Podcast: Play in new window | Download
So maybe your website isn’t quite performing as well as you’d like. Maybe you’re just not seeing much conversion, or at least the conversion rates you’re hoping to reach. How do you go about figuring out where you may need to start tweaking things to improve on that?
A/B split testing is a great way to test a few different variations and see what jives better with your audience. But where do you start? For a website with a large number of pages, that can be a pretty daunting task if you start testing variations of every page. Your best bet is to first see what individual pages – and aspects of each page – are performing and converting well as is, and then start working on those that need help.
Chris Dayley is the founder of Dayley Conversion, where he uses his expertise and knowledge of using scientific, data-driven testing to help businesses improve their customer’s user experiences.
Rich: Chris Dayley, founder of Dayley Conversion, is a digital marketing entrepreneur with a passion for helping businesses improve website design through data driven testing. His company provides full service A/B testing for businesses including design, development and test execution, to help businesses learn more about their users and improve onsite conversion rates. Chris, welcome to the show.
Chris: Hey, thanks for having me.
Rich: I’m so excited about this. We do not talk about conversion optimization nearly enough on this show, so I’m very excited to kind of jump right into it with you.
Chris: I’m excited to be here.
Rich: Excellent. So how did you become so interested in conversions and conversion testing, where did that come from?
Chris: So I got my start, my first real job was doing SEO and I fell in love with just the digital marketing space in general. So I spent a couple of years in the SEO space getting traffic to websites, I dabbled a little in PPC and social media, but bottom line was I was getting traffic to websites.
I ended up in-house for a company and I went in for my annual performance review and we had just been killing it for these traffic metrics. We had gotten ranked for some highly competitive terms and we had tripled our organic traffic over the last year, all these great winds in terms of traffic. But I was caught really off guard when our CEO came back to me and said, “So what. You’ve gotten all this traffic, but did it generate any dollars, what did that do for the bottom line?” And I didn’t really know at that point. It turns out after I was digging in I started to realize that a lot of this traffic I was generating wasn’t converting.
And the company I was with at the time didn’t have anyone that was doing conversion optimization so no one could really help me get this traffic to convert. So I started it off as a pet project. I ran my first test and I had no idea what I was doing. It was probably the worst run test of all time, but I knew we had to do something.
So I ran my first test and saw a small jump in conversion rate, and that just excited me. The fact that prior to that test I had just been under the assumption that if you get good traffic to your site, it’s just going to convert. And that was my first indication that if you have great traffic but the site experience is poor, then people aren’t going to convert. So I just fell in love with this idea and it’s really been an exciting last several years for me to just kind of start to learn what drives people to make decisions. Some of the psychology behind it is just fascinating to me. So that was kind of a long winded answer for you.
Rich: No, that was great. And I’ve always been fascinated by the psychology that drives conversions, sales and all that sort of stuff, so I like that, too. But the one takeaway that I really liked from that is I so often talk about quality traffic and the importance of quality traffic over traffic. And what you’re saying is that quality traffic is important, but it’s not enough. You need to do things on your website that gets those qualified visitors to become qualified leads. And that’s maybe the piece that a lot of us are missing out there. We’re really good at bringing the right people to our website. But maybe we’re not good enough at getting them to take the next desired step.
Now before we dive too deeply into those waters, I know that one of the things that’s kind of prevented me from doing more in terms of conversion optimization is just the idea that it’s something that big companies do. So as a small business owner, I don’t have the time or the bandwidth to really focus on that. I’m just trying to get more search traffic and more digital ad traffic. What do you say to people like me who think it seems like too big a nut to crack or too much work to optimize for conversions?
Chris: Yeah, I get that all the time. Especially – like you said – from small business owners, and usually I hear one of two things. Number one, it’s too time consuming for the ROI. And number two, I don’t get enough traffic. For both of those concerns I have the same response, and that is, if you run the right kinds of tests it doesn’t take as much time as you might think. And number two, no matter how much traffic you get to your site, you can still test and optimize.
