About Cara Harshman
The former Content Marketing Manager and Blog Editor at Optimizely, Cara Harshman is a preeminent voice in A/B testing and website personalization. She wrote the book A/B Testing on behalf of Optimizely’s co-founders and has spoken around the world. She dislikes writing autobiographical paragraphs but loves telling stories to live and digital audiences.
In her Learn Inbound talk, Cara will explain how website personalization is a marketing strategy that allows you to create unique, valuable experiences that can explode your sales and grow your business. Personalizing your marketing may be a daunting idea right now, but after Cara breaks it down, you’ll realize why embracing it early will be transformative, highly lucrative, addicting, and not creepy. If you want to be on the cutting edge of digital marketing, you need to learn how to personalize your website for each user.
- The what and why of website personalization
- Case studies highlighting how top brands are using personalization today
- Common pitfalls to avoid and actionable tips to start delivering personalized website experiences
Hello, everyone. I am so thrilled to be in Dublin, Ireland with you. I'm coming from San Francisco, a long way, but I'm making the most of my time in Dublin and Ireland and Scotland too. So I actually, specifically, chose that song, "Staying Alive" by the Bee Gees, if you guys don't know it, because I wanted to share a story that is only really applicable to a captive audience of inbound marketers that will nerd out about personalization and A/B testing and SEO and things. So it's a short story that I'll just start with. So I used to work at Optimizely, and I was there for four years. And in that time, I kinda came to be known as the rah-rah person in the company. Like I knew people, I could get things done. They went to me if they needed like something ridiculous to happen.
So when we were launching Personalization, the sales leadership team came to me and they were like, "Cara, we need to get people excited about Personalization, this new product we're launching. Can you do something to get people excited?" And I was like, "Sure. I can do something. Let me think about that." So I went back to my little marketing den and put my thinking cap on and came up with a whole litany of parody songs in the theme of personalization. And I performed these songs live in front of the company to get people excited like in between the weird, or not the weird but the like less than exciting talks from the sales team on how to sell this thing. So one of the songs I came up with was, "Ah, ah, ah, ah, personalize, personalize. Ah, ah, ah, ah, personalize..." and anyway, you get it. So I did a bunch of that. And I hope you'll ask me later after the talks about my full rendition of "No Scrubs" by TLC that I gave the breakdown of why Optimizely is better than VWO, if you know about that competitor. Okay, anyway.
But the story that I came here to talk about today starts in 1998 with a man named Jeff Bezos. At the time, he was about four years into building his small book retailer that came to known as amazon.com. And back then, he was starting to do a lot of interviews. And he did this one with the "Washington Post" newspaper in '98 where he said this very momentous statement, "If we have 4.5 million customers, we shouldn't have one store. We should have 4.5 million stores." So Jeff, back in 1998, was kinda setting the stage for this phenomenon of internet marketing and way that we do business today known as personalization. Now, personalization, I shorten to P13N because there's 13 letters in between P and N. So typing out personalization is really long. Anyway, so saying personalization, showing unique content to a user in real time based on something you know about them, has now become pretty ubiquitous, especially within the B2C community.
A consumer website today wouldn't even think about launching a marketing campaign that was one size fits all. It would be completely kind of like ludicrous to them. This is a snapshot of my Amazon home page, totally personalized, knowing that I'm looking for Chaco sandals and a revolutionary cooking device called the Instant Pot. Does anyone know about this? Woo. Yeah, someone's cooking rice with an Instant Pot out there. Okay. So everything about this consumer brand is very personalized to me, but the B2B world, probably a lot of you are B2B marketers out there, we're kinda lagging behind. But recently, all this huge wave of new technology, new data is available. It's a bonafide explosion of software and availability of data that lets us do personalization in real time.
So this is just a glimpse of the different providers we have that allow us to do personalization, serve unique content to people in real time based on something that we know about them. So for data providers, these third parties that you can integrate with your own website, we have people like Clearbit, Demandbase, KickFire, DiscoverOrg. And then there's even personalization engines, companies that are just built to deliver personalized experiences, such as Optimizely, Cupid, Get Smart Content, Google Optimize. And then there's even strategy partners, these agencies out there that are specialized for bringing you personalized content, helping you create, helping you deliver these experiences. RedEye and conversion.com are from Europe and the rest, Funnel In and Sierra Metrics are two that I highly recommend from the U.S.
