivariate testing, it’s important to note what is meant by “visitor
value.”
Let’s say you sell a widget for $100.
If 100 people came to your website,
and only one person out of those
100 purchased your widget, then
you would have a 1% conversion
rate. And each visitor would be
worth $1 to you ($100 per sale / 100
visitors to make a sale = $1 visitor
value for each person). That’s a
simplistic explanation, because you
need to also consider your
advertising, fulfillment, and other
costs. A more realistic example would be if after all your costs, your widget netted you $100 in profit. Then at a 1% conversion rate, each visitor would be worth $1 in profit to you.
In other words, the visitor value is dollar amount each visitor is worth to you.
Double your conversion rate to 2% in the example above, then your visitor value also doubles
to $2.00 (for every 100 visitors, 2 of them buy x $100 profit per sale = $200 profits total—now
divide $200 by 100 visitors and you end up with a visitor value of $2.00).
So your visitor value is a useful number to know.
It’s also important to know your cost per visitor, cost per lead, and cost per sale.
Let’s look at a simple chart with some example figures showing how to determine each of
these:
1 Cost per visitor $1.00 $2.00
2 % opt-in 10% 25%
3 # visitors required to get one lead
(100% / % opt-in)
10 4
4 Cost per lead
(# visitors required to get one lead x Cost per visitor)
$10.00 $8.00
5 % of leads who eventually purchase 3% 8%
6 # leads required to get one sale
(100% / % of leads who eventually purchase)
33.3 12.5
7 Cost per sale
(# leads required to get one sale x Cost per lead)
$333.33 $100.00
Figure 1
The highlighted figures are numbers I made up for this example, but they also point out which
numbers we can control to some degree, either directly or indirectly. With the exception of the
first values (cost per visitor), the highlighted numbers are the ones we can improve through
testing.
The chart above in Figure 1 shows two different fictional traffic sources. The first source costs
you $1.00 per visitor and the second costs you $2.00 per visitor.
The second line (% opt-in) is the percentage of those visitors who give you their name and
contact information for follow-up. That is, the percentage of those visitors who become leads.
In this example, note that 2 ½ times the number of visitors (25% of the traffic) become leads
with the $2.00 traffic source as they do with the $1.00 one (10% of the traffic).
Line 3 shows you the number of visitors that are required to gain one lead. The $1.00 traffic
source converts to leads at 10%, so 10 visitors
(100% / 10%) from that source are required
before one lead is gained. The $2.00 source only
requires 4 visitors (100% / 25%) to gain one lead.
In line 4, we simply multiply the number of
visitors required to get a lead (line 3) by the cost
per visitor (line 1). In our fictional example, we
can see that the more expensive traffic source
($2.00) actually costs us less to gain a lead
($8.00) than the cheaper $1.00 source (which
costs us 10.00 to gain a lead).
Line 5 shows the percentage of leads gained that
eventually purchase our product. The values I
supplied there are fictional examples. In real life,
you would replace these values with the actual
conversion rates of your leads to customers. This
is not your website conversion (which would be
the conversion percentage of total visitors to
customers).
From there we can now determine the number
of leads that are required to make a sale (line 6)
and what it costs us to make one sale from each
of these traffic sources (line 7).
I’m illustrating the math here to show you how
the process works, but in most cases your
tracking software will do the math for you.
With a good marketing funnel in place, you
ideally want to match each message you portray
to your market’s desires. That means everything
from the initial pay per click (PPC) ad, article,
ezine ad, or whatever method you use to funnel
traffic to your site—all the way on down the line.
Everything.
What Is A
“Marketing Funnel?”
With the “marketing funnel” model, a
person makes a small purchase or
supplies their contact information, and
over time they are “funneled” towards
more and more high-end products and
services, step-by-step, by selling them to
the next level.
So by “funneling” (some call it
“backending” or “up-selling”—Dan
Kennedy calls it "gathering the herd")
prospects into paying customers, you’re
setting the stage to provide tremendous
value to them. So much value, in fact,
that your customers begin to look
forward to receiving content from you.
And with that value comes the
opportunity to take your customer to the
next level, where you can sell higher-end
goods to them.
And this isn’t a one-sided benefit. Both
you and your customer benefit by this
relationship. Your customer benefits
when he gets even more value…
something he really wants. You’re
helping him in that regard. And of course
you benefit as well by slowly graduating
your customer to your “A” list, where
you can provide even more value.
Your opt-in/name squeeze page
Your email autoresponder sequence
Your audio and video podcasts
Your teleseminars
And certainly your sales page
You see, the real purpose of testing is to
develop the most profitable end-to-end
marketing funnel. So by serving up one version
of a PPC ad/squeeze page/sales letter combo
to one group of prospects, and a completely
different version to another (depending on
their wants and needs), you create a more
targeted message, and your ROI climbs.
So it’s not just about conversion. Look at the
chart in Figure 1 one more time. It’s about
improving those numbers that you have some
degree of control over, either directly or
indirectly. The highlighted bolded numbers.
Given the examples in that chart, I’d choose to
spend my advertising dollars in the $2 per
visitor source over the $1 per visitor one. Even
though I’ll get fewer visitors, look at the
results. My cost per lead, and more
importantly, my cost per sale are both cheaper.
And if I took it even further, maybe I’d find out
that the lower cost per sale actually gives me a
lower ROI when back-end sales are factored in
as well. Maybe the extra money spent
acquiring each sale is worth it in the long run.
