How To Split Test Tiny Website And Landing Page Elements To Gain More Organic Inquiries
Our latest video provides a step-by-step process for identifying website and landing page elements that don’t convert so you can fix them through split testing.
As Shane Sparks, Co-Founder, and Sasha Tiede, Front-End Developer and Conversion Rate Optimization Analyst explain, this small but crucial part of the process will make or break your test. It’s also where most beginners to split testing make mistakes.
Watch this video to learn how to:
- Identify problem elements on your website and landing pages using Google Analytics and other website data and insights
- Hone in on the opportunities to increase conversion presented by those problem elements
- Design hypotheses based on your data that will consistently produce meaningful test results
- Replicate this simple but powerful process at your school
Follow their process to get big wins from testing tiny web elements.
This video is a recording of a breakout session at our 2020 Enrollment Builder Best-Practice Virtual Conference.
Read the full video transcript below.
Sasha Tiede: Yeah. And it’s focused around a website as well.
Shane Sparks: Yeah. So applicable to really all aspects of the school, but we’re going to focus on websites today, just given we don’t have a lot of time. Okay. Sasha, you’ll share your screen and fire up or PowerPoint.
Sasha Tiede: You guys see my screen? Really. There we go. How about now?
Shane Sparks: Yeah. Awesome. Okay. Well, let’s get going, Sasha. So we’re gonna split testing 101.
Sasha Tiede: Yeah. So, the topic I was assigned is how to split tiny website elements to get more organic inquiries. It’s quite a daunting topic, to be honest, to summarize. There are so many variables that can affect your outcome. The path to success will be different for everyone. A website consists of so many testable elements. We have page forms, navigation, links between pages, and on and on and on. On top of that, each website is unique. So basically, we have countless variables to consider and no clear solution. From a mathematical standpoint, this problem would fit nicely into something known as chaos theory. So ER has, has a wealth of knowledge about this kind of stuff. Over 13 years, we’ve got case studies, a huge testing library, a lot of stories to share. So it was really tempting just to jam as much of that knowledge as possible into this 20-minute webinar for some fun case studies, and hope some of it sticks and is useful to you. However, instead, I’d like to talk about a tiny bit of the split testing process that is crucial for any success and is often executed very poorly, and I hope you find it more useful than case studies.
Shane Sparks: Can I just, yeah as well. What I most appreciate about Sasha is she has a degree in math, very analytical mathematical mind that counters my, promoter marketing minds. Right. You know, we’re, we’re on the opposite ends of the, I don’t know that the social styles model and what I most appreciate about Sasha is when I try to try to jam in an idea that she thinks is dumb. She could see she kind of bristles and then pushes back despite her feelings that she should probably just go along with their boss. So I view that as great integrity. So, operating with creativity, sorry to interrupt.
Sasha Tiede: Yeah. So this is the basic structure of any split test. You start with a goal to increase website conversions. A hypothesis, if we change the button color, then our web conversion rate will increase. Then you test it, and then you get some results. What I’d like to focus on is how to get from your goal to a hypothesis that will actually give you meaningful results. Why I think this is so important the only way you’ll ever see big wins from testing tiny web elements are if you get this part right, it is the most important part of the process. So if you, if you don’t do this right, then the rest of it doesn’t matter. And I also believe it’s where most beginners, myself included making the most mistakes. And lastly, the available resources that I’ve found online are, are poor in the sense that they don’t really offer any tangible process that you can replicate and use in your own kind of testing.
Sasha Tiede: What is the most common approach, especially when you’re first starting out, you start with a goal and you go directly to your hypothesis. So you’re essentially starting with a goal and going directly to a solution.
Sasha Tiede: In testing, we call this the let’s throw it, let’s throw stuff on a wall, and see what sticks approach to split testing. so you began testing under the assumption that you know, nothing and you just keep throwing stuff until you see what sticks
Sasha Tiede: The problem with this approach is, in general, you are not likely to learn anything. They’re very highly unlikely to produce a lift cause you’re not really considering, in your hypothesis, how the changes that you’re making will actually change the user’s behavior. And they’re not likely to solve a real conversion issue with your website, like what makes us think that this button was a problem?
