When a visitor comes to your website, is your content compelling enough for them to buy? Are errors on your website turning potential clients away? Are you A/B testing correctly? There are many factors that play into boosting your conversion rates, and Khalid Saleh is an expert on turning people from interested to a lifelong customer. Khalid is the CEO and co-founder of Invesp, a leading provider of conversion optimization software. We discuss how you can learn to create an online presence that people trust, and how to use the data that can change your business forever.
Hello, and welcome to Marketing Speak! I’m your host, Stephan Spencer, and today, we have Khalid Saleh, and he is the CEO and co-founder of Invesp. It’s a leading provider of conversion optimization software and services so in other words, Khalid is a conversion expert. He knows all about how to really ramp up your conversion rate and turn a lot of lookers into buyers. He’s also co-author of O’Reilly’s Conversion Optimization: The Art and Science of Converting Prospects to Customers, which is a great O’Reilly book so, he’s a fellow O’Reilly author. In 2016, he’s now gearing up to launch his technology startup, FigPii. FigPii is a one-stop platform for everything related to conversion rate optimization. He is a recognized expert on marketing strategies, presented at marketing conferences across the globe, and he’s also a frequent guest on media outlets including CNN, BBC, Sky, France 24, MSNBC, New York Times, National Public Radio, and more. Khalid, it’s great to have you. Thanks for joining us!
Thank you for having me. I’ve been a follower of the podcast for a while now. I’m glad to be on it.
Oh, awesome! Awesome! Let’s start by-you know, this is all about conversion so, let’s talk about some of the biggest conversion mistakes you’ve seen. What are some of the biggest train wrecks? You don’t have to name any names, but what are they doing that’s really stupid?
Well, I can classify them into several categories, but I think one thing that I see consistently is that people are always looking for quick fixes, short-term hacks, and they’re just looking for something really quick without really understanding that conversion optimization is just like any other type of marketing SEO and PPC-just traditional marketing. It’s really not a light switch that you turn on and off. You really have to have a strategy, you have to think through the strategy, and then implement that strategy. That’s one area that we see nowadays because everybody’s into conversion optimization yet, they’re really focused on the tactics and forgetting about the bigger picture. Something else, now, if we get into actual projects, I’m still shocked-we’re in 2016, not many people know how to look at the data, whether it’s Google Analytics, Omniture, or whatever software that they’re using to track their websites. I was working with a client recently. They have a conversion optimization team with five people and we say, “Oh, great! This is going to be wonderful working with them.” We tell them, “Okay, let’s look at your Google Analytics,” as there is an optimization team-an analytics team. I look at the Analytics, and not a single funnel set up. I’m like, “Wow, how do you guys figure out what’s going on?” The general metrics we have-three to four million visitors a month-and I’m like, “How are they navigating around the websites? What are they doing?” They just looked us and they’re like, “Well, Google does that for us,” and I’m like, “No, it doesn’t.” Those are kind of the two main things that I see nowadays. There are other things such as everybody’s into conducting AB multivariate testing. However, they’re doing it really without thinking. They’re just throwing things at the wall. We worked with a client recently, and we’re doing an audit for them before we get started so they’re like, “Wow, we’ve done tremendous amount of testing,” and I’m like, “Well, how many tests have you conducted in 2015?” They’ve conducted 155 tests, which is incredible. You’re talking about an AB test every two or three days and like, “Okay, this is great.” I mean, unusual because you’re concluding test really fast, and I’m like, “What was the uplift from 155 tests?” and they said, “Zero,” so it’s going back to the same point. They’re just throwing things without a strategy and just crossing their fingers and praying that something will stick.
Wow! That’s a lot of tests for not a lot of value or zero value.
Okay, now, let’s kind of go back to some basics here because you threw out a few terms that I think we need to clarify for our listeners. You mentioned, for example, AB testing and multivariate so let’s distinguish for our listeners what’s the difference between AB testing versus multivariate testing. I know the answer, but go ahead.
AB testing is when you create a new variation of a web page. Let’s say you have your home page and you say, at a very simple level, you say, “You know what? We’re not sure if the headline on the page works really well so, we’re going to create the second page,” a second version or a variation or challenge of that home page, and you can create, of course, multiple variations, and then you split the traffic between those different variations so that’s an AB test. A is the original, B-sometimes referred to A-B-C-they’re all challenging the original design. You set up the test so, they have those different designs. You split the number of visitors that you have between those different designs, the original and the challengers or the variations, and you compare the number of conversions that you’re getting so, change the headline-which of those variations or is it the original that actually generated higher conversion for us? That’s AB testing. Multivariate testing gets into a bit more lower level of detail where you say, “Okay, so I have my home page or I have a particular web page on my website, and I’m going to test different variations of the headlines,” so for example, I have the headline-let’s say a really silly headline, “Welcome to our website,” and I say, “You know what? That’s not going to work well. Let’s try something else,” like in a second headline, “Increase your sales,” the third headline, “Increase conversion,” so you’re testing different headlines, but then you take another element-let’s say, the image. The image on the side that shows, maybe, the founders. You say, “Well, maybe this is not working well. Let’s bring an image of a business meeting. Let’s bring a second image of, maybe, us conducting a one-on-one interview with the client or something like that,” so you take different elements and you create different variations so, it’s multivariate for each one of those elements. AB testing works really well for websites with less number of visitors and less number of conversions. Multivariate testing is popular, but amongst enterprise websites because you can test many, many different variations and many, many elements.
