This is an in depth look at CRO, provided by Paul Savidge, the conversion kingpin. When finished reading this epic blog post you surely will walk away with a lot of information, but before you leave, let him know what you think and if you have any questions, he will be here to answer them.
What Is Conversion Rate Optimization ( CRO )?
First sentence out of the box I am going to say this is going to be a long one. I will start off with a simple answer to the question; what is CRO? The science of getting more conversions out of the already existing traffic to yours or your client’s site. It only goes downhill from here. lol
So a more detailed definition… Coming from Wikipedia: “The method of creating an experience for a website or landing page visitor with the goal of increasing the percentage of visitors that convert into customers. It is also commonly referred to as ‘CRO’. “
The absolute basic math behind Conversion Optimization is:
conversion rate = number of desired actions / number of visitors X 100
If 1000 people view your home page and 12 call you for a consultation that would be a conversion rate of 1.2%
To clarify the definition even a bit further I will list a few examples of what Optimization IS and IS NOT:
Conversion Rate Optimization Do’s:
- Know Your Numbers: If you don’t know what traffic is and what it is doing ( or not doing ) CRO is not for you.
- Find The Leaks: The absolute best place to start is where you are losing the most. Half the battle is looking at point one, and knowing where these spots are.
- Run Tests: This is where the science creeps in fast with CRO. Your site becomes a lab, and you run test to optimize performance.
- Test Everything: Once you get “The Bug” you will anyway! lol
- Keep Testing: I find that many changes have a fall off point where the gains are diminished. Continual testing keeps your site at peak performance.
Conversion Rate Optimization Don’ts:
- Small Sample Sizes: If you cut a test short you stand the chance of publishing the wrong variant, and having to go back in and test over – wasted time.
- Don’t Be Scared: We all have a tendency to replicate sites that are successful, but don’t be scared to push some limits and be different.
- Don’t Let The Guy / Gal That Writes The Checks Dictate Results: I personally… have a hard time with this one at times. I want what I want. There are more than a few times that testing tells me otherwise… but I have conflict with the “Look” sometimes. TESTING, and the RESULTS are the boss!
- Don’t Under Estimate Small Gains: 1% here and 3% over time start to add up. It’s not super sexy like a 300% gain from a single element But, the end goal is reached the more elements you increase conversion with.
Why Should You Care About CRO?
You should care about CRO for a few reasons. One way or another, you are paying for traffic, and a high conversion rate means a better return on that investment (ROI). It’s far more cost efficient to convert visitors you already have than it is to obtain more visitors. Basically CRO is free.
It lowers your overall Customer Acquisition Cost ( CAC ). Doubling your sites Conversions in essence will cut your acquisition costs in HALF. CRO gets you around your end user’s limited attention span. Getting your visitors to find what they are looking for sooner, so they don’t find it somewhere else.
CRO is important!
- Higher Conversion Rate = Better ROI
- Maximizes Overall Profits
- Cost Efficient vs. Finding More Visitors
- Decreased CAC
- Work Within Your Visitors Limited Attention Span
I think now that we have a better understanding of what Conversion Rate Optimization is, we can start breaking it down a bit. So I am going to throw a question at you… what are we Optimizing? Rates or Conversion? Conversion right…
Get updated for new awesome content as it happens, enter your email below.
What Is A Conversion?
There are more than a few ways to look at this. There is the big daddy conversion “The Sale”. Then there is the trail of smaller conversions along the way that gets the CM there. The first little conversion at the search engine page to click your link is really no less important than hitting “Buy Now” and continuing through your purchase process.
The most important aspect of the whole topic, and honestly not to many sites define “Conversion”. What is a conversion? What does it take to make it happen? I personally think this is big stuff. We are talking PhD over the top math stuff here right? Well I just so happen to have a resource… Dr. Flint McGloughlin over at Marketing Experiments aka MECLABS. ( http://www.marketingexperiments.com/ )
If you go over and check out the Research Directory on the MECLABS site you will see they are test running fools. Over the years this brainiac of a guy has figured out an equation for Conversion. The equation is as follows:
( I told you it only goes downhill from the easy explanation )
Bunch of letters and numbers. I will define the letters for you so you can make some sense out of it.
