What Is Split Testing
& How Can It Make Me Money
Why bother to do "Split Testing"
Taking the time to do split testing is in our experience the biggest differentiator between average and highly successful online marketers.
In fact, of all Alliance Software's clients, the clients who make the most money online are the ones who regularly split test sales messages to find out which ones generate the highest number of sales.
The measurement they'll talk about is Conversion Rates - the percentage of customers who are "converted" from a lead into a sale.
Most web-sites start with pretty poor conversion rates. It's not uncommon for a web-site to have a conversion rate of 0.25% - or even lower. This means it takes (on average) 400 visitors to a web-site in order to make 1 sale.
Using smart split testing techniques, some of Alliance Software's clients have lifted their conversion rates from next-to-nothing, to above 2%. In some cases, we've gone well over 10%.
And at a 10% conversion rate, the difference is significant.
A site which previously received only 2-3 sales from every 1,000 visitors would suddenly receive 100 sales for every 1,000 visitors.
That's a 4,000% increase in sales!
Imagine what that would do for your business's bottom line!
It's this "secret" which is largely responsible for generating the biggest profits for some of our most successful clients.
More customers with the same advertising expense
Think about this from a different angle -
If you're paying for Pay-Per-Click advertising to bring visitors to your site, then you're investing a few cents to pay for people to visit your web-site under the expectation that a small percentage of those people will end up buying from you.
But to get the biggest return on your advertising budget, you need the maximum number of those people to buy from you.
Think about it for a moment, by using split testing, and increasing the conversion rate of your website, you're effectively getting a higher profit per customer who visits your web-site.
So Split Testing can also help you to improve the results of your online advertising.
So what is Split-Testing?
Web-sites are effective selling tools when they're used in the same way as businesses use their salespeople - to deliver a sales message.
Split testing is a way of testing two different sales messages to find out which one is the most effective.
Let's use an offline example of how this might be done..
Say you were a salesperson selling steak knives door-to-door. You want to sell your steak knives to as many people as possible, so you decide to test two different sales pitches to find out which one works the best.
For each door you knock on, you try out one of the two sales pitches on whoever answers the door, and you record how many times each sales pitch resulted in a sale.
You might repeat this process for a few hundred tries, until you'd tested both sales pitches on enough people that you could be certain that one would work better than the other.
After testing both sales pitches to enough people, and tallying up the results, you could see which sales pitch ended up doing the best job of convincing people to buy, and making the sale.
This process is known as A/B split testing.
It's the easiest and most common way of split testing a web-site in order to find which (of two versions) works the best.
How often should I split test?
If you're able to improve your sales (even by just a small amount) each time you run a split-test - then surely you would want to split test as often as possible!
In fact, out of all of Alliance Software's clients, those who are making the most money online are the ones who are constantly split testing different versions of ads, web-pages, e-mails (and practically anything else!) in order to achieve the maximum number of sales!
Taking the approach of making continual improvements to your marketing and sales materials is something every business should do if they're after the maximum number of sales possible.
So in short, you should split test as often as possible, on as many elements as possible.
Where can I run a split test?
The short answer to this is "You can run a split test any time you ask someone to make a decision".
Every time you offer someone the opportunity to buy your latest special, or click on one of your ads, or read a special report on your web-site - you can run split test.
Popular elements which Alliance Software clients split test regularly are:
- Sign-up / Name Squeeze pages
- Product information pages
- Sales pitch pages
- Autoresponder e-mails
- Newsletters
- Advertisements (Pay-Per-Click ads, banner ads, etc)
Which items should I split test?
There are literally hundreds of elements on web-pages, e-mails and ads which you can use to split-test. But of those, there are only a few dozen which usually affect sales covnersion rates - from word usage to colour usage, headlines to text boxes.
Having conducted countless split tests on behalf of our clients, we've been able to gather data on exactly what has worked and what hasn't over time.
Of course, we can't reveal confidential client details, but some of the "big" factors which have created some of the biggest improvements in split tests have included:
- Headlines;
- Pre-headlines and sub-headlines;
- The opening paragraph.
- Formatting and colour useage;
- Calls to action (sales closes);
- P.S.'s at the bottom of salesletters;
- Guarantees;
- Testimonials;
- Sign-up boxes and ordering pages;
- How images are used;
- etc
Testing different versions of these elements has often created some of the most significant improvements to our clients' sales conversion rates.
Give me some example of split tests you've run
In a recent example, Alliance Software built a web-site for a client who was runnning a seminar.
As part of his promotion, this client sent out emails to a number of qualified prospects. This email invited them to view a page aimed at convincing them to attend his seminar.
Unknown to the clients, they were randomly allocated into three equally sized groups and each group was directed to a different page. The pages they were sent to were identical except for the headline and first paragraph. The three headlines were each written to appeal to a different reason to buy a ticket.
The difference in results were amazing. Nobody in the first group placed a single order and only 2% of respondents in the second group ordered. However, 6% of respondents in the third group ordered - an excellent response rate!
Something in the message sent these people was far more appealing.
Although the difference between the three web-pages was only minor, the third page achieved a much better response rate.
Obviously the client chose to use the web-page that generated the highest response rate as the basis for promoting his seminar to thousands of people over the coming months.
Can you see the effect that a different presentation made in the number of orders generated?
Is there anything I need to be careful about when split testing?
One of the important considerations of split testing is making sure you've got a big enough sample size.
Let's say that you flipped a coin twice, and both times the coin landed on heads - your data would say that every time you flip the coin it's going to land on heads.
But you know this isn't true!
You haven't flipped the coin enough times to get any accurate data.
The same applies to split testing web-sites. The more people you test on, the more accurate your data will be in your split test, and the less likely that "chance coincidences" are making their way into your data and skewing your results.
This can be very important when you're making assumptions on which version of a sales page might work best for the next 1,000 or 10,000 people who visit your web-site.
But on the other hand - if you're paying for traffic to your web-site, you want to send the minimum number of people through your split test so that you can start using the best possible split test option as soon as possible - and as a result achieve the best return on investment from the traffic you're paying for.
There are several calculations which you can use to work out the optimal number of people to send through a split test before you can expect to have accurate results. The team here at Alliance can help you with these calculations.
Are there any problems with split testing I need to know?
One problem with traditional A/B Split Testing is that you can only test one variable at a time.
For example, if you wanted to test two different headlines, two different PS's, and two different sign-up boxes for your name squeeze page, you would need to conduct three separate split tests - one each for the headline, the PS's and the sign-up box.
And three separate split tests means that you need to send three separate sets of visitors to your web-site, with each group of people large enough to give you an accurate result for your split tests.
This can be very time consuming.
However, there is an advanced form of split testing known as Taguchi Testing.
Taguchi Testing is also known as Multi-Variate Testing because it's able to test multiple variables (ie - several different elements) at the same time.
This can be a very powerful tool to have if you want to optimise several different elements on your sales pages because it means you can achieve the same result as you would from many different split tests - but do it with less visitors, in less time, and with less cost.
For clients who want to optimise their sales pages quickly, Alliance Software is able to provide the marketing expertise, software, and technical know-how to help you to perform Taguchi Testing on your sales pages.
I'd like use split testing to increse the conversion rate on my website
Click here to contact us and we'll call you to talk through your requirements.