Right, so you have to have a certain amount of traffic coming in to have statistical significance on a multivariate test. It takes a lot more traffic. Actually, it’s all about the conversions happening, not really the traffic. If you have a lot of traffic, but you don’t have a lot of conversions, you still won’t get statistical significance in order to run the multivariate test and get useful data out of that, correct?
Exactly, exactly. Kind of back-of-the-napkin calculation that I always tell people. Look at the number of conversions that you have. For each 100 conversions, you can introduce a challenger against the original so, let’s say, you have 200 conversions-that’s actually the original or your original design has one-it takes 100 and then a challenger takes another 100 so, you can only conduct an AB-one challenger against the original. If you have 1,000 conversions, you can test nine challengers against the original so, that’s an AB test. The only time you start talking about meaningful multivariate test is when you start getting really more than 10,000-15,000 conversions per month, and not many websites have that. Lots of people are focused on the traffic, “Well, this is how many visitors I have!” It’s not about the visitors, it is actually about how many conversions you are getting within a month.
Yup. There are different ways or technologies around multivariate testing if you do have the number of conversions required. Taguchi Method is a terminology, but there are also different tools that you could utilize for a statistically significant multivariate test like Optimizely does multivariate. Visual Website Optimizer, I believe. Do Optimizely and Visual Website Optimizer both do multivariate testing?
Yes, both Optimizely and Visual Website Optimizer or VWO-their acronym. They both provide. For multivariate testing, they support it out of the box. What I’m seeing which is sort of interesting, I was talking to clients a couple of days ago-what I’m seeing is, although website have a lot of conversions-and we’re talking about enterprise websites that have maybe 50,000-60,000 even 100,000 conversions-they’re shying away from doing multivariate testing, and it’s an interesting phenomenon. The reason they’re saying, “Well, okay, so we’re changing all these elements, we’re trying different headlines, different images, different buttons, and different colors, and you have a winning design, which is great, however, you cannot rationalize why this design won.” Why is it that this particular headline with this particular image with this color worked out? Well, it just shows that, but we cannot rationalize so it’s difficult to build on that test result for a second test. What you see is, more and more companies are moving just to conducting AB testing with the limited sets of number of challenges against an original. I was mentioning to the client that I thought it was sort of interesting because Google used to offer a product that offered multivariate testing. That was about seven or eight years ago-Google Website Optimizer?
And at some point, they decided to stop-just kill the product and moved to Content Experiments, which only provides AB testing, and everybody wonders, “Why are they doing that?” and I think Google had the data to say, “You know what? Most websites don’t have the capacity to conduct multivariate testing, and if they conducted, they don’t know how to run it, and if they don’t know how to run it, they don’t know how to build the next experiment so, let’s just kill it.” Now, it took us about four or five years later to figure out, “Oh! That’s the reason Google decided to shy away from multivariate testing.”
Right. Let’s talk about which tools are the best sort of tools since we talked about briefly Optimizely, Visual Website Optimizer, Content Experiments inside the Google Analytics, which used to be called Google Website Optimizer. What are some of the best tools that you would recommend for a smaller business, for a larger business, and what are some of the mistakes that you see in implementing different tools?
When you peel off all the marketing and advertising, I believe all these tools look very close to each other. The three most popular tools that we see-of course, Google Content Experiments that comes free so, if you have Google Analytics, it’s available to you. However, it really requires some technical experience and know-how to be able to implement the test. It’s not the most user-friendly, which is unfortunate. There are three tools-Optimizely, VWO, and the third one I have to mention is, Convert. They all provide a visual editor. Literally, you give them your website address, they mail inside their editor, they pull up the website, and you can just drag and drop, click and point, and change the header without actually having to talk to your engineering or your development team. Lots of small businesses love those tools. Omniture is built towards a more enterprise level testing. Its interface leaves a lot to be desired. I haven’t seen Adobe updated in a while. It’s kind of the same reporting that we’ve seen for the last three or four years. There might have been some of the functionality, but they really have fallen behind. I’ll plug in our tool over here, FigPii, which is about to go into private beta next week or so. We’re also going to be providing an AB testing tool. We’re not providing multivariate testing. It’s going to have very similar functionality as Optimizely and VWO. Any of those other tools, you’ll get the same functionality.