But I’d never know that information if I
didn’t test.
So unless I get new information telling me otherwise, I’ll stick with the $2 visitor source. And
once that dries up, only then would I move to my next-highest profitable traffic source (the $1
per visitor in this example).
What Is A
“Marketing Funnel?”
(Continued)
Notice how the width of the funnel gets
smaller towards the bottom? The width
represents the number of customers at that
height, or stage, of the funnel.
However, the smaller the width, the more
money they are spending with you. In fact,
the amount of money they spend with you
can be thought of as being inversely
proportionate to the width of the funnel
(more or less).
So the 20 percent responsible for 80 percent
of your profits are at the bottom of the
funnel. The other 80 percent that give you
20 percent of your profits are towards the
top.
This distribution is a general observation
and not a mathematical absolute. It might
be 70/30 or 90/10 or somewhere in
between.
This is no accident. Your “A” customers,
your biggest advocates, are in the smallest
segment of your customer base…the
bottom of the funnel (but the top in terms of
the value you deliver to them).
How Fast Can You Spend A Million Dollars?
he Ad Budget Myth: Companies who think in terms of an advertising budget are missing
the concept of return on investment. If you were to perform the analysis in the chart
shown in Figure 1 on your website, you would know how much you can spend to
profitably acquire a visitor, lead, and customer. And once you get a system in place where you
have a positive ROI, you should be thinking, “How fast can I spend a million dollars?”
That is, if for every dollar you spend you get two or three or four back, why would you want to
limit yourself on spending? You should be spending as much capital as you have available,
because you’re only going to get more back. It is quite literally selling five-dollar bills for ten and
twenty-dollar bills. Only a business not thinking in the right mindset of abundance (and financial
common sense) would limit their advertising budget under those conditions.
However, since too many businesses don’t test,
don’t know the ROI of their ads (matched to each
traffic source), or couldn’t tell you their visitor
value, they have no idea how much they can
profitably spend. So the advertising budget is
usually an arbitrary figure that’s put into place,
based on guesses and assumptions that have little
bearing on the bottom line. And as a result, it limits their potential for growth.
Another metric you should know is your customer’s total lifetime value. That is, the average
amount your customer spends with you over his or her lifetime. Because that will tell you how
much you can spend acquiring that customer. Obviously you want to do better than break even.
But to even know that break-even point, you need to know the total lifetime value of your
customer.
That figure should also be matched to your marketing funnel, from where you get your traffic
to the methods you use to acquire them as a lead, the steps to sell them on a front-end sale,
and all back-end marketing. All of it. For life.
When you look at it that way, the visitor value metric explained above is rather simplistic,
because the calculations I gave only factor in the front-end sale. You need to factor in
everything.
If that sounds rather daunting, relax. It’s always better to start somewhere, and if you get the
metrics in that chart down, you’ll be ahead of most of the pack. You can always expand your
tracking later on.
The key here, however, is to gain an understanding of where to spend your money. If you don’t
test, you’ll only be guessing.
You have a certain amount of money to spend on traffic. You want to spend it as wisely as you
can. You want to get the highest ROI on that money. That’s the bottom line. And it really
doesn’t matter if your site converts at 25% or 0.05% to do so.
Perhaps this is nothing new to you, but let me ask you this: do you conduct this kind of testing
now? Do you really? Right now could you tell me…
Your cost per visitor
Your cost per lead
Your cost per sale
Your visitor value
Your lead value
Your customer lifetime value
Whoa! What’s that I threw in there? Your lead value?
Yes, just like your visitor value and your customer value, you determine the value of your leads
by taking your adjusted income (the amount you net after paying for the traffic) and dividing it
by the number of leads you acquire.
So let’s recap…
Your visitor value is your adjusted income divided by the number of visitors
Your lead value is your adjusted income divided by the number of leads
Your customer value is your adjusted income divided by the number of customers
Your cost per visitor is your total number of visitors from a source divided by the total
cost for that source
Your cost per lead is the number of visitors required to acquire one lead multiplied by
your cost per visitor
Your cost per sale, or cost per customer, is the number of leads required to get one sale
multiplied by your cost per lead
So ultimately you’ll want to track all of the above, and to do that, you’ll also need to track your
conversion ratio.
And once you know those figures, the way to improve them is through testing.
There are many tools that will track those figures for you. The Copywriter’s Toolkit shows you
some of them, where they are, and where to start.
Now let’s get back to multivariate testing. As I
mentioned earlier, multivariate testing allows you to
test many elements of a sales letter simultaneously.
Why use multivariate testing over split testing?
Primarily for 4 reasons:
1) Split testing one element at a time makes it
more of a “sequential” process, as you start by
testing the elements that you suspect have the
greatest impact on your conversion ratio (e.g.
your headline) and work your way down to the
elements with a lesser impact. That process
can take some time as you find you have to
test more and more elements. With a lot of
elements to test, that time can increase
exponentially.
2) More often than not, you’ll discover that your
original control ends up having the higher
conversion rate over the challenger, which can
negate your time and effort invested in the
test.
3) Split testing is more of a "hands-on" approach
that requires you to constantly end a test and
start a new one with a different element being
tested each time. In addition, the conversion
rate changes you'll experience from each split
test can be comparatively smaller than
multivariate testing, since you're testing one
factor at a time instead of many at once.
4) By its very nature, split testing doesn't test the
interactions between the multiple elements
being tested. As a result, it won't reveal any
"hidden" gains produces by certain element
combinations. Multivariate testing, on the
other hand, has shown that such interactions
do exist and can affect the overall conversion.