A lot of people will say they first need to work on the traffic side of things and get enough traffic to their site and then once they have enough traffic they’ll start to optimize. What I see happen a lot of times is companies, because that’s their first focus – whether it’s PPC or social media or SEO – they devote so much time and resources and strategy into these, that once they have enough traffic they’re just so deep into this that they’ve got blinders on by that point.
And so getting back to the conversion optimization part, if you have low amounts of traffic and if you have very limited bandwidth, I would just say you need to be efficient with the kinds of tests that you run. And usually I suggest running bigger tests. A bigger test to me is one that is testing lots of things on the page at once. A lot of people in the A/B testing community hate these types of tests because it’s really hard to figure out when you’re testing more than one thing at the same time, once you have a winner it’s hard to decide what you can actually attribute the lift in conversion to. So if you have changed some images or colors and you have a different headline and different amounts of content and you have a winner, it’s hard to decide was it because of the headline or was it because of the image.
When you have a small amount of traffic and when you have very limited time to put together tests, those are the kinds of tests that are going to really drive ROI and really be worth your time, versus some of the smaller tests that I like to run with larger clients, like tweaking headlines.
Rich: Yeah. That’s what I’ve heard in the past, that for A/B split testing you’re testing one thing. You’re testing the photo or the headline or the message or the font, but not that you’re throwing up everything against the wall. It sounds like what you’re saying is you found success by throwing everything up against the wall in trying two very different home pages or ads, whatever it may be. Am I getting that right?
Chris: Yeah. And again, this isn’t something I’d really suggest for larger businesses, but for smaller businesses. One of the problems with testing smaller things is that when you have a limited amount of traffic and a limited amount of bandwidth those small tests where you’re just testing one thing at a time, they usually only have small lifts, as well. So usually what I’ve found is the larger the change is, the larger the impact.
So if you’ve got a limited bandwidth and a limited amount of traffic, you want to go for a big impact, because you don’t want to waste three months worth of traffic testing a headline that only lifts conversion rates by 2%. Right?
Rich: Right. So let’s say we decided Chris makes sense and were going to test some stuff on our website. I can think of three types of pages that I might try to test for conversions. One would be the homepage, one would be some sort of services page. For me maybe it’s getting people to sign up for our email list services, or for a dog trainer it might be about coming to a specific course that I’m teaching on dog training. And the other type of page might be something like a squeeze page or a landing page. Do you have recommendations on where we might want to start, or do we test all three things, or is there even another option that I haven’t thought of yet?
Chris: That’s another really common question that I have. Usually what I suggest – and this is something I do with my clients – is go through and look for where you think your tests will have the highest impact. So that’s kind of obvious. But the way that you can do that is to look at the amount of traffic that you get to each page and look at the current conversion rate.
So one of the difficult things with homepages, people always think of homepages as the first place to start testing because it gets the most traffic. A common problem with homepages is they’re probably isn’t one conversion point. Usually a homepage is a filter to help funnel people into other pages of your site.
Rich: Exactly. It’s almost like walking into the lobby of a building.
Chris: Exactly. And then you try to figure out where to go from there. So one of the problems that I’ve seen with companies that start on a homepage is they’ll run a test, and either they won’t have enough clearly defined success metrics or they won’t get that big of a lift. Because again, homepages can be somewhat ambiguous for users. So if you’ve got 8 pages on your site, or 800 pages on your site, there’s so many places that people could go. And if you try and increase traffic to just one of those pages, that might not be the page that your users want to go to.
So that’s not to discourage people from testing homepages. But again – going back to your question – if you’re looking for one place to start, I don’t typically recommend the homepage. I typically recommend starting somewhere down the funnel.
Now if you’re looking at other pages – you talked about maybe a services page and a landing page – with those I would look at the amount of traffic and then look at the current conversion rates of those pages and ask yourself, “If I had the same percentage of lift on both pages, which one would generate a higher ROI for me?” It’s probably going to be the landing page, so that’s how I would kind of suggest it.