Oh, wow! Very cool! Are you bootstrapping that or did you get funding?
Cool! When is the tool going live?
I was talking to our technical team just before this recording-we expect them to go into a private beta in about a week. We actually have some big-name clients who are already doing some testing on the tool. They need some very specific features and the nice thing when you’re working with a technology startup, they can build those features for you so we’ve been doing some of that. Next week, private beta about a month, and then we’ll go public beta, and then we’ll set a live production data after that.
Got it. Very cool! Congratulations and good luck! That sounds amazing. You have a book out that I have over here on my bookshelf. What are the plans for that because conversion changes maybe not as quickly as SEO, but it’s pretty fast-moving, right?
When did the book come out? When is the next edition coming out? How many editions have you had? Kind of give us a little lay of the land in terms of the book. Also, what would our listeners expect to find in the book, if they were to go pick it up at the bookstore or order on Amazon?
Very good. The book came out back in 2010, and if I recall correctly, Stephan, you’ve introduced us to the editor at O’Reilly.
Yeah, that’s right. It’s been a long time!
Yup, there you go! We both got the same editor, and you were helpful with that as well. We haven’t updated the books since it was released. It was translated into several languages, including Russian, which I did not know until very recently-about a couple of weeks ago. We just hired a lady from Russia. She was looking to doing some research and she came to me. She said, “Hey, do you know that your book is in Russian?” I’m like, “No. I didn’t know that.” She said, “Yes, it is. You probably want to check with the publisher.” I emailed O’Reilly and they said, “Yes, we’ve translated it to Russian. We forgot to tell you,” so they translated it into a couple of languages. We’ve been actually talking for the last couple of months about doing an update-the second edition. When we first published the book, I think we’ve had about 25 case studies at that point. We’re looking at adding a total of 100 more case studies to make it kind of a substantial second edition. Also, at that point in time, the book covers the conversion framework, which is the methodology that we use in our consulting practice to help companies increase their websites’ conversion rates so, the thought is to take all these case studies, all the mapping that we’ve done, and all the learning that we’ve done over the last six years since that first edition came out and incorporate that into the second edition just to kind of put things into perspective. When we published the book, we had conducted, I believe, at that point 400 AB tests, and since then, I think we have, as of the end of last year, done 3,000 AB tests. We’ve grown tremendously in the amount of learning and knowledge that we can share with anybody who’s interested in conversion optimization.
Wow! Is that book in progress or are you just about to start it? Are you in negotiations with the publisher to do this second edition? Where are you at with that?
What we did is, we’ve spoken with our old editor at O’Reilly and mentioned that we’re interested and they said yes, there is interest from their side, and we’ve started updating. We’ve been actually updating chapters internally for a while now for maybe about six to seven months so, we’re adding a lot. We’re kind of trying to see, “Okay, what are some of the case studies? What are the insights that we can pull?” and that actually was quite a much larger task than we expected because we said, “You know what? Let’s just pull all those 3,000 tests that we’ve conducted. Look at the design, see which ones we can use, what was the theory behind and the hypothesis behind every design, the screenshots, and the uplifts,” so it’s been quite an interesting and very rewarding experience because as we’re doing that and we’re updating different chapters in the book, we’re actually also writing lots-quite a bit of guest posts and sharing some of that knowledge on different industry blogs so, it’s been rewarding in multiple ways.
Mm-hmm. Right. What would somebody expect to find in the book? What’s the structure of it? What are some of the key takeaways that they would get?
Sure. The book starts with laying out the math behind conversion-what are some of the criteria that you want to look at? What are some of the things that you want to evaluate on your website when you are starting a conversion optimization process, and if you’re running specific page campaigns. All of that is laid out in the first couple of chapters in the book. From there, the book goes on to discuss the conversion framework and the six elements of the conversion framework. What we say is, in order for you to persuade visitors to convert, you have to understand the elements that are really going to be helpful in converting that visitor, or convincing them to convert, or the visitor saying, “You know what? I’m not interested. I am going to leave.” Those elements are broken into two categories. There is what we call, “website-centric” elements or factors and then there are “visitors-centric” factors. Visitor-centric factors are the factors that you don’t have control over. They’re really based on the visitor so, the visitor persona, for example, people shop in many different ways, and so clients tell people, “I’m somebody who is very spontaneous. Any e-commerce website would love to have me. I shall buy with a credit card in hand and ready to place an order very quickly.” A lot different than, for example, my partner who takes her sweet time, reads every piece of copy, and reads the reviews. It’s a completely different experience, and you have to appeal to both of us. Also, in the visitor-centric factors, there’s the complexity of the sale. In your website, if you’re selling items with an average earned value of $2,000, it’s a lot different than selling items at $30 on average earned value so that’s something else that you want to keep in mind-the visitor persona and the complexity of the sale. Then, the cycle-the “Where is the person landing on your website?” Are they registering by? Are they still early on in that funnel? You want to keep that in mind. We discussed each one of those in a chapter of its own, and then we discussed also the website-centric factors. Website-centric factors include things such as trust-if I don’t trust you or if I don’t have confidence in your website, there is no way that you going to be able to convert me. They’ll include things such as F.U.D’s or Fears, Uncertainties, and Doubts. We all have them, correct? You land on a website looking, “I don’t know this website. Is this the right place for me? Do they have the right items?” You want to be able to anticipate the visitor’s FUD’s and address them on the website’s different locations. Then, use of incentives. How do you incentivize your visitors to actually act right now as opposed to saying, “No, I’ll bookmark the website, and I’ll leave. I’ll come back to it later.” Then, finally, engagement-where you move from “Okay, I’ve converted the visitor,” into a long-term relationship, “I’m bringing the visitor back here, and he/she is referring his/her friends, colleagues, and family back to the website.” We take each one of those factors, present a theoretical discussion around it, the theory behind it, and then we give practical examples on how this would be implemented on the websites.