Sasha Tiede: So why is this approach so popular well, button color is an excellent way to describe AB testing. It’s not really a meaningful way to improve your website. And why is it so common? Well, myself included when I started, Oh, I just started by reading a bunch of case studies and you get things like changing the button color from red to green equals 72% improvement, yada yada, yada. but the truth is there’s a lot of hard work and knowledge and experience that goes into those tiny changes, for, for huge result type case studies and case studies just kind of show the fun part of it. And then you go to, you know, after you’ve read a bunch of case studies, okay, well, what should I test and then you go back online, you Google, what should I test and then you get expert advice from people like Neil Patel.
Sasha Tiede: Who’s, he’s an expert in the space, I guess. And he writes articles like “obvious AB tests You should run your website”, serif font or, Sans serif font, number of columns on your landing page, background, and patterns and images. Those are the top three. And if you want to get gains from these kinds of tests, you need like millions and millions of visitors a month. And like, I, in my experience anyway, like it’s not, it’s not obvious that that’s the case. When you start reading all of these, all of these articles. Here’s an example in practice, this is my first or second split test. So this is the homepage of one of our clients, and the goal was to improve web conversion rate. So this was my solution, changing the form design on the website’s home page will address, low web conversion rates. Why I, I don’t know.
Shane Sparks: All right. So sorry, I got kicked out for a sec. What motivated this What was the big idea here
Sasha Tiede: What the test
Shane Sparks: With the test with the forms
Sasha Tiede: What was the big idea I was throwing stuff on a wall pretty much right. At the point. Yeah. Yeah.
Shane Sparks: So, Hey, so you had a mandate go to go improve conversion rate. How about it and their creative spirit. And the moment was, Oh, what do I do And I, Hey, let’s make it look a little bit better, maybe.
Sasha Tiede: Exactly. Yeah. I didn’t. I personally didn’t like the way the form looked. That’s what it was based on my personal opinion yet. Totally. Yeah. And of course, it wasn’t a real problem in the first place. So nothing happened, the results were inconclusive, and I didn’t learn anything from the test except for maybe not to do tests this way.
Shane Sparks: Yeah. I think you’re also a little hard on yourself. And if we just tie this into organizationally our journey through this kind of navigating how to test and find actually useful things as a company, we started, we started certainly on just very simple AB tests. There’s a red button better than a blue button. Is this headline better than that headline? You know, it was two forms better than one form and, you know, and they would bear fruit, but, you know, we would show improvements on some of the tests, a bunch were inconclusive, but what we found or what was always frustrating is that it wasn’t cumulative. Right, so if we had four successful tests and they each showed an improvement, you know, button in the forum and a headline and a this, those didn’t accumulate together, right. They weren’t, we didn’t get exponential gains off that we just, it would level out and we would get some improvement, but that it would kind of stall. And that proved frustrating. And that’s partly, partly what motivated changes are getting a little deeper and more thorough in our testing regimen. The AB tests are certainly gratifying. It’s fun to know if, you know, one thing’s better than another, but ultimately it’s superficial. I think that’s kind of your point.
Sasha Tiede: Yeah. That, and I mean, to test these tiny superficial elements, without any knowledge and not enough, like website traffic to get, you know, you’re, you’re very unlikely to get gains on, on small things that have no thought behind them, basically. So how do you get big wins from small changes in your website and I’m not saying it’s not possible. You just need to put some thought into them. That, that question directly translates into this question. How do you identify a tiny element on your website that is causing a big problem and fix it? So without a process to hone in on the tiny problem elements, you’re essentially trying to find a needle in a haystack. So, in short, the answer is to use your data insights and analysis to design hypotheses that produce meaningful results. So your data and your insights, you can use to hone in on problem elements, and then you can use a heuristic analysis. I’ll talk about that briefly in a second to come up with a possible solution.
Shane Sparks: And so in the data section here, so I assume everyone has some kind of Google analytics, so they can, you know, mine and see what, what is going on. Can you explain what a click map is and how it’s different than a scroll map, or we don’t have to go into every single one, but just walk us through it a bit more.