With multivariate testing, you can put many elements
to the test at once and let your market decide which
It’s Really About Saving
Time And Money
Adding more elements and
variations can increase the number
of split tests required exponentially.
For example, let’s say you have 8
different elements on a sales page.
And let’s say each of those
elements has 3 variations.
Let’s do the math: 8 elements
with 3 variations each is actually
38 , or 6,561 combinations to test
(if you wanted to measure the
interactivity between each
variation). That’s just not practical.
Nobody will run that many split
tests.
Yes, you could test each of the
eight independently and miss out
on the gains made from the
interactivity. And you can also just
test the one or two elements likely
to give the greatest gains and be
done with it as well.
But if you really want to squeeze
every last bit of performance out of
your letter, multivariate testing is
the way to go.
With multivariate testing, you
don’t have to test that many
combinations.
That’s because it’s not necessary
(or practical) to shoot for a 100%
confidence level.
For example, let’s say your tests
reveal a conversion rate of
8% +/- 1% at a 90% confidence
level (meaning 9 times out of 10
you’ll get between 7 and 9 percent
conversion). 90% may be an
acceptable confidence level for you.
ones are the winners. In this manner, your sales page will truly “evolve” over time to produce
the one with the highest conversion rate.
There are a number of products on the market that will perform multivariate tesing.
In fact, Google has released a beta of their own tool, the Website Optimizer. Although the tool
is designed for users of Google Adwords (and is integrated with Adwords), the tool tests all
traffic to your site, regardless of the source.
Google Website Optimizer
http://services.google.com/websiteoptimizer
The tool is still in beta testing, but is now open to the public. It’s free to apply and to use the
tool, and the only requisite is that you have a Google Adwords account.
If you’d like to learn more about multivariate testing, I recommend David Bullock’s site, shown
below. David has a variety of resources to help you make the most of your testing, and he helps
to boil down some of the more complex aspects of the algorithms down into manageable bites.
David Bullock’s Site - davidbullock.com
On one of the videos in the Copywriter’s Toolkit, I walk through a demonstration of using
multivariate testing, and I also point out some useful resources to gauging your results. For
instance, if you use a “bare-bones” multivariate testing tool, you may find Blair Gorman's
Taguchi-Based Ad Optimizer tool useful (see below).
Blair Gorman’s AdComparator - adcomparator.com
Another good resource for learning more about multivariate testing, as well as seeing others’
actual test results is the ‘ol trusty Copywriting Board.
The search feature at that forum (and generally any forum) is one of
the most useful ways of finding the specific information you’re
looking for on demand. Chances are it’s already been discussed in
length at some point, so before posting your questions, try the
search feature first.
The Advanced Search will usually allow you to drill-down on the most
relevant topics posted, and you can segment your search by date range, person who
contributed to the post, the forum the post was made in, and much more.
For example, a quick search on the Copywriting Board for “multivariate” yielded the following
thread, showing specific multivariate test results that were posted:
Putting It All Together
Let’s take a look at a simple diagram that shows what a typical split test might look like.
Figure 2
The split testing software (or engine) is
representing by the gears in Figure 2 to the left.
When visitors arrive at the website, the
engine checks to see if it’s a repeat visit or
not. If it is, the visitor is sent to the same
sales letter as last time (to ensure the same
offer is shown).
If it’s a first-time visitor, a sales letter is
chosen at random, and the visitor is sent
there. The engine records which sales letter
the visitor is sent to (either by setting a
cookie on the visitor’s computer or saving the
visitor’s IP address), so return visits can be
routed back to the same sales letter.
As you can see, there are two or more sales letters, which may differ by only one element on
the page, or they could each be completely different from one another.
Now let’s compare that to multivariate testing, which, as you may recall, takes a different
approach. Instead of sending the visitor to one of several versions of a “static” sales letter, the
sales letter is assembled on the fly and dished up to the visitor.
Let’s first examine how a single element (a headline, in this case) is selected to be included in
the sales page.
Figure 3
Like the split testing engine, the multivariate
equivalent keeps track of which headline
(and all other elements) is served up to the visitor.
If the visitor has been here before, the same
headline is shown. If not, a new one is chosen
at random.
In this case, however, it may not be a true
random choice, where each headline has an
equal chance of being shown. It starts out
that way, but as testing progresses, the
winning combinations are assigned weight
values, according to how well they perform.
That is, headlines that are likely to produce
more sales (or whatever the goal is) are
chosen more frequently than those that are
less likely.
This is how a sales letter will ultimately
evolve to show the most profitable
combination and phase out the least
profitable.
Figure 4
Figure 4 to the left shows how different
elements are chosen independently of oneanother
and assembled into the sales letter.
The sales letter can either be built by the
engine, or it can be comprised of a starting
template, where the chosen headline, lead,
etc. are substituted where appropriate.
Each new visitor would get served up a
different combination, and repeat visitors
should get the same combination they got last time.
Now that you’ve seen how both a typical split test and multivariate test work, let’s combine the two.
Why would we do that?
So we can get the best of both worlds. A split test is best when you have two completely
different sales letters (like when you have a new letter challenging the control). Multivariate
testing is more ideally suited when you want to test a bunch of headlines, leads, video versus
no video, a header graphic versus none, and more—all at the same time. But it does not handle
two or more completely different sales letters well.