Right, and so these different factors, you then come up with experiments based on like, “Okay, so inside of incentives, do we try to incentivize people to take an action based on a sale or some sort of clearance? Having taken action based on urgency and scarcity, I guess, are your two best friends, right?
Let’s dive or deeper into that-how do you take these website-centric factors and turn them into useful experiments?
The first thing you want to decide on is-what pages do I want to have on my website? Do I want to look at my product, let’s say you’re an e-commerce website-should I be looking at my product pages, my category pages, my cart page, checkout, and my home page-all these present an opportunity for you to optimize and increase the website conversion rates, and to answer that question, you have to look at the analytics. Dig deep and look at how the visitors are flowing. Let’s assume that you’ve done that so you have decided, “Well, okay, it’s my product pages,” the next question is, “Well, how do I use those different conversion framework elements to really determine what changes should I be making on the website?” The way we do it is, we take the page-the web page-and we start analyzing it from the perspective of each one of those elements. We look at the page and we say, “Okay, I already trust the elements and trust breaks into 52 different factors, are they present on the page? Are they too present on the page?” Sometimes, when you try and emphasize that you’re trustworthy, you’re actually going to have a negative reaction to that like, “Hey, they’re overdoing it over here, what are they trying to hide?” You analyze each one of those elements. You look at, for example, let’s say, incentives. I have a product and you want the visitor to act right away when they land on the page. Well, does the visitor-when you’re looking at the pages-the visitor thinking, “Well, there’s nothing special. There’s no sale. I can come back next week, and I’m probably going to find the same product.” You evaluate that against, for example, let’s say, “We only have three items left in stock,” so that’s scarcity or urgency-“Order in the next two-and-a-half hours and get it delivered to your house tomorrow.” There are different things that you want to do. Are those present on the page? F.U.D.’s, for example, fears, uncertainties, and doubts-you’re evaluating the page from the perspective of F.U.D.’s. What are the biggest concern that people have when they land on our product pages? Maybe they’re worried about pricing. Maybe they’re worried if this the right product for their needs. We need to be able to address those, and as we’re evaluating the page from different elements and perspective, we’re listing the top problems that we see on the page. A good conversion optimization specialist can take a webpage and pinpoint anywhere from 30 to 50-and I’ve seen even 150-problems on a page, which is quite a bit, and you don’t want to fix that many. What we do then is, we prioritize those items. There’s a methodology that we go so trust is very critical, then addressing F.U.D.’s, then after that, you want to move into incentives, and then kind of one of the lower elements is engagement, and where the person is, and the buying cycle. When we figure out the problems on the page, that’s when we say, “Okay, now, it’s time to create a new design to address those top three or five problems that we’ve seen.” Those designs now have the theory and hypothesis behind them. This new design is really going to address trust-so, because we added some social trust icons and we showed people how many people placed and order for this item. This other design addresses F.U.D.’s because it provided a guide for people’s “Is this the right item for you or not?” Each design that we’re introducing provides an insight and is solving one of the problems that we see on the page, and that’s when we move on to say, “Now, it’s time to.” We have the designs, we do our AB tests, split the traffic, and let the visitors be the judge of the quality of the designs. Which one works better for them? Is it the original design or is it one of our designs where we fixed some of those other problems?
Right, so some of the tests are kind of unnecessary because you just know that they made a bad choice with whatever design or layout element, and you just need to fix it, right? There are some things that are just braindead simple-you have to fix these things-and then there are other things that are like, “Okay, well, we’re not sure which one’s going to perform better so, let’s run a test.”
How do you do that? What’s the process for that?