Sasha Tiede: Sorry. That, that was my original plan that I’m looking at the timer, and I’m like, eh, but, so yeah, so from the data, almost everybody will have, Google Analytics and the rest will come. If you, you know, once you start establishing a testing program or you see the need for them. So a click map, we’ll just show you where users are clicking on your site in terms of like percentage. Scroll maps are actually quite useful. It shows what percentage of users get, how far down your page. Form analysis shows like where users drop off on your form What, what fields are causing hesitations session recordings are literally just recordings of how users are interacting with your site. So you can watch individual, browser site,
Shane Sparks: Super interesting, by the way, right. You know, when you learned a lot from those by understanding that people tend to, they don’t really read things in a linear way, right It’s not a top to bottom when you watch somebody interact with a program page or a landing page, they’re skipping around all over the place, like, you know, start at the top and come back. So it’s really a kind of a non-linear interaction, which informs the design. Right. It’s super interesting.
Sasha Tiede: Oh, for sure. Yeah. It can be time-consuming, though, to go through the session recording of your session recording. And then just sorry, quickly under, under the insights. So test archive, you won’t have that if you don’t have a testing program yet. Survey data is just like the voice of prospect kind of data, but you have some business knowledge, like, you know, what do your prospects want out of a program or, you know, what, what motivates them to pick a certain career path. And just very quickly the heuristic analysis, there’s many different models, but a really popular one for CRO is a lift model by a wider funnel. And basically, they just say analyze your, your page or your element or your form or whatever you’ve honed in on, and optimize the conversion drivers, which are relevance, clarity, and urgency, you know, is it relevant is my message clear is there any sense of, like, should they take an action now and then, reduce conversion inhibitors, which is anxiety and distraction. So what, what, what elements could be distracting my user from taking the next step or, or converting essentially,
Shane Sparks: And anxiety would be, in the context of the school that would be around credibility maybe, right Like this play with testimonials, employer endorsements, those kinds of things, or certification?
Sasha Tiede: Or like even, I mean, probably the most simple example would be a phone number field on form. Oh, you know, I’m fine picking my program or, or the campus, but when you ask for a phone number, I’m like, eh, don’t want to do it.
Sasha Tiede: So, in a nutshell, to go from a big goal like this to a hypothesis that will produce meaningful results, one identifies the opportunity. So you gotta find what, like where people are spending their time and where are they converting. We don’t want to optimize a page that gets no traffic or a page that is highly unlikely to produce a conversion, like a blog post. And once you’ve honed in on, on that page or element, find the problem, this is kind of logical. And then, once you find the problem, come up with a solution. So, and you’re gonna do all of this stuff in the middle here using the data insights and the heuristic analysis. So I’m going to quickly just go over an example or two if we have time. So the goal here would be to improve website conversion rate, and I find it super useful to, you know, map out your site visually and then fill in Google analytics data.
Sasha Tiede: So you can try to hone down on your opportunity for optimization. So from Google analytics, you can answer questions like where, where are people spending their time paid, you, you report. Okay. So in this example, we know that they’re spending their time on the homepage program pages and our blog posts. Cool. Where are they converting? Well, they’re converting, mainly on our program pages, which is awesome. It makes sense on our home page and very little on our blog posts, sorry. I’m like all of these conversion rates like I’m just focusing on the pages that get traffic here. I’m not focusing on the stuff that doesn’t matter at this point. And then you ask, how do users navigate your website? You can also get this through Google analytics, our top past year of homepage or program page or homepage blog, post, or a program page to blog post, where are they spending their time and where do they leave our site from so where are they engaged Looks like the blog posts mostly, which makes sense, not really to engage on our homepage, but that also makes sense because they’re trying to get somewhere else kind of thing, in our program page. And then is it right. So, where are people landing and then leaving the site from?
Shane Sparks: So in this example, Sasha, we’ve thought higher engagement on the blog pages, but low conversion and homepage we’d probably assume as a catchall, you know, it’s just, it’s an aggregate of the rest of the website. They just kind of ended up back there and then make a decision to inquire. And so the program pages are the highest converting pages, with the lowest lower exit rate. Well, not middling, and so if we’re not driving them to the program page, that’s, we’re, we’re missing an opportunity. Is that what you’re saying
Sasha Tiede: Yes, essentially. Thank you, Shane. Yes, very well put, thank you. So your blog post looks like there’s some opportunity here. You know, should we, should we optimize the conversion rate? Well, probably not because people, I mean, just kind of common sense. It’s just an article. It’s not. No, but a specific program we offer, but we do notice that there’s no, there’s no arrow back here, right people are visiting our program pages and people who visit the program pages, they care about blog. So people who are reading our blog might care about our programs.