So personalization is so big that Forrester has been doing some surveys about it. And last year, this survey from 2016 showed that 68% of technologists are prioritizing personalization today. And what are they prioritizing? They're looking at content on the website, promotion offers, product recommendations and reminders, all sorts of things that we do as marketers, as technologists, as UX designers. So there's still, though, this kind of shroud of fog around exactly what personalization is, how to do it in a way that's not creepy or taking advantage of the data that you have about your visitors. A lot of people don't want Big Brother to be watching, which is very understandable. We don't want you to be creeped out when you're surfing the internet.
So I want to share some stories, some building blocks, if you will, about how to do personalization in a delightful way that's meant to convert your users, give a better experience to people, and just have a lot of fun as a marketer because it's a pretty big playground when you unleash all this data you have, especially behavioral data, about your visitors. Okay. So this starts with two big blocks. Kinda think about it, it's like the foundations of personalization. First is who are you personalizing for? Which audiences are you building for? And number two is what are you showing them? What is the experience that you're delivering that's personalized to your visitors? So it starts with this audience. I define an audience as a group of visitors with a specific need that deserves unique experience, because not all types of visitors have a specific need or deserve something unique.
You might have lots of different visitors, and you probably do. You have many audiences already. If you have people like a power user or an enterprise customer or an advocate, those are all user personas already that are kind developed that you can now take and turn into audiences for personalization. Okay. So breaking down this big block into a few different smaller blocks. So the way that we're slicing this audience starts with behavior. What is your visitor doing? That's the first way to divide it. Number two is demographic. What is innate about this visitor? What is unchanging? Who is this person? And the third one is context. What's unique about the visitor when they're on your website? How did they get there? What brought them there? What referral link, etc.?
So the three building blocks are context, what is unique about the visitor right now, behavior, what is the visitor doing, and three, the demographic, who is the visitor? These are the ways that you take the huge 100% of your visitors, the huge pie, and slice it into smaller audiences that will then personalize the experiences. So this is a lot of information to take in. So let me break it down with a few specific ideas. Okay. So behavior, here's some real examples of audiences that I've seen companies create as audiences. So behavior, people who have downloaded specific content from you. They could be in one audience. People who have viewed a specific page. People who have abandoned a cart. People who have read a specific category of content. Those are all examples of behavioral audiences. Then it gets really complex. I couldn't go too complex here, but I'll talk about that later.
Then context, referral source, device, browser. Are they on an Android versus an iOS? Are they from a specific marketing campaign with a UTM source? Are they a new versus returning visitor? These are all different contexts of your visitor that you can then tailor the content for. And then last is demographic. And this is where those third-party providers really come into play is what industry are they in? Are they retail, financial, healthcare? All very different needs. Gender and company size too. So to help illustrate more of these, I'm gonna give an example of a real company from each of these different categories. The first one we'll talk about is a contextual case study from a business called Secret Escapes. Secret Escapes is a flash sale luxury site. You can book vacations, spa treatments with them. I don't really use it, but they have a lot of customers. They're doing really well.
So they had this dilemma that a lot of us have. They have great ads on Google. They're doing their SEM well, but people are clicking on them but not converting on the landing page. So they had a hypothesis to try a more personalized landing page based on the search term that the visitor searched for. This is called symmetric messaging. It's kinda standard now. I'm starting out with a vanilla, not vanilla, 101 flavor...vanilla is still a good flavor...with a 101 example. So in the generic version, they showed this default experience to everyone. No matter what you searched for, like Ireland vacation or spa, like, mask treatment, they would show this landing page, "Join now for free. Save up to 70% on luxury hotels and holidays." But then they decided to personalize it and dynamically insert the key word that someone searched for on the landing page.
So when someone searched for "spa breaks," this is new landing page. The image is personalized based on a spa to give you the feeling like you're there and they put "Spa Breaks" into the headline. So they made this an A/B test, the personalized versus the generic. And guess which one won? 'Meh," that's what I heard. Cool. Okay. The personalized experience won. Thirty-two percent more people in this segment clicked on the more personalized landing page with the key word that they were looking for. Not really surprising. If I'm looking for something, I want the search result to reflect what I've searched for. Okay. You get that, pretty straightforward.