Whenever we start with the client, one of the first things that we do is, there’s quite a bit of qualitative research that we spend almost anywhere between three to six weeks where we’re doing it in different iterations. At the very simple level, we take the website and we say, “Okay, what are the most common trends that people go through when they’re placing an order with a website?” Maybe somebody searches for an item, they click on your link from a Google organic search, they come to the website, and they go through the checkout process. That might be very common. Another thing is, they might be clicking from Facebook, landing on a landing page, and placing an order. A team of conversion optimization specialists, which usually is anywhere between three to five, is going through the process, mimicking exactly how a visitor goes through the process, and they’re recording everything that they see on the website. Are there things that are throwing them off? We start making a list as we’re going from one page to the next. This list can go fairly large. We just did it for a recent client, and it refers to just a single iteration. We came up with, I think, 27 different items of things that we see on the website, and this is even prior to talking about conversion framework. Then, we started quantifying those and classifying them-this is a must-fix right away or this is a testing idea. For example, let’s say, we land on our landing page, a product page, and the dropdowns are not working, or the image just sticks over the whole page. You don’t want to test that. What you do is you say, “This is a must-fix. We need to fix it quickly,” so we add those for us into a report that we give to our clients, which we call “Stop the bleeding” reports. It’s where we say, “If you don’t test it, fix it because it’s not about usability, it’s not about persuading visitors, it’s about making the website actually work as anybody would expect it,” and then we move on. After that’s done, we move on to the actual testing on the website.
So, a “stop the bleeding” report might be how many pages? And then, a full conversion analysis of all the opportunities and recommendations for tests and so forth would be how many pages? Give us a sense for what these deliverables are like?
Yeah. It varies. I was just talking to a client yesterday. They just signed up and they’re like, “Well, how big is our report going to be?” I tell them that it truly varies from one website to the next. I’ve seen websites that had a “Stop the bleeding” report of about one page with three items after we’ve gone through nine different routes. We found things, but we’re like, “Okay, the website is performing at par. We don’t need to be fixing when there’s a lot of testing ideas.” I’ve seen websites, literally, with 25 pages of things that we need to fix right away just to get the website to a usable level where, “Okay, this is fixed.” Of course, whenever you’re generating those reports, you’re doing them on desktop versus mobile because it varies from one website to the next, but with most websites, we’re seeing the majority of the traffic come on mobile. We’re talking about 50-60% of the website’s traffic getting to mobile.
What would be some of the differences between mobile users and the behaviors that happen versus desktop in terms of conversion?
This is sort of interesting because what we see and I’m talking about data maybe from about 20-25 different websites and we’ve kind of websites. The majority of their traffic nowadays, we’re seeing anywhere, I’ll say, upwards of looking 45% and upwards, some websites with close to 80% of their traffic is coming on mobile. However, when you start comparing the conversion rates, what we’re seeing is that, the mobile conversion rate is about half what you see on desktop. Now, some of this is due just to visitors. They’re checking and they’re still not used to actually placing an order on a mobile website although we’re seeing less and less of that. They’re just checking an item on the mobile experience, and they might even add it to the shopping cart, and then later on, they get home, they get to sit on their desk, they pull up the item on desktop, and then they convert. We’ve seen some of that. However, I would say majority of the websites are still struggling with actually building websites with a good mobile experience. Sometimes, they just merely take their desktop version of the website and just make it responsive, and that might not work all too well. The screen space on a mobile device is much more limited. The amount of information that the visitor is looking for is a lot less than what he wants or expects compared to a desktop. For example, a lengthy description that is available on desktop and product specs, on a mobile device might not make any sense so you have to be careful.
What are your recommendations in that regard?
Responsive is kind of the absolute minimum that I think everybody jumped into after the Google recommendations and the Google algorithm that talked about like the need to have mobile website design. From a conversion optimization perspective, we prefer having a separate mobile experience. If somebody is coming on mobile device, you direct them to an “m” or “mobile.yourwebsite.” We have a lot more control over that mobile experience. In order for us to do an AB or multivariate testing, we run tests either specific for mobile or for a desktop. They’re separate swim lanes. The visitors are coming in one device, they’re not impacted by the test we’re running. Looking at this visitor coming on desktop, they’re not impacted by the test running on a mobile experience. That’s what we recommend for most of our clients. Let me ask you, Stephan, although you’re doing the interview. From an SEO perspective, what would you recommend to our clients?
Yeah, that’s a great question! If you do responsive design, that’s usually going to be fine from an SEO perspective. If you do dynamic serving, which I think, overall, is the best experience because you have the opportunity to serve up more customized HTML for mobile users, as we just described. The little “gotcha” that people need to look out for, if they have dynamic serving, is they have to make sure that they’re sending the very HTTP header-that alerts Google to the fact that, “Hey, we’re varying the HTML based on the user-agent,” so Vary: User-Agent is the HTTP header that you would send before sending the HTML, and then you’re good to go. I’m not a big6 fan of mobile specific websites like “m.” or “mobile.” I think that’s kind of old school. It’s just more stuff to manage and so forth, but if you do do it, you just need to make sure that you bi-directionally link the two sites together-the desktop and the mobile site-and the desktop site’s going to have a route alternate tag that points to the mobile equivalent of that page, and then on the mobile equivalent, you’re going to rel=canonical to the desktop version. That will eliminate, hopefully, the duplicate content that might happen when you have two separate sites like mobile and desktop.