If you have a services page and you have a really, really low conversion rate on that page, a small lift isn’t going to make that big a difference, because you’re already not converting very many people. If you have another page that has a higher conversion rate, then a lift is going to continue to play off of those strengths. If you have a lift on a page that has a 10% conversion rate, you go from a 10%-11% conversion rate. But if you have a lift on a page that has a 1% conversion rate, then you’re going to go from a 1% to a 1.1% conversion rate.
So that’s really kind of what I look at when I’m looking at prioritizing pages. You can go into more depth than that but that’s probably the easiest way to start looking at things.
Rich: Alright, so what I’m hearing is, maybe temporarily putting a hold on making changes to the homepage, taking a look at some of these services pages and some landing pages, and seeing how much traffic they’re already getting and what their conversion rates are already at, because that’s going to make a bigger difference. And maybe also one of the other factors in here is if you do have different products at different price points,that you may want to factor that in as well. If you’re getting even a 10% increase on one page and a 10% increase on the other but one of those pages is a much higher price point, then maybe that’s the first page to start on. Would that also make sense?
Chris: Absolutely. So you want to weigh those things additionally. And one other thing, on the homepage, the one place that I would suggest you start on the homepage is if you have a really high bounce rate on your homepage. Bounce rates affect a lot of things. They are starting to affect your search rankings, but they also affect the effectiveness of the rest of your site. So if you have a really high bounce rate, I’d suggest that a high bounce rate is anything over 60% bounce rate, and that obviously varies from industry to industry and depending on what kind of traffic you get. But if you have a really high bounce rate – meaning you’re losing a lot of the traffic to your homepage – then it might make sense to look at your homepage. But if you have a pretty decent flow of traffic to your site through your homepage, that’s when you don’t want to touch it at that point.
Rich: That makes sense. So I’m going to use myself as a guinea pig here. So let’s say that I’ve got a page on flyte’s website about search engine optimization and another one about social media and another one about email marketing, and at the bottom of each one I’ve got some sort of call to action about a half hour free consulting and then it’s a click to the contact form. Am I testing to see if my conversion on those pages getting somebody to click on that link to head over to my contact form, or do I need to somehow track also or separately the number of people who convert on my contact form who get to the contact page and actually fill out the form? Am I getting too into the waves, or do you get what I’m getting at here?
Chris: I see what you’re getting at. You want to track both metrics. And this gets even more complicated when you have an even deeper funnel. So if you have a product page and that product page takes you to a checkout page, and then that checkout page takes you to the payment page, and then the payment page takes you to the thank you page, you want to track every step of that process.
You might get a really high click-through rate on one variation that you’re testing, but maybe you put, “get free money from Obama”, and so people are clicking on it and then they get to the page and realize they’re not getting free money. And so you want to track all the way through the funnel. And like you said, you get kind of into the weeds at that point, because if you have a small lift in one of those numbers and then you have a slight dip in one of those numbers, it’s kind of hard to figure out should I go for the one that increases the click-through rate or the one that increases the form completion rate.
But those are absolutely things that you want to take into account and you want to measure. I suggest measuring as many things as you can doing your tests.
Rich: Alright, as you’re talking I’m thinking I need to, for each one of my services pages, I need to be measuring what’s the click-through rate on that free consult offer that I’m making. So let’s say that taking one of those service pages – and I know a lot of our listeners have pages that are similar in whatever industry that they’re in – what kind of things would you recommend that we should be testing on that page? How do I run a side by side comparison? What are some of the things – based on your experience – should I be looking at? Is it the copy, is it the images, is it the title tags, is it everything, is it the size of the button?
Chris: Yes.
Rich: I was afraid of that answer.
Chris: So I’ll break it down a little bit and hopefully make it a little bit more approachable.The first thing that I suggest when you have a page like this is install a heat map on your page. There’s lots of tools out there that do that. There’s a free one that you can use called Hotjar – and it’s just one snippet of javascript that you can put on your page – and that will give you an idea of how far down the page people are scrolling and where people are clicking on the page.