Next is an example from a company called Citrix. Citrix is like the epitome of a B2B company. They provide networking solutions for big companies that work remotely. They used to make GoToWebinar, but then they've sold that...fun facts, I didn't know if anyone knew that...to LogMeIn in 2016. A cool fact. So Citrix, like any other B2B company, has a ton of audiences. They have people in the retail business, in the healthcare, in the financial services, in the education. So one message just doesn't fit for all of these different audiences. One e-book won't really serve all of them. And as marketers, I'm a content marketer, I'm killing myself to create all these e-books and all these different messages for all these audiences. So personalization is giving me an opportunity to serve that content to the people who really need it.
So that's exactly what they did. They came up with some ideas of how to personalize their home page. So this was the default experience that everyone got, "Citrix partners with Microsoft to deliver the next generation of Azure remote app." Great, but what if I'm from the educational industry? What do I want to see? I don't really know what that means. I don't really know if I care about that. Okay. So for the personalization side, they used a third-party data provider. In this case, the used Demandbase to look up the IP address of the visitor and matched that with the industry. So they could see first-time visitor, never been here before. I can tell you're from the education world. So I'm gonna show you this nice photo of a teacher in a classroom and a headline that's targeted for teachers.
They did this the same way for the SaaS or IT, secure your data, drive innovation, and they did this for healthcare as well, enable healthcare mobility. I could only fit the top part on the screen, but they changed content below, the white papers they were showing, the CTAs and everything was very personalized. At first, they tried this with just a test group of three audiences before rolling it out to all their verticals. All right. So what they did is they tried this one versus these three, many more. And again, the personalized experience won. So a little bit more behind the hood here, or under the hood here, they were talking to anonymous visitors.
They'd never seen these visitors before. They were using Demandbase to look up the IP address and match that to the industry. And they are using Optimizley in this case to deliver the personalized image and the personalized headline to the visitor. Okay, but this experience didn't only win by a little. It did really well. It increased banner clicks by 30%. It decreased the bounce rate on the home page by 7%, 10% increase in page use per visitor, which is pretty big, people were more engaged and going deeper in the site because of the home page, and they saw an increase in 4% average session duration. So overall, this just worked really well. Great example of B2B. I love this example.
Okay. The last one, which is my favorite because it's complex, but really insightful, comes from Brooks Running, an e-commerce company that makes running gear. So they were trying, in this case, to decrease returns because returns are the bane of our existence as consumers and as retailers. I don't like returning stuff. They don't like waiting for me to return, and all that. So what they did was they looked into behavior of people who return stuff to see if they could find any data correlation to see, okay, what are the indicators of people who are going to return an item and how can we prevent this from happening? So they ran some data science things. They did some analytics. They looked at their return data and saw that people who had a high order value were very likely to return.
Now, high order value's supposed to be good. Like you're buying a lot of stuff. That's great. You're making money, but in this case, people who had a high order value often were buying the same exact shoe, just in different sizes. So they were unsure of what size shoe they wanted. They were like, "Am I a 7 or am I a 7.5?" So they ran a very simple, like a stupid simple, experience to personalize to the segment of visitors from the likely to return cohort and just said, "Not sure which size to get? Our expert customer service can help you figure out the one for you so you don't have to return the other." And IU had made it look better than it was. Like the pop up was just so boring and plain, but this is all it took to improve the experience of the entire...decrease returns, basically. Yeah.
So let's see what happens. They pitted this against the generic, just like nothing generic. And the return, the personalized one, won as well. They decreased returns for this audience segment by 80%. So that's pretty massive. And not only that, they also increased conversion rate by 2% because people were more confident that their shoes were gonna fit, and they were gonna add some socks in and add, you know, more other things. And also, their customer happiness scores went up. So all because they just looked into the data to find out who was likely to return and to give those people a very specific experience. They did really well.