That makes sense.
Because you don’t want to have mobile results in your Google search results. You want all that collapsed and only for the desktop versions to show, and then you’d do device detection to see if the users on the mobile device, and then you would redirect them to the mobile version of the site.
That make sense.
So, yeah, that’s basically how you do it.
Very good! Although I’m being interviewed, I have an amazing SEO expert on so I must ask this question.
Yeah, why not? Let me jump to a related question here-what are some of the key points that are listeners need to know about page speed from a conversion standpoint? Because it is important for SEO, and it’s becoming more important over time-you want a fast-loading webpage. However, from a conversion standpoint, it’s even more important. You could just destroy your conversion rate by having slow-loading pages so, what are some of the key points we need to bear in mind in regards to that?
Definitely page load time and speed time impacts conversion. When I tell our clients that what you need to figure out is, if the pages take more than five seconds to load, then that will definitely impact your conversion rate tremendously. You want to, at least, reduce it to five seconds so, if you have a webpage that’s taking 10 seconds, reducing to five seconds will see anywhere between 8-15% increase in conversions. Now, between five to three seconds, you also see an increase in conversions, but that’s not going to be as significant. You might see anywhere between 3-6%. Anything less than three seconds is going to have a lower impact, and it becomes like the ROI on the investments might not be there.
Right, got it! Let’s jump back to different kinds of A/B tests. You ran 3,000 AB tests at your company over the last number of years and you learned a ton, what would be some of your favorite tests that you would recommend everybody run like on headlines, on hero images, and on price points. What are some of the no-brainer kinds of tests that must be run on your site?
Some of the things-because you’ve already mentioned hero images are just a no-brainer. What works for one website may or may not work for another website. You really have to try different hero images so, with difference-a single person, multiple individuals in the image, where are they looking? Are they looking at the call-to-action? Are they looking at the visitor? We’ve seen really different data points for different clients and where that works better. That is one type of test that we like to run for our clients. In terms of different headlines, that would actually depend on the type of website that you have. If you are testing different headlines on landing pages for a PPC campaign that you’re running, definitely those headlines can have an impact. If you’re testing different headlines on a home page or a large website, it would not have as much impact. A third type of test and, perhaps, one of our favorites is the incentive-based testing. When I say incentives, by that there are three different types of incentives. Its factors that you add on a website so you can encourage the visitor to act and act right away. There is pricing-based incentives-that’s very clear. You’re giving a discount so you’re doing some price testing. It’s giving 10% discount on my items-would that increase my conversion rates by 30-40% or not? That’s one type of incentive testing. The other type of incentive-testing is what we call, “urgency testing.” In urgency, you are playing with the time. We’ve done some wonderful testing with a recent client where they were offering a discount to all of their website visitors-a 15% discount. We said that instead of throwing a blanket discount of 15%, why don’t we split the website traffic into two groups? One group sees the 15% discount while the other group sees the 15% discount, however, there is a counter that shows them that this special pricing or discount is going to run out in three days so it’s counting down. Just by adding that counter and by adding that urgency, we see about 30-32% increase in conversions. The last type of incentive-based testing is what we call, “scarcity.” In scarcity, what you do is, there is a limited quantity available of the item that you are offering. For example, if you’re an e-commerce website, this is very clear, “I only have three items remaining on stock. Order right now!” but even if you’re, let’s say, a consulting company-“We’re only accepting two more clients.” That’s a great way to encourage people to contact you right away. If you have a training program-“Only 10 spots available!” If you are, for example, running a conference-“Only 17 seats available or remain before the conference sells out.” In this case, visitors are actually competing each other, and when a visitor converts, there is less quantity available-thus, it also encourages the remaining visitors to convert even more. If I look at it from a large, kind of like zoom-out, there are three different types of testing. There’s what we call, “element level testing,” and a clear way of doing that is, testing different headlines and different images so you’re testing a single element. Those are really quick to implement, and they have anywhere between 4-6% or 4-8% impact on your conversion rate so that’s one type of testing. The other type of testing is what we call, “page level testing,” so now, you’re testing different elements on the page-the page layout, the colors, and you’re moving things around. It’s a bit more complex. It takes more time to implement, but has more impact on your bottom line so you can anywhere between 10-15% uplift in conversions. The third type of testing is what we call, “visitor flow testing,” and the best example of the visitor flow testing is testing a single-step checkout versus a multi-step checkout. Of course, there’s many other applications to this type of testing. This is the most complex testing to implement, however, it will have the most impact on your bottom line.