For example, if you had two sales letters—one an 8-page letter that was more benefit-focused,
and the other a 20-page letter that focused on story-telling, where the problem was
introduced, agitated, then solved (by your product)—how would you set this up with
multivariate testing?
It would be cumbersome and a lot of work, to say the least.
Multivariate testing is used when you have a lot in common between the two sales letters, with
the differences being the specific elements being tested. If the two letters are completely
different, how do you separate the elements to be tested? You might as well use the entire
sales letter as an element (which would in essence be the same as a split test).
Let’s imagine we had two completely different sales letters we wanted to test. And let’s take it
a step further and say that in each letter, we wanted to test different headlines, leads, graphics,
layout, the P.S., close, and much more.
In this scenario we would be sending visitors to one sales letter or the other (i.e. a split test).
And from there, each letter would be optimized for the best headline, lead, etc. (i.e. a
multivariate test).
How would that look?
Figure 5
Here you can see that the visitor hits the split testing engine first, which then sends them to
sales letter A or B.
Once arriving at the selected letter, the multivariate testing engine takes over and assembles
the headline, lead, or whatever elements are being tested.
Note that because these are two completely different sales letters, the actual elements being
tested in each can be completely different.
That is, in sales letter A, the headline, lead, offer, and layout are tested in this example.
In sales letter B, the headline, offer, and whether or not to show a header graphic or include
video is being tested.
That’s because the elements being tested would be specific to each sale letter.
There are two ways you could set up the scenario in Figure 5 above:
1) You could set up your split testing software to send your visitors to your sales letters just
like you normally would. Then, each sales letter would invoke the multivariate testing
engine to substitute the selected headline, lead, and every other element being tested in the appropriate location.
In this manner, your split and multivariate testing software are working together, even if
they weren’t designed to function in that manner.
This solution is ideal when you your multivariate testing software does not handle split
testing well, and when it lets your sales letters invoke (i.e. call upon) it when a specific
element needs to be inserted.
2) If your testing software handles both split and multivariate testing, you can set it up to
work like the diagram in Figure 5. Check your software documentation to see how you’ll
need to set it up.
And remember… to get maximum effect, you’ll want to conduct this kind of test for each
unique traffic source (see my article on page 35 for further explanation and some examples).
A Step-By-Step Guide
sing the methods I’ve described
in this report, you can take two
or more completely different
sales letters, come up with different
headlines and other elements to test
for each one, then set it up once and
let your engine take it from there.
The result is a self-evolving sales letter
that only cares about one thing: the criteria you define.
That can be more sales, more subscribers or leads, or some other goal.
As more and more traffic is funneled to
your site, your sales letter will learn
and adapt along the way, until you’re
left with the most profitable
combination of everything you put in.
Here’s a step-by-step approach to
setting it all up:
1) Start with two or three sales
letters that have different
hooks, or themes.
Or test a “newsy” advertorial
approach against one that tells
a story. Or a benefit-driven one
against a video sales letter.
The point is to have two or
more starting points, from
which you’ll set up each one to
evolve in its own way.
U
A Few Points To Remember
Ok, you’ve set up your system to handle your traffic
and serve up the pages like I’ve shown you in this report.
And you’re starting to see some impressive results.
Then, for no explicable reason, sales grow stagnant.
Opt-ins drop off. Conversions start fading.
Here are some tips to keep your engine well-oiled and
constantly producing for you:
Garbage in, garbage out. Your website is only going
to evolve based on the “raw materials” you supply.
Give it boring unconvincing drivel as sales elements to
test, and you’ll find that no combination will be
profitable.
Yesterday’s winner is today’s has-been. Consumers
become more savvy and immune to certain sales pitches
as time progresses. Remember that what worked
yesterday may not always work today.
Get the biggest bang for your buck. Start with the
things that are likely to give you the biggest gains first
and go from there. Your headline, lead, and P.S. are all
ideal choices to test first, because they tend to have the
greatest influence on your results.
Be alert! Watch what other marketers are doing and try
to see what’s working and what’s NOT working for
them. That can give you ideas for your own testing. The
Copywriter’s Toolkit contains a video that shows you
how to spy on them and gain the results of their testing
without doing any of the work or spending a dime.
A niche is a niche for a reason. Remember that what
works for one niche market may not work for another.
Give them what they want. Keep a pulse on what your
prospects want and “dangle the carrot” in your copy.
Stay fresh! Periodically introduce new tests with fresh
copy, graphics, video, and more. Continuously try to
beat your control.
Be creative! You never know what you’ll stumble upon
that’ll convert like crazy!
2) In each sales letter, identify each element you’d like to test. That could be the headline,
fonts, colors, anything. And as each sales letter may have its own unique elements to
test (e.g. a video sales letter might have two different videos as a single element to test,
while the non-video sales letter might have something else), it might be easiest to tackle
each sales letter one at a time, as though you were just doing multivariate testing on the
one. Create all the different versions for each element. Then when you’ve finished with
the first letter, move on to the next.
3) Now you should have something resembling Figure 6 below.
Figure 6
Obviously yours would differ, but you can see that for each sales letter, we have
different headlines, leads, offers, videos, etc. And each set of elements is unique to that
sales letter.
For example, both sales letters are testing different headlines, leads, and offers, but only
sales letter A is testing the header graphic. Only sales letter B is testing different video.
(NOTE: Depending on the theme of each letter, Headline A on one sales letter may be
completely different from Headline A in the other. The same goes for all elements.)