Sasha Tiede: So the opportunity would be, sorry, the user journey from blog posts to a program page, and then you go over and you have a look at your blog, and you realize that the only, the only link back to your program pages from your blog posts is for the main navigation and user spending a lot of time on the page. So odds are, they’re probably getting to the end of your blog post, where there is no navigation back to your program page’s point. So a solution adds links from your blog posts to your program pages, and then your hypothesis just naturally falls out of there. Oh, if we add links from our blog post to program pages, then we’ll increase the conversion rate because this will make the user journey to convert shorter and easier. I hope this is,
Shane Sparks: And to clarify: a hypothesis is simply a, we think this might be the case. Like it’s just a scientific inquiry, essentially. Right so we, the, the opportunity problem solution or the pre-steps to get to the thing that you think may be true, which is what initiates the test, is that right?
Sasha Tiede: Correct. Yeah. But it can be educated in a form through your data and you can hone in on the problem as much easier this way.
Shane Sparks: Versus, hey, I wonder if blue’s better than red, right. Or, hey, I wonder if this photo is better than that photo, right. That’s the this is the mature model versus a let’s guess kind of a testing model.
Sasha Tiede: And it’s quite effective, especially, you know, if you’re just starting out, you’ve probably got some serious opportunity. And just by making the site map and filling in analytics data, you can, you can find stuff, and odds are it’s, your, your tests are going to have a real impact on, on user behavior as long as you’re honing in on the right thing.
Shane Sparks: No. Cool. Right on.
Sasha Tiede: I’ve just got one more similar kind of example to illustrate the same thing. So let’s just assume it’s a different site, the same goal improve conversion rate, in this case, are we are traffic as the homepage and our medical program pages and our IT program pages. So where are you just converting? That kind of makes sense?. Just to note, so like this, these two pages right here is one of our sites, like it’s a, it’s a general program page. So it talks about all of the medical programs or all of the IT programs that we offer. And then there are links to the specific pages. And this is how users navigate our website, typically homepage to our general page or homepage to a specific program page. And we’ve identified that through Google analytics as well, people are relatively engaged on our program pages.
Sasha Tiede: Good. And then we do the exit rate. So where’s the opportunity from here? We’ve got a lot of traffic going to it, a lot of traffic going to our medical program pages. But it seems like there’s a, there’s, there’s an issue here. A lot of people are going to this IT general program page that talks about our IT programs, lists the specific programs and links to them, and people are just leaving there. Plus, these pages convert quite well when people do get on there. So where’s the opportunity? The click-through rate from our IT general program pages to a specific program.
Shane Sparks: And it’s also ancestry. So if we’re looking at this kind of diagnosis, it’s also interesting that the time on the page is actually longer on the IT programs. And can we assume there’s a correlation between those things: the longer somebody is on a page, the more likely they are to convert
Sasha Tiede: Yes. We, we have done some analysis on that. I forget exactly what the sweet spot was, but, yeah. The longer users spend on your page to convert. Yeah. Yeah.
Shane Sparks: So, yeah. So an interesting one. So we’ve got the, IT, people are spending more time-on-page. We’re having trouble, we’re losing too many of them before they get to the page. Like the individual program pages. If we know if we get there, hey, 70% of them are 7% are going to convert. Is it fair to assume that the, I know the psychological makeup of a medical person would be different than an IT person
Sasha Tiede: Well, yeah, that’s kind of what you got to hypothesize, I guess.
Sasha Tiede: Yeah, so looking at, so like our IT general program page and our medical program page, in this case, they both share the same layout. So we can kind of roll out that it’s a problem with the design because they both have the exact same design. So then you can probably, confidently say that, that it has something to do with the copy and the value proposition on the IT general program page that is not enticing enough for them to click through, to learn more. So the solution is to change the sort of, Oh, sorry. Why did I get to the, the value proposition that was doing a heuristic analysis? So, I went, I, I looked at that page, and asked like, is it because is my message clear are there, is there anything causing anxiety, reasons people should hesitate? And I honed in that it’s most likely the value proposition for these specific prospects. That is the issue that I think I’m not, I’m not reaching them.