Yes, so the fast-action bonuses are the sorts of things that you see a lot of times in the info marketing space. It’s where people are selling training programs and so forth and like, “The first five people who register during this webinar will get, let’s say, a free event ticket,” so you get the online training that you’re buying, but you will get a free three-day seminar ticket as well if you are in the first five people or first whatever number of people who register.
And then, another one is, if you make the purchase while this webinar is still taking place or before the end of the day, then you’ll get this free event ticket. Those tend to work really well, and they really focus heavily on coming up with incredible deal sweeteners so that people just feel compelled to take fast action because if they don’t take fast action, the likelihood that they’re going to buy, let’s say, a $1,000 or $2,000 online training program is pretty small. Any other bonuses that you’ve seen that are particularly impressive to you?
I mean, all the other stuff that we like to do and it’s really showing, and I speak a lot about e-commerce because many of our clients are from the e-commerce space, but if somebody is in the cart page or product page, being able to upsell them-offering complementary products works really well. Another tactic that really works nice is, for example, if they offer $100 free shipping, and somebody added an item for $70 and they need that $30 more to get to that free shipping, instead of just telling them, “Well, add $30 more worth of items,” give them those specific recommendations-you have the database, you have the products, and you know what helps them. Let’s give them a few items that are $35 or $45 and really just make them click right away on the product page or on that cart page because this way, you’re actually not only increasing conversions, but you’re also increasing the average or the value that you’re getting from the same customers.
Mm-hmm, and how does email kind of figure into this? Because if somebody, say, they abandon their cart, and an email gets sent out, let’s say, 72 hours later, “Hey, you’ve abandoned your cart. Come on back and we’ll give you a discount or give you some sort of incentive for completing your purchase,” that’s kind of part of the conversion funnel, right? Where does this email fit in?
Email is extremely valuable. What we’re seeing is, typically, people who gave you their email and the people who receive your email are then more committed to your website so, typically, if you divide traffic between different channels, what you’ll see is that the email channel will convert better compared to your paid channels and your organic channels because there’s that relationship, correct? It’s about bringing people back. I like, especially for those cases, where a visitor abandoned their cart. They’ve already added items to the cart, and they decided to abandon, bringing them back is always a really nice way after you need to set the right triggers and figure out, “Okay, how long does it take people to buy? Is it 24 hours? 72 hours? Do we even send them an email a week later?” What you have to be careful with is, with offering them incentives so, do we offer them $10-discounts?” That’s definitely a good way to bring them back, but are we training our customers or visitors to add items to the cart and leave? We actually have seen a client recently who fell into that. People know that at the end of the month, they send these promotional item discounts on the item you added to the cart so people will just add items to the cart and wait until the last three-four days of the month, and then they’re getting the promotional discounts, and after that, they’re coming and converting.
Oh, that’s funny. Yes, you don’t want to train them to do those behaviors. You’re basically incentivizing them to do stuff that you end up discounting on so that would be a bad outcome. What is the kind of the optimal checkout process? You talked about briefly the multi-step checkout versus single-step checkout. If it’s multi-step checkout, what’s optimal generically? Three-step, four-step, or five-step?
I think there’s really no right answer to this, but what we found out is, the higher the average order value so, let’s say, you’re selling an item for $500 or $1,000 or $2,000, people feel very uncomfortable with a single-step checkout. They feel like, “Oh, one step, you’ve asked me for everything,” and they need that multi-step. I think it gives them some comfort as they’re going through the process. We’ve actually tested that this for a furniture retailer where they went from a multi-step checkout to a single-step checkout thinking, “Hey, this is going to help us increase conversions!” Now, the average earned value for people buying from their websites such as about $2,000, and when they made the switch, their conversion actually dropped by about 22%. We’re like, “Okay, let’s go back to the multi-step checkout.” If you’re going to multi-step checkout, three steps seems to be optimal with a final step to allow people to review the final order and make sure everything is good. They feel comfortable like, “These are the items I’ve added here,”, it all makes sense. “Let me go ahead and place my order.”
Yup, got it! What about colors? Does this make a difference? Of course, we’ve all heard of BOB, right? The Big Orange Button. Does BOB always work or sometimes, do green or blue work better as a button color? What about other types of color usages like maybe in a headline or in the price or whatever so, what are the best practices? I know your mileage varies, depending on the site and your target and everything, but generically speaking?
Yup! What I say, when it comes to color, what you need is, you want your calls-to-action to stand out, but also in a nice way because sometimes, you’ll use some really nasty colors, and I’m like, “Why is it like that?” They’re like, “Well, we’ve read somewhere-“and that’s where you need to go, “I’m not going to know. It needs to send out, but it needs to also flows with the rhythm of the website. I recall a study where Google tested, I think, 4,000 different shades of the blue registration account creation on the main website. It’s on www.google.com. They saw a 7% increase in conversions. Now, I say the change in colors, unless you’re doing something really bad with your website, would not have such a huge impact. There might be a few outliers here and there, but for most of its size, it has minimum impact.