So these are the three blocks of building your audiences. That's what we're talking about so far is the context. What is unique about this visitor right now? What is the visitor doing, their behavior? And who is the visitor? Okay. So that's the who. Now we're gonna talk about the what. What are we showing them? The what is the experience. It's kinda the full message you want to send to the audience. And it, too, has three little blocks inside. One, the location. Where is the experience on your site? Three, the content. What is the actual message that you're going to share? And four, or three, the measurement, like how are you measuring to make sure that this is actually performing better than the generic version? Because personalization is a hypothesis, it's not a guarantee that it's always gonna work better. That's really important to remember. Okay.
This is what I just said. Where on the page is the experience? What is the message? And how is it performing? Okay. So for this one, I'm gonna share an experience case study of a full website redesign that they did at Optimizely. I was still here when we did this, but they've done a lot more since then. So at this stage in Optimizely's history, we were trying to move from a quantity-based lead generation machine, like just tons of leads, just get them in the door and then convert them, to a very quality-focused lead generation machine. So in other words, we were kind moving towards an account-based marketing, like very targeted account creation and account sales working. And we weren't just going after everyone. We weren't fishing with nets, so to speak. We were fishing with spears for those very targeted companies. Okay.
So our home page went from...this is high level. Our home page went from this, just all we wanted was your email, that's all we really cared about, test, personalize, and optimize, to this for Sony. Only visitors from Sony, to this for Target, to this for people from the "New York Times," to this for people coming from Microsoft, to this home page for people from the retail industry. If we didn't know what company they were with but we knew they were retail, they would see this one, "Let's optimize digital experience for your shoppers," with a picture of retail, to this for people from the travel industry, a nice wingtip, and this in the afternoon. If we knew nothing about your company type, your industry, but we knew what time zone you were in and we knew it was afternoon. And then this one, my personal favorite, for people surfing the web between midnight and 4 a.m. We're like, "Why are you here? Why are you checking out software in the middle of the night? It's weird."
So I'm gonna break down kinda the full thing of how we were doing this from the location, the content, and the measurement perspective. Okay. So the first one is the location because if you remember our first site, we just had a headline and a button. There wasn't really much there to work with in terms of personalization. So the first thing I'd say about location is you need to create space to do personalization. Oftentimes, this means redesigning a website, adding opportunities for CTAs, for content call-outs, for images or for more headlines. So this picture right here is an actual mock-up of the site that our designer, John, drew when we were thinking about how to redesign the site.
So the next step after you have made space is to just kinda lay out your site like, "Let's print it out and just paste it up there and then circle..." This is so low fidelity, but it really works, all the places that you could personalize like, "Okay. We could easily change that image. We could change that headline. All those logos, we could change, and content and everything else. You could just say, "What is personalizable in general?" You could say it all. Okay. Then the next thing you do is you think about designing the experience. What is the specific message you're gonna send, and where in the journey are you gonna say it? So this one for us, we were really trying to hit all the different audience data types. We wanted to get behavioral data in there, demographic data, and contextual data. So I'll take you through a few examples.
So here, the hero image and headline change is based on demographic data. This is what we know about the visitor. We're not inferring this based on behavior. It's like Demandbase sending us information that says, "This visitor is from Walt Disney IP address, show them that." And then with Salesforce, we got pretty like geeky here. We personalized this section to the actual account manager for that account. So like they were a customer at the time, or they still are a customer, so we said, "If you are curious about something, schedule a meeting with Joseph." Pretty cool. Okay. Then for Sony, this is what we changed, the hero image and the headline again. Same thing for Target. And then when visitors are unidentifiable, we don't know what their industry is or who they are, we just go with time of day.
And then when these modules are also changing, so we made sure we added enough space, like I said, for personalization. So the one on the left changes for behavior, the one in the middle changes based on their website, and then the one on the right changes based on who their account manager is. So the thing that I will say...Oh, they have the [inaudible 00:23:30] there. Cool. Okay. And then this section changes based on demographics. So this is if you have customer case studies, this is one I love, showing all your different customers in the different industries that you work with. So if someone is coming from a media site, they'll see logos of the media companies that Optimizely works with. Because they don't really care what e-commerce sites you work with, they want to know who of my peers trusts you.