Because one of the things that’s really challenging sometimes is knowing where your potential problems are. It’s not always easy to spot, especially when you look at Google Analytics. Google Analytics will tell you how many people went from one page to the next, but Google Analytics doesn’t tell you how they got there. Did they click on the button that you think they’re clicking on, or did they click on the navigation, or did they click on a footer link or one of the buttons that was lower down on the page?
So you want to figure out where people are engaging currently, where is people’s attention currently, and then that will give you a good idea of where should I focus my attention. If 90% of the people are clicking on that button that’s above the fold, then you probably want to make sure that you focus your attention on that above the fold area. If you’re getting a lot of engagement all throughout the page, then you want to look at the entire page.
So one of the things that I usually will recommend to my clients, is once you’ve got that heat mapping data back, cover up all of the stuff that you already expected your users to click on. So if you think that they’re supposed to click on the button, then cover up the button. Just cover it with your hand or put something over it, so that you can focus on the other things. Because one of the things that I find almost every time is that a large percentage of users are going to be clicking on stuff that you didn’t expect them to click on.
So they might be clicking on images that aren’t actually linked anywhere, they might be clicking on text that you have that is hyperlinked somewhere else, you might see that most of the people are clicking on your footer links. And if people are clicking on your footer links, that means that you don’t have a very good landing page.
So if you cover up the stuff that you were hoping that they’d click on or that you expected them to click on, you’ll start to see other stuff that your users are clicking on and that will start to give you some ideas of what you can test.
Rich: Alright, good to know. How long should we run a test? How long before we think we have some information that’s going to make a difference?
Chris: Well, the first thing is longer than you probably want to run it. Most business owners, as soon as they see an increase in any kind of metric, they want to call a test because they want to start capitalizing on it.
So the first thing that you want to look at, almost every testing tool out there will calculate the statistical significance for you. So if you use Optimizely or Visual Website Optimizer, both of those have free accounts that you can sign up for. So if you start with one of those – or any other testing tool out there – it will calculate a statistical significance for you. Now I don’t place a whole lot of stock in it, but this will give you a first reference point to look at. My suggestion is you want to get at least a sense of the statistical significance.
Now if you didn’t do well in statistics in high school or thought about statistics for ten or twenty years, basically the way that you can think about statistical significance is just the confidence that you have a winner. That’s the most basic way that I would describe the statistical significance.
Rich: Right. Because you might have a day where one seems ahead of the other, but actually that day you got mentioned in some online news magazine and it linked to your site and it’s completely skewed because those aren’t typically the kind of people who come to your website, for example.
Chris: Exactly. So the first thing that I look at is the statistical significance, and the second thing that I recommend for almost any company – whether you’re big or small – is to run it for a minimum of a week. Because what that does is that will normalize out your “day to week” variables. Most websites convert very differently on the weekends than they do on the weekdays, sometimes Mondays convert better than Wednesdays, so you want to get at least one of every single day of the week in your test.
Because what I’ve seen happening in some businesses is they’ll run a test during the business week from Monday through Friday, they’ll call the test on Friday and implement the winner, and then all of a sudden over the weekend their site goes crazy and their conversion rates are way lower than they normally were. That’s because they didn’t test that winning design on the weekend. So you might have something that wins during the weekday but not during the weekend – I see that all the time – and so you want to get your 90% statistical significance, and you want to run it for at least a full week.
Rich: Alright, I feel like I skipped a step as I’m thinking about this, because once we decide what we want to test – like you mentioned a couple of software services – so what do we do. Are we hiring a company like yours, are we able to get some software, so any of these things play nice with WordPress, and do any of them have a negative impact on the SEO? I know that was like four questions bundled together, but answer it however you best see fit.