This is where you would need to set everything up in your testing software, either using
your split testing engine in conjunction with your multivariate, or one engine if it can
handle both. Consult your testing software documentation to determine the right way
to define it.
4) Now let the traffic decide what versions are most profitable as your engine does its
thing. You may need to periodically tell your engine to update all the stats (i.e. visitor
values) so it can start showing the more profitable ones more often.
5) Once you’ve arrived at your best-performing combination, it becomes your control. Now
come up with new versions of each element to test in the control letter (and introduce
new elements where opportunities exist). You may also want to create an entirely new
version of a sales letter and go back to step 2 to begin the process anew. That way
you’re always trying to beat your control.
For Maximum Profits, Test By Traffic Source
There’s no question the web has made it easier than ever to test your marketing campaigns and
come up with new controls.
In the old days, testing was time-consuming and expensive. You had to segment your list and
mail out multiple versions of a sales letter. Then you had to wait for two or three weeks.
Display ads in magazines took even longer, typically 3 months or more to gauge results.
Nowadays testing can take place in days or even hours, depending on how much traffic you
have coming to your site.
If you don’t test your sales letter online, you really are missing out on some easy and cheap
ways to maximize your sales and leads. Software to automate the process is as plentiful and
cheap as ever. I even bundle a free tool with my Copywriter’s Toolkit.
But have you ever considered testing by traffic source?
What I mean is, do you have an A/B sales letter evenly split between ALL of your traffic as a
whole, or do you have different tests in place for different sources of traffic?
The difference between the two can make quite a difference to your bottom line.
Here’s an example to illustrate what I mean.
Suppose you sell a weight loss product, and you sell to both men and women. Each of the two
will likely require a different sales message to maximize your chances for success.
If one source of your traffic is from an ad in a women’s ezine, do you want to send them to the
same sales page as you’re sending men from another ad?
More than likely you do not.
So when you’re split testing two different sales letters to see, say, which headline performs
better, you should have an A/B test set up for both the women’s traffic source and the men’s.
That is, you really have each test conducted independently of the other.
For Maximum Profits, Test By Traffic Source
It doesn’t matter if the sales letters in the test are identical (although in this example they likely
wouldn’t be). What matters is that the test results should be tallied for each traffic source.
Here’s another example: if I’m sell inbound and outbound call center services, I might have 2
selling points that primarily set me apart from my competition:
1) Price - my call center services are cheaper than most
2) Features - we offer more features than our competitors
But let’s say I’m not necessarily sure which of the two will appeal to my prospects more. So
when I write my sales letter, I create 2 different versions, one with a headline and lead that
appeals to bargain-hunters, and another that appeals to those seeking the most features.
Now I go ahead and test the two to see which one comes out on top.
Except…if I’m getting traffic from people seeking call center services for computer support,
they may be concerned more about price, since many of them view such operations as a “cost
center”.
And if I’ve got traffic coming in from people who want to conduct outbound telemarketing to
their house list, they may be more concerned with features, since theirs is a “profit center”.
You see, if I just did my testing for all my incoming traffic as a whole (using just those two
sources as an example for now), the bigger traffic source will decide the winning sales letter.
In other words, if the computer support traffic is bigger than the outbound telemarketing traffic,
and they’re more concerned about price, then overall sales might be highest with the sales letter
that has the price headline/lead.
But this is misleading, because if you split tested the two sales letters independently of each
other for each traffic source, you might find that the price one wins for the computer support
prospects, and the features one wins for the telemarketing crowd.
But you would never know this unless you tested in this manner.
Make sense?
Now you’ve got a sales letter customized for each crowd coming to your site. And you would
continue to test from there to maximize sales even further.
The above article was taken from my blog, http://www.JohnRitz.com
What To Test
hen it comes to testing, everything on your website is a possible candidate.
Even things you assume won’t make a difference can affect results under the right
conditions.
When copywriter and “Conversion Doctor” Eric Graham decided to test the impact of privacy
policies on conversion, he discovered that while only 11% of visitors clicked on privacy policy
links, of those who did, 17.4% went on to make a purchase.
Compare this to the fact that the average overall conversion rate for the websites in his study
was 2.9%, and you can see how visitors who click on a privacy policy link are far more qualified,
and closer to making a buying decision than those who do not.
But throwing everything under the sun into a test is neither practical nor necessary. For one
thing, the more elements you decide to test, the more visitors you’ll need
to make your results statistically valid. And traffic costs money.
Even if it’s “free” traffic, testing irrelevant things on your website can
divert you from testing things more likely to make a difference, costing
you potential sales.
So you want to start with elements that are likely to make the biggest
impact.
Start With Your Headline
Testing your headline is a must, because it’s your biggest opportunity. I’ve
already discussed how five times as many people on average read your headline over your body
copy.
Let me tell you a true story to drive the point home.
A client offered to pay me a sizable bonus if I could beat his control. This was on top of the fee I
was already getting for writing a new sales letter.
Well, before I even wrote the sales letter, I gave him some new headlines to test. And just like
that one of the headlines increased his conversion, giving him a new control.
And another thing happened, too: by testing the different headlines up front and seeing the
effect they had on conversion, I found out which “hook” would likely make the biggest impact
in the letter I was writing.
So testing can do more than simply boost your conversion. It can tell you things about your
target market that’ll ultimately indicate what they want. Then you can go ahead and give it to
them.
In the very first sentence of this report, I talked about a single change you can make to your
website right now that’ll take you only minutes to do, but could increase your sales and
conversions by 300% or more.