Sasha Tiede: And then the hypothesis will just naturally fall from there and how you go about changing the value proposition on, on that page. I mean, there’s so many different ways you can go about it. And that’s where your, your tests are born, but at least you’re, you’re focused in on the problem area. and likely, you know, not just the problem area, the problem with the copy on this page and the value proposition within the copy of that page. So you’ve really honed in on your opportunity. And now it’s just about testing different value propositions for the prospects. And if you do that effectively, then you’re going to learn a lot about your IT prospects. And then you can use that learning on those learnings on other pages or your landing pages, or your calls to action or your ads or whatever.
Shane Sparks: And, and these kinds of insights inform other parts of your marketing, right? So you’ve mentioned landing page tests. We’ve got Chris Cunningham here was one of our Adwords analysts. So Chris and Tammy and Chris, and the Adwords team use insights that we get and other areas on, you know, how either copy points that resonated or quirks of the market. You know, the things that would make an IT person different than if someone who’s pursuing an allied health medical kind of program. And they can use that to inform the ad buying and the, you know, just shaping the pitch and the, you know, the ads and the keywords and the sales proposition and other parts of marketing. Right. So it’s not just about converting more traffic off your website, it’s about gaining insight into your market and how best to communicate with them. That fair?
Sasha Tiede: Totally. Yup. Yup. This would be a really valuable thing to learn, actually, this contrived example. Yeah. Pretty valuable test.
Shane Sparks: That was a good one. Should we have about four minutes Does anyone have any questions? Is this… Alan go for it
Attendee: Yeah. Sasha. Thank you very much. I really like what you’ve done that. 1. Can I get your slides because I’d like to go over that and utilize it myself? I think I got a meeting with you on Thursday or somebody from the team.
Sasha Tiede: Oh, me ABM, right?
Attendee: Yeah. So we’ll give our website to somebody who doesn’t know what they’re doing that drove into the tree and we have like a scroll monster, so now people scroll thousands of times you go right past the quiz, it’s dropped off a hundred percent and we asked the guy, what do you think? And he said, Oh, put it here. So now we’re going to come to you and I have all the faith in the world you’re going to save our sites.
Sasha Tiede: A lot of things to say about it. Yeah.
Attendee: Oh tell me about it. I was spin mad, anywho. Would I be able to get to the slide set of Sasha?
Sasha Tiede: Absolutely. Yeah. A hundred percent.
Shane Sparks: We’ll send it all out. Any other questions? So we could bring up an example.
Attendee: I lost her. This happened before when they bring up the slides, they suddenly get muted?
Sasha Tiede: Maybe I have to do something.
Attendee: Well, I had a really basic question, and I think I know the answer already, but I’m just checking myself. When doing a split test, how much should you keep it kind of in a pure AB test or how much can you fragment it out? And to kind of give an example of what I’m asking, we recently did a test where we have a PPC microsite, and we also have launched landing pages and we split our PPC traffic, you know, 50 50. The thing that we’ve run into is that our landing pages are, producing as far as leads, I would say probably about 65% greater than our PPC microsite. However, the conversion, as far as enrollments, is way under what the PPC micro-site does. So we’re wanting to tweak the landing pages to see if we can keep the lead volume, but increase the enrollment rate. Would it be best to kind of tweak that now and still have the traffic split 50 50 between micro-site and landing pages, or should it only be a pure test between one or the other without a third option?
Sasha Tiede: I’m sorry, can you just repeat the last part of that there
Attendee: Yeah. Sorry about that. Yeah, we, so we actually were wanting to throw a third variable in there basically to simplify it. We did an AB test. We got some good results, but we’re wanting to tweak it even more. So is it feasible to throw in a third variable or should it strictly be an AB between one or the other?
Sasha Tiede: It, it kind of depends on how much traffic you get, you know, to have a third variable in there, can be effective if you, if you have a ton of traffic, like our, our sites get like on average, maybe 5,000 bidders, visitors a month, and we won’t test more than two variations at a time using that traffic. I mean, I guess it also depends on the changes that you’re making. So like, if they’re huge changes and you can still get meaningful results from…