Got it! What would be the most important piece of advice that you would give people who have been listening to this episode? You’ve dropped a bunch of knowledge bombs on people already, but what would be something that we haven’t already talked about that would really be a critical thing, kind of a take it home and apply it next week sort of piece of advice?
I say, here’s a small investment that everybody can make, which will pay back tremendously. Find 5-10 people, average users, who would actually use your website, invite them to a free lunch, and with a free lunch, tell them, “Hey, I want you to shop on my website,” and observe how they actually shop on the website. Watch them and don’t say a word. Just watch them add items to the cart, and go through the checkout. If you’re willing, I would give them a credit card to go ahead and place an order with your website so, watch that, watch the experience, and learn. You will be amazed seeing actual users use your website and the interaction they have with the website.
Right, and do you film that? Are you taking notes with a clipboard? What are you doing when you’re kind of extracting information out of that?
What we do whenever we do something like that, we’re actually recording the session on the laptop when the user is actually using it. We’re taking notes so, we’re writing them in a non-intrusive way because sometimes people will get these and be like “Well, what are they doing?”, and if they allow us, we can even film the user. Many people don’t feel comfortable, but at a minimum, you can bring in a screen-recording software so you can record their interaction, and you sit next to them quietly and just jotting notes down, seeing what buttons they’re clicking on and nothing is happening, seeing how they’re navigating from one page to the next, and making a huge list of items to fix on your websites.
Yup, that sort of idea of bringing users in or just a small audience and doing kind of an informal study on them-there’s a whole chapter dedicated to that concept in Steve Krug’s excellent book, Don’t Make Me Think, and for listeners, if you have not read this book, you need to-
Definitely! It’s a must-read for anybody who’s doing online or even offline, by the way. It’s just amazing because just the concepts, you can apply anywhere.
Yeah. So, how would people get in touch with you if they wanted to work with you or hire your firm to do conversion optimization and drive up lots more conversion?
Yes, I am.
There you go! I’m trying to do Snapchat. It’s an interesting experience because people interact with you one-on-one, and it’s been interesting. It’s actually khalidhajsaleh. Maybe we can put it in the show notes.
Yeah, we will.
All right, well, thank you, Khalid! This is fantastic! You are just a fount of knowledge and wisdom about conversion and turning lookers into buyers so, thank you for sharing all that. Listeners, be sure to check out the show notes on www.marketingspeak.com. There will be a checklist there, which you can download. It’s a nice PDF of actions to take from this episode. Thank you for listening. This is Stephan Spencer, signing off. Catch you on the next episode!
- Invesp Marketing
- Conversion Optimization: The Art and Science of Converting Prospects to Customers
- Google Analytics
- Visual Website Optimizer
- Don’t Make Me Think by Steve Krug
- Khalid on Twitter
- Khalid on Snapchat
Your Checklist of Actions to Take
☑ To find out if there are issues with your website, go through and mimic the exact steps a visitor would
take in the purchasing process. Would you purchase from your website again?
☑ Page load time and speed can impact conversion. Reduce your page load time to around three to five
seconds to see anywhere between an 8-15% increase in conversions.
☑ While there can be many issues on a website that need to be addressed, functionality is at the top of the list. If a landing page, a product page, or a dropdown isn’t working, fix it immediately.
☑ For A/B testing, start by looking at the number of conversions that you have. For each 100 conversions, you can add another variation to test, and see what works the best.
☑ Really think through your optimization strategy before you begin to implement it. Ask yourself, what is the big picture? What would you like the end result to be?
☑ When you overemphasize that you have a certain characteristic, such as being trustworthy, you can
receive a negative reaction. Don’t overdo it.
☑ There’s a methodology to how to fix website-centric factors and errors. First, focus on trust, then address FUD’s. Lastly, you want to move onto incentives and engagement-or the buying cycle.
☑ Add credibility to your website. An easy way to build and address trust is to add social icons and links,
and also to show how many people placed and order for an item.
☑ From a conversion optimization perspective, you should have a completely separate mobile experience. Direct mobile users to a “mobile.yourwebsite” version, and you will have more control over your user experience.
☑ Be sure to touch base with clients via email at least once per month. An email channel will convert better than a paid or organic channel, because they have given you their information already, and you are bringing them back.
About Khalid Saleh
Khalid Saleh is the CEO and cofounder of Invesp, a leading provider of conversion optimization software and services. He is the co-author of O’Reilly’s “Conversion Optimization: The Art and Science of Converting Visitors into Customers In 2016, he is gearing up to launch his technology startup Figpii, a one-stop platform for everything related to conversion rate optimization. A recognized expert of marketing strategy, he has presented at marketing conferences across the globe. Khalid has been a frequent guest in key media .outlets, including CNN, BBC, SKY, France 24, MSNBC, New York Times, National Public Radio, and more.