- C = Probability Of Conversion
- M = Motivation Of User ( When )
- V = Clarity Of The Value Proposition ( Why )
- I = Incentive To Take Action
- F = Friction Elements Of Process
- A = Anxiety Over Entering Information
The largest variables overall to overcome are “Motivation” and the “Clarity of Value”. Motivation as I work with it, is where in the buying process a visitor is. I have a broken the buying process into 3 segments.
- Information Seekers: These can be those in the very front end of the buying process to people looking for information to write blog posts lol
- Comparison Shoppers: This category is just short of committed to buy, they are looking for the right “Value” – This category can be boosted into buyers with the right amount of information provided.
- Actual Buyers: In a commerce based environment a great percentage of this category will be repeat visitors to your site, once they decide to pull the trigger and purchase. In lead generation and other online intents this is usually not the case. There is a point of urgency that there is only 1 chance at a conversion per visitor.
Motivation as I see it has more to do with traffic based variables more than anything. It directly relates to your ability to target the right segment offsite than anything to do with onsite. As seen in point #2, “Value” is important. So the Clarity of Value within your offer has to be more than obvious. More specifically 3 times more than obvious! THIS is gained through onsite interaction.
Incentive to take action… This one is played out many different ways and kind of dependent on what the offer is. If you are developing an E-mail list, it could be a free report. If your conversion is a phone call… it could be a “free Consultation”. If your focus is commerce, it could be a discount, or free shipping etc.
Friction and Anxiety
I am separating these 2 variables for a reason. Of the 5 variables required to make a conversion these are psychology based variables. Motivation, Value, and Incentive are very tangible and I think easily overcome. Friction and Anxiety at times can be a bit less pragmatic.
Generally with friction you want to reduce the drag that someone may feel in converting. An E-mail opt-in for example. Reducing the number of fields required is considered reducing friction. As I discussed in my first article ( 7 Awesome Tips For Conversion Rate Optimization ) you can manipulate friction within an opt-box to either get more calls, or more E-mail correspondence.
Anxiety – this directly relates to trust. Having a secure site seal… testing the “Your e-mail is safe with us” statement below a e-mail opt-in. I have seen the overall site color has been a factor ( Oh boy was that test a stretch, but it paid off in the end )
Some considerations that have to be looked at in overcoming anxiety is overall site performance. Broken links, buttons that don’t work, disconnected phone numbers etc. all add to the overall anxiety in creating a conversion. There are tools that can test much of this for you, and if you feel your site is not performing to a level it should, I would without question start looking there first.
Knowing Your Numbers
In the next section we are going to start breaking down some testing methods. Before we get there, I want to STRESS the importance of traffic, and more importantly your understanding of YOUR sites traffic.
The image above is the traffic across one of my sites. The red arrows indicate Saturdays. Very quickly you can tell that weekends suck, and Monday is the best day of the week for me. By no means by intention this site draws a “Professional” at work crowd.
So why is this important? When you are testing, you want to “try” and test across the same traffic patterns. If you start a segment of a test on Monday and it ends on Thursday. On Friday you start a new segment… it’s safe to assume the impressions across your test area over the weekend will not be the same Traffic segment that is present during the week.
On this site specifically, I test Monday, Tuesday, Wednesday, and Thursday only. Testing into Friday Saturday and Sunday in general has proven to provide wacked out results. In recent months I have actually gone as far as to start exchanging elements of my site to better target my weekend traffic. ( These changes go up Friday evening and are pulled Sunday in the mid afternoon. )
These changes are possible because I know my numbers. The truth… it was a fluke test that was run over a weekend that indicated some possible potential. That lead to some pretty hard core testing over the next I think 3 months to optimize the site for the weekend traffic vs. the week day traffic.
TESTING TESTING ONE TWO THREE TESTING
I use 3 different types of tests; Multi-Variable, Serial Testing and A/B Testing. The last one – A / B testing is without question the Test method that is preferred. That being said… It cannot always be used based on a few factors.
Based on some factors I will discuss here, I simply do not use this method. In the above image Variable “A” is the control ( no changes made ) “B” is a headline change. “C” is a button change. And then “D” is a Headline and Button change. So you run the traffic across the test page. Visitor 1 gets page “A” visitor 2 gets page “B” visitor 3 gets page “C” visitor 4 gets page “D” visitor 5 gets page “A” and so on.