And then the last section on the site, on the home page, was the call-out for contents and promotions. So if we knew that someone was 10 miles away from an event we were hosting, we would share like, "Come to our event in Los Angeles. You'll love it," or if we're having a webinar or anything that's related to something that we know about them, this is the ample spot for the content marketer to take advantage of. So one thing that's to note with personalization is that it creates a content need. If you are designing seven experiences for seven different audiences, you're gonna need content for each of those. So this is often a good way to judge, okay, does this audience deserve a unique experience? Do we even have content to support it? If you do, then great. Then that's an automatic slam-dunk for a personalized experience, but if you don't, then maybe it's not worth it, especially at the beginning when you're just getting started. I would really recommend starting with just a few audiences and not going too overboard with a lot. It's gonna confuse you and it's going to be a content overload.
Okay. So that last section of the what is measuring the experience. So for us, this is not just a, "Fun, let's try our new product and launch it." We really wanted to generate new engagement, engage the right kind of people, and we wanted to measure this space on two levers. We wanted to see qualitatively, what are people saying? Is it tanking? Do people hate the look? And we also wanted to be data driven, like we are, and measure the actual conversion rates. So the first one we saw was qualitative. We monitored the tweets, the Twittersphere, of all this. And we saw people are actually tweeting about our home page. That's pretty cool because no one really tweets about your home page. So when that's happening, that's awesome. And people were sending emails, like their account managers to their account reps saying, "Oh my gosh, look, your face is on Optimizely's website. That's so cool."
So okay, that was awesome. That was fun and like, "Yay," but next we had to look at the hard data. So we pitted this old version of the site against the new personalized version. And the personalized version won, not by much, but we just wanted to see that it was not tanking all of our results. If it was upticking in any way, we were happy with that. Okay. And it's also was knowing that this personalized site did not automatically just like make all of our metrics soar. It took a while to get there. We ran it for a long time, but we did see that 1.5 increase in engagement. People were just clicking on the site more, which is great, 113% of increased views to the solutions page, which was like our products page. That's excellent news. And then 117% increase in "starts the account create" process. And this is not "creates an account," this is clicks on the button to start an account. So this told us a few things.
One, we are finding a new local maximum on our way to finding that global maximum of top site performance, but also that we have some more work to do, that people aren't actually creating new accounts, they're just starting it. This is actually really handy because it says, "Now you have to go optimize that account create form because it really sucks and people aren't actually finishing it." Yeah. So that's what we found out. And since then, Optimizely has been doing a lot of really interesting personalization, like on their pricing page, they're changing the page, how it looks and how the solutions are ordered based on what you've looked at. They're doing blog recommendation based on what you've read, like recommended for you, that kind of thing.
So okay. Here's the building blocks to design your experience. Make sure you're looking at the location, like where on the page is the experience? What is the content and the message that you're showing, and how is it performing? Now I want to leave you with a few tips to just get this shit done because it's a big investment, and proving ROI from it takes some time, but it's worth it. So if you do a little bit just to start it today, then I think you will see that it really pays off. Okay.
The first one is don't create too many audiences at first. You're gonna run out of content and into confusion. So if you want to show results quickly, just pick like new versus returning visitor or customer versus not customer, logged in versus not logged in, or enterprise versus SMB. There's probably lots of ways you could just slice your visitors in half and create different experiences just for them today. And then treat personalized experiences as hypotheses. They should be carefully crafted and measured. Again, we measured every single thing that we've done. There's a whole back group. So 5% of traffic is always seeing the generic version. They're never seeing the personalized. So you can make sure that you know, you can prove like, "Okay. This is actually working." This is actually increasing whatever metric you are trying to move.
And then lastly, do personalization because it's valuable, not just because you can. We are very privileged to have access to so much data that we do about the way people interact with our sites and who they are. So we need to be responsible and use that for good and not just to sell someone something that they don't want. So use it for good. Don't be creepy. And use these building blocks, the who, the what, the context, the demographic, the location, behavior, measuring in content to create experiences that dazzle and convert because there's a lot we can do. We just have to put our heads together to do it.
So thank you so much. I wrote this guide to B2B website personalization that's hosted on Clearbit's site. It's really informative, really in-depth, like 6,000 words. So if you want to learn more about this, I highly recommend you go there. And if you want, you could download this deck on my website. Thanks so much, everybody. See you later.