That’s the power of testing new headlines. (Yes, it’ll take you more than a few minutes to write
the new headlines, but it’s time very well spent).
So I recommend at a minimum you start there and do that as soon as possible.
After your headline, I’d move to your lead next (the first few paragraphs of your body copy,
including any deck copy), and then perhaps your P.S., arguably the most widely read part of
your sales letter after the headline and lead.
And don’t forget to test your offer.
How did you arrive at your price point for your product?
Most likely you saw what similar products in your niche were selling for and decided to offer a
comparable price.
That’s a good starting point, but you may find that by raising your price, you’ll make fewer
sales, but make more profits when you factor in support costs and everything else.
Or you may find that by lowering the price, sales explode, resulting in an avalanche of orders
that shoot your profits through the stratosphere.
The bottom line is you need to find that ROI “sweet spot,” and the only way I know of to do
that is by testing your offer.
That could mean more than just testing your price. You could test splitting up the payments,
using a “bill me later” option, adding or removing certain bonus premiums (yes, sometimes
taking away bonuses can increase your conversion and/or ROI), or forced continuity, an
overlooked buried treasure in many cases, among others.
In fact, the sheer number of things you can test is mind boggling.
I already mentioned you should start with the things that are likely to make the biggest
difference.
Let’s take it a step further.
One of my “hobbies” is learning about and watching the advancements made in the field of
quantum physics (yes, I do have “normal” hobbies, too). I like to read about it, and I’ve even
toyed with a few experiments of my own. I also like reading about astrophysics as well.
So I’ve read a lot of books about Niels Bohr, who articulated the Copenhagen interpretation of
quantum mechanics, and Albert Einstein, among others.
Now, one of the things about Einstein that used to impress and inspire me was the many
“thought experiments” he used to perform.
See, Einstein wasn’t a mathematician, but he often saw the mathematical abstracts in his head
and intrinsically understood them. In fact, he often enlisted help from trained mathematicians
to help work out the math on paper for many of his theories.
Einstein would think things through in his head and would just “get it.” He rarely performed
experiments in the lab. Only in his mind, where they were later tested and confirmed by others,
including his infamous theory of relativity.
Now we all can’t be like Einstein in this regard (nor do we have to be for our purposes).
But these thought experiments can be applied to marketing as well.
Here’s what I mean…
Obviously we can’t test every combination of everything. But we can conduct our own “thought
experiments” to zero in on those things that may make a difference. And then follow-up with
real-world testing to confirm them.
In the Taguchi world these thought experiments are known as design of experiment (DOE), or
Taguchi design of experiment. The purpose of DOE is similar to the purpose of multivariate
testing: to study the effects of many variables at a time on a desired result. DOE was originally
created to study the effects of sunlight, water, fertilizer, and more on agricultural production.
Later it was expanded to include the chemical industry.
The word experiment is just another word for test. It implies that the design of it includes a plan
and purpose.
So a thought experiment, or DOE, is really the mapping out process of what you plan to test.
Shown below in Figure 7 is an example of such a DOE.
Figure 7
This DOE is designed to test the effect of raising the price of a product on the overall return on
investment (ROI).
Simply testing a new price with the existing sales letter (i.e. a split test where nothing else
changes but the price) is certainly an option, but here the DOE takes it a step further and tries
to identify the specific elements that may influence the buying process with the new higher
price.
That is, the new headlines to test may introduce a different buying mindset. The payment
options (split up into multiple payments, bill them later, etc.) also are a factor. And finally some
subtle changes to the call to action can influence results as well.
If we were to do this as a split test, the order shown (from top to bottom, in the direction of the
chevrons) would be the likely order we would test.
So in the split test scenario, we would change the price of the product first. Then do a headline
split test, taking the winning headline and applying it to the next split test (the price plan). Then
the winner there would be applied for the final split test (call to action).
However, in a multivariate DOE, we would conduct one test with everything included, so we
include the interactions between each element in the test. In reality we’d probably also add the
price itself as an element to test (in addition to the price plan, which is already included).
These thought experiments will become second nature to you once you really begin to
understand your prospect and get to know him as an individual (that is, a “single person”
encapsulating all the typical qualities of your prospects as a whole).
Once you understand his wants, his needs, his fears, what he loathes, what he loves, what
keeps him up at night, indigestion boiling up in his esophagus. How he talks. His education level.
What he secretly desires above all else.
And here’s the icing on the cake: you can find out all of that and more when you test.
So your test results will give you additional ideas on other things to test also.
Of course you’ll research your market in the traditional ways, too.
But remember when I talked about my headline test above?
How it helped me to decide which hook, or theme, to ultimately
take with my sales letter?
Well, with pay-per-click (PPC) ads and the like, it’s even easier.
You can craft different messages in your PPC ads and see which
ones get the highest click-through rate. And you can do it quickly
and cheaply.
That’s powerful stuff, and it tells you a lot about your prospect, and maybe where to focus at
least some of your testing on your website.
Do enough of these, and your “thought experiments” will start to form on their own.
Actually, you already do this. Everyone does to some degree…even the most die-hard testers
(especially them). Because testing everything sounds good in theory but is difficult to do in
reality.
Do What Works For You
I used Einstein’s thought experiments as a vehicle to get you thinking about your prospect and
the kinds of things they might respond to.
The key word there is “might,” because that’s where testing comes into play.