At the end of the test cycle One of these variations should be a winner… we will say that “C” the button only variation is the clear winner. That’s great right? Increased conversion over the page. Wait a minute… What happens if I go in and find a better headline now? I test with an average of 1000 visitors per variable. the above test would take 4000 visitors to complete.
As I see it… and more so as my personal methodology sees it, I am better to test the Headline first ( 2000 visitors in total ) and then test the Button ( 2000 visitors in total ) I end up with 2 points of gain vs. the one point that Multi-variable testing can provide. I am over here shrugging my shoulders I just don’t get it… lol.
I use this method more than I am willing to admit at times. When dealing with the specific testing that I have specialized in ( E-Commerce implementing WordPress and WooCommerce ) many changes are template based changes. So implementing A / B testing in that specific environment can be done.. But just the coding to do so is extremely time consuming.
Serial testing is testing a single element for 1000 impressions, and then making the element change and tracking for the next 1000 impressions. This test above and beyond the other 2 methods REQUIRES you understand your traffic and its patterns. Looking at the above traffic chart, I start this test on Monday and it takes till Thursday to get the needed traffic, I will wait till the following Monday to start the 2nd segment of the test.
This type of testing is subject to the whims of a sites traffic. If for some reason there is a boost in the test period, on one segment and not the other, the test can be skewed. If for some reason in the 2 segments of testing you lose a SERP ranking and traffic drops. Again the result can be skewed.
There have been times that I have run this type of test, and run it more than once across a test set and the numbers were just jumping all over the place. Let me tell you it is frustrating! I have since implemented a preliminary test across any Serial Test’s that I do. I run a simple A / A test of the elements.
A / A testing is exactly like A / B testing, but there is no variation change. The desired outcome with this type of testing is a 50% per side type result. I will tell you it does not always happen that way. Because at times I do not have the advantage of truly knowing the traffic I am testing across, this type of testing gives me some re-assurance that the flow of traffic is consistent by nature.
A / B TESTING
The big daddy of CRO test… the A / B test. Out of the 3 testing processes I have gone over here, this is the easiest of the bunch. 2 Variables, be it 2 headlines, or 2 images or 2 button text changes. A total of 2000 visitors across the test area, 1000 per each variation and may the best variation win! lol
To run this test, you will obviously need some outside help. This test can be set up in Google Analytics. ( My primary test tool of choice ) There are many sites that offer testing platforms as well. Visual Website Optimizer ( VWO ) https://vwo.com/ is one that I use. There is Optimizely, Maxymizer, many of the landing page platforms have the functions built in to their back ends.
Why 1000 Visitors per Element?
So we have gone over the basics here and I am betting you are asking why 1000 visitors per side? The answer here is obviously going to be mathematical… The equation used to determine that A is a winner over B, and by a point of significance is bell’s theorem ( you can look it up if you want )
I personally test to a point of 95% assurance. meaning that to either side of a decision there is a 2.5% allowance. In other words I can say that I am 95% sure the results are accurate give or take 2.5%. The variable within the equation to make that happen happens to be 1006 to be exact, but 1000 is more than fine for me.
Below is a an image of a statistical significance graph.
To give you an idea of how tight that ends up being. If you were to draw a line straight up where the Zero is, That would be the 50 / 50 line. as you move to the +1 or -1 mark that would be the 75% mark. Moving to the +2 or -2 mark is that 97.5% ( 95% overall ) mark we are aiming for, and the gap for 2 to 3 ( + / – ) is that 2.5% tolerance.
If you were okay with an 80% assurance rate… the number needed there is 600 each side. Using the chart above.. Your statistical significant jumps to 10% per side. The colored section of the graph jumps a bit past halfway up the next segment.
Wrap it up
So there it is… that’s my world! What I think is so awesome about Conversion Rate Optimization is the ability to make it as complex or as simple as possible. Simple is usually the preferred method, but there are those times that complexity kind of takes over. The secret is knowing when the complexity is not so needed.
Some of the things I did not really get into in this article are some of the other testing things that can be done such as Heat Mapping, or Click Mapping as examples. These are very “Trendy” right now.. they provide the WOW factor for clients. ( I am obviously not a big fan ) They do however have their place.
Click mapping is actually kind of cool, it allows you to see where end users are clicking on your page. Such as images or perceived buttons. ( This type of testing can isolate and identify Anxiety based page issues ) Really not a test I run a lot of, but there have been times it has come in handy.
So that’s it… any question or comments hit me up below!