Chris: I’ll try to go through those in linear fashion. So in terms of WordPress, pretty much every testing tool works really well with WordPress. Again, these are all just – like any testing tool out there – they’re just implemented with a single snippet of javascript that you put in the header of whatever page you’re testing. And then the testing tools will usually have some sort of interface that allows anybody – and I mean non coders or non designers – to do some basic things.
So for example, if you use Optimizely they have a visual test builder where you can click on any element of your site and move it or restyle it or delete it or add something to it. Those are very basic tests. If you want to get into more advanced testing or bigger scale testing – like I was mentioning earlier – like full on redesign testing, at that point you’re probably better off either using internal resources. So if you have a designer or a developer then you can use your internal guys and they can usually figure it out pretty quickly, or hire an agency to help you out with the test implementation.
What happens a lot of times is business owners or marketing managers have this big idea for a test, they’ll go in and start to implement and realize it’s a lot more complex than they think. Then if they have internal resources they’ll go in and talk to their developers and their designers and tell them they have to put this in the queue, and then you have to prioritize that against everybody else’s stuff. And then people are wondering why you’d prioritize a test like this when they don’t even know what the impact is going to be.
The reason I actually started my company was a lot of larger company’s resources are really tied up. So I was in house for a company and was running tests, and it would take me two to four weeks to get one of my developers to work on one of our tests. And I was like, there is no way you can keep running effective tests if you have to wait four weeks in between every one of them. So we actually ended up hiring an agency to help us so that we could quickly move from one test to the next.
So in terms of testing tools, pretty much any testing tool out there will play well with WordPress or just about any other development platform that you’re on. I usually suggest Optimizely, if people haven’t tested before, because they have a free account that gives you up to 50,000 visitors for free. So you can run most tests inside of their free account.
Rich: We talked a little bit about SEO. So I’ve heard rumors that sometimes when you’re doing these sort of tests it can have a negative impact because Google comes by and gets confused. I’m not sure if there’s any truth to that or not.
Chris: So there’s not any truth to that. I’ve run thousands of tests that judge this impact. So first of all, Google really likes A/B testing, in fact, they even have their own A/B testing tool. It’s not very good, but they do like A/B tests. The reason they like it is because what you’re essentially doing is you’re optimizing for the user. And that’s really what Google cares about. Google cares about websites that are providing a good user experience, so I’ve never seen them penalize someone for A/B testing.
Now what happens after you test, if one of your variations has no text on the page and the page before had a ton of text, well it obviously might affect your rankings there if your content was really valuable. But if what you’re doing is you’re just optimizing for a better user experience – so you’re testing different images, different buttons, different colors on the page, different layout – but the content is relatively the same, then you’ll probably get a better benefit. Like I mentioned before, they do look at onsite metrics, and so if you have a better user experience you’re probably going to have lower bounce rates and higher click-through rates, and you’ll probably have higher time on site and on page. And those are all metrics that Google likes to look at. So more often than not I see a positive impact on SEO. Like I said, the only time that you might potentially worry about a negative impact is if you are removing large amounts of text from a page.
Rich: Makes a lot of sense. This has been great and I know I could probably ask questions for a while, but I know we’ve got to let you go. Where can we learn more about you and what you’re doing online? And if people maybe want to outsource some of this to your company, where can they find more out about you?
Chris: Sure! So my company website is dayleyconversion.com. You can also look me up on Twitter at @ChrisDayley. Or check me out on LinkedIn or anywhere else and I’d be happy to chat with anyone or answer any additional questions.
Rich: Awesome. Chris, thank you so much for your time today.
Chris: I appreciate you having me here.
Show notes:
- To learn more about Chris Dayley and what his company does, check out his website, or follow him on Twitter.
- Chris mentioned a couple of free programs to use to help you out with your own testing:
- Rich Brooks is not only the President of flyte new media, he is also the brains behind the Agents Of Change Digital Marketing Conference. If you like the information and guests here on this podcast, you’ll appreciate the impressive lineup of speakers he’s assembled for this year’s conference. Have you gotten your tickets yet?