However, I encourage you to develop your own approach to the “big picture” of testing. You
must do what works for you. If “thought experiments” (or DOE or whatever you wish to call it)
works, then great! If something else works better, I say “go for it.”
The truth is I tend to take a more “holistic” approach when deciding what to test. Instead of
thinking about testing a new caption under a photo or testing some handwritten text in a
certain spot, I try to envision the bigger picture and how the pieces fit together to match the
message I’m trying to convey as a whole.
That’s the same approach I showed in Figure 7 above. In that DOE, I could have just split tested
the new price with the same sales letter otherwise unchanged (and I would likely do that also).
But with a wholistic approach, I’m looking at the bigger picture. Because a straight split test
might be a failure in terms of raising my price and hoping for a bigger ROI. But when I include
the other elements to test in my DOE, I may get better results.
Or I may get worse. Such is the nature of testing.
Sometimes that approach works for me, other times it just muddies the waters.
When Science Meets Art
If you do enough testing, you’re going to find that a lot of changes have a negligible effect on
conversion, and sometimes results aren’t reproducible.
That’s because there are so many factors that go into why a visitor decides to make a purchase
at that precise moment in time.
Given enough visitors (samples), the results tend to be more statistically valid, but you must be
careful not to overanalyze your results.
Everything from the time of day and day of the week to the weather at your prospect’s location
probably has an influence of sorts, but much of it will be in the form of “noise.” This is where
you can go a little bit overboard with your analysis.
The truth is, your prospect’s state of mind probably has a lot more to do with anything, and
how do you measure that?
This is where the science intersects with the art.
Maybe you’ll do better developing a Zen-like approach instead of using thought experiments or
DOE.
Even genetic algorithms, powerful as they are, cannot entirely predict the markets on their
own. And so perhaps a self-evolving website is only the beginning. It’s where you take it from
there that ultimately matters, I suppose.
Personally I think there’s a subconscious state of awareness you can achieve when it comes to
testing and knowing what to test, but that’s a little outside the scope of this book. And
extremely difficult to put down on paper, even if I wanted to.
And evolution? Perhaps that’s a question as much for the artists as it is for the scientists among us.
But until you develop your own testing methodologies, there are some common starting points
worth testing that have shown to make a huge difference in conversion. They are listed below in Figure 8.
Figure 8
Common Elements Worth Testing
Header graphic
Pre-headline
Headline
Subheads
Lead (opening sentences or paragraph)
Bullets
Offer
Bonuses
Call to action
P.S.'s
Privacy policy
Layout
Fonts, styles, and type
Testimonials
Scarcity/takeaway selling
Credibility/proof
Audio
Video
Other graphic elements: photos, clip-art, graphical seals, etc., and their captions
Popups and other interactive elements
Conclusion
’m not an expert on testing. I’ve gotten better at it with practice, as will you. I’m also not an
expert in statistics.
But remember when I said you don’t need to be a marketing genius to test properly? Well,
neither do you have to be an expert in statistical, Taguchi design of experiments, or even
software engineering.
There is software available for your website that will automate much of the testing process,
letting you concentrate on the big picture.
You’ve undoubtedly noticed that I didn’t go out of my way to recommend any specific testing
software up to this point.
I’ve mentioned Google Website Optimizer, which is free, and overall not a bad place to start.
Also free is Google Analytics, useful for tracking all kinds of information about your visitors and
buyers, including costs and visitor value.
Some folks have asked me what I use for testing software. The truth is I’ve tried several, but
haven’t found the right tool within my budget that precisely fits my (admittedly non-standard)
needs, so I took some off-the-shelf software packages and modified them.
Even that didn’t provide me with everything I wanted. So currently I have a custom solution
under development. I may or may not release it to the public at some point.
But there’s certainly no shortage of options out there, and Figure 9 lists some of them. I have
not used all of them. I would recommend getting the opinion of someone who’s used the one
you are considering purchasing before you buy.
• Muvar – PHP software you can install on your site to give you a basic working
multivariate testing solution. Automatically optimizes your headlines, leads, formatting,
and other elements according to how successful they are in advancing the sale.
• Verster – Hosted solution means you don’t have to install any software on your site.
Simply paste snippets of code in your web pages that Verster provides, and you’re on
your way. Verster provides hosted split-testing or multivariate testing solutions.
• Multitrack Generator – Multitrack Generator is ideal for multi-paged websites, but can
be a bit complex to set up. However, it does use a proprietary algorithm developed
specifically for internet marketers.
• KaizenTrack - The algorithms KaizenTrack is using for Multivariate Testing are the ones
that were developed by Dr. Kowalick (the man who is the inventor of Taguchi for
Advertising).
• IntraTrakker – Another multivariate testing solution. A fairly new tool that I haven’t had
a chance to review yet.
• GoldbarOne – Marc Goldman’s complete e-commerce system, including testing
components. I’ve been using GoldbarOne® for about a year, and it really is packed with
features. You have to see everything they have, and they offer a 30 day trial for free.
Like Verster, GoldbarOne® is a hosted solution as well. Currently they offer split-testing
only (as opposed to multivariate testing), but I believe a multivariate tool is in the works.
Figure 9
My Copywriter’s Toolkit also includes a split testing tool, and it’s one of the few tools that I
myself still use.
In fact, some of the content in this report was taken from the
Copywriter’s Toolkit Companion Guide to the video series, from
the section that covers testing.
The toolkit covers much more than testing, though. In short, it’s a
collection of Web 2.0 and classic profit-pulling direct response
tools, resources, and software, designed to boost your website
sales and save you time and money.
For example, you’ll get access to the largest collection of proven
sales letters in the world. Online sales letters for sure, but also
stellar print display ads—including controls written by nearly any legendary copywriter you can
name. Gary Bencivenga, John Caples, David Olgilvy, Eugene Schwartz, Gary Halbert, Joe
Sugarman, Jay Abraham, Dan Kennedy, Joe Karbo, John Carlton, you name it. Legally swipe
these ads and adapt them to your own product or service to get proven winners right out of the gate.
After all, why stare at a blank page when you have this monstrous library of home runs to
choose from.
And if you’re looking for headlines, leads, a P.S., wording of your guarantee, or a proven call to
action to test against your current control, you’ll certainly have plenty to choose from.
And then there’s Web 2.0. I’m sure you’ve heard about it.
Wikipedia calls Web 2.0 a "second generation of Web-based services that emphasize online
collaboration and sharing among users." However, for anyone selling a product or service, I
think of Web 2.0 as a new type of interactivity, designed to get your visitors more involved in
the buying process. And as a result more sales for you.
That is, much has been discussed online about the social bookmarking aspect of Web 2.0. I
think of those methods as more of a traffic generating technique.
The kind of Web 2.0 I'm talking about is designed to boost your response and sales, help you to
make more informed business decisions based on what your visitors actually want, and save
you time and money.
Tools like my Interactive Tracker (iTrack), which will let you measure just about anything your
visitors do on your website. Things like how far down they are scrolling. The devices on your
page they’re interacting with. All without requiring them to refresh the web page.
Remember, if they interact with anything on your page, you know your headline did its job,
which is to get them to start reading your sales letter.
So the Interactive Tracker tool works best when it's combined with other elements your visitors
can play with on your website, like my "Interactive Johnson Box."
Sixty years ago, direct mail copywriter Frank H. Johnson was experimenting with ways to
increase the impact of his sales letters. Instead of presenting the offer towards the end of the
sales copy, he decided to try highlighting the offer in a rectangular box centered at the top of
the sales letter. The result was a tremendous boost in response, and the "Johnson Box" has
been used ever since.
The "Interactive Johnson Box" is a Web 2.0 version of the device, with dynamic content the
prospect can interact with.
And then we get into things like "tear sheets," a time-tested proven direct response tactic.
In direct mail, a tear sheet is an advertorial-type "article" that looks like someone tore it out of
a magazine. Tear sheets are popular with direct mail as a great attention-getter and credibilitybuilder,
and they're worth testing on your website as well.
I'll show you a tool that can generate them for you in minutes.
You can see an example of one here (http://www.copywriters-toolkit.com/tear).
How about another tool that rewrites your sales letter on the fly, based on choices your visitors
make? It's the ultimate "message to market" match, because your visitors don't see it happen.
As far as they know, they're seeing a sales page that matches their interests perfectly...
continuing the conversation in their minds.
That's the power of my Dynamic Sales Letter.
And if you want to take some of the techniques I discussed in this report to the next level, you’ll
want to learn how to "spy" on your competitors and top marketers, learning what kinds of
things they are testing, and more importantly, watching the winners that emerge. That way you
get the benefits of their testing without doing any of the work or spending a dime of your own
money.
Listen, it's not that difficult to do, but you need to know a few "techie" tricks (don't worry, I
explain it all in plain English, so you don't need to be a computer geek to benefit from these
profit-pulling methods).
See, long before I was a copywriter, I used to be an IT guy (way before it was even called "IT").
And I'm uniquely qualified to know how things like web servers, web browsers, and the like
function from end-to-end. And more importantly, how YOU can take advantage of these
technologies to get a "behind the scenes" glimpse of some of the most successful online
marketing campaigns.
Think that'll spark some profitable ideas? You betcha!
And in the Copywriter's Toolkit, I don't just tell you how to do all these things for yourself, I
SHOW you via screen snapshots and explanations in the companion guide. Plus you'll see me do
it all LIVE in the 12 videos, where you can pause, rewind, and follow along with me.
Anyone can do this, when you see me pull up the actual web pages and watch what I do.
There’s really much more in the toolkit—too much to explain it all here.
But you can get more information at the Copywriter’s Toolkit website
(http://www.copywriters-toolkit.com), as well as my blog (http://www.JohnRitz.com).
The Next Step
In this report I’ve showed you the importance of visitor value and how to determine your costs.
I’ve shown you how to test the right way, to maximize your return on investment (ROI) as much
as possible.
I’ve shown you the importance of testing by traffic source, because prospects from separate
sources will have their own expectations, and they’ll respond to a more precisely-matched
message. That is, you’ll do better by continuing the conversation already going on in their
minds.
We’ve talked about how advertising, when done right, is not an expense, but a profit center.
I’ve shown you the relationship between your market, your offer, and your sales copy, and how
all 3 must be spot-on for maximum revenue.
We’ve even covered the basics of a marketing funnel, and how many businesses aren’t
maximizing theirs.
And finally, I hope I’ve been able to show you how a self-evolving website could be just the
opportunity you’ve been waiting for to grow your business even further, and squeeze every last
bit of profit available to you.
How much can your website conversion rate, profits, and (perhaps most importantly) ROI improve?
Well, of course that’s going to vary, but it’s not unusual to see gains of 55% to 300% or more.
And by working with a professional copywriter with a proven track record, you have an unprecedented opportunity to reinvent your business and grow your business for many years to come.
