Episode 1: How testing can improve usability and increase ad income

Listen to Episode 1.

Transcript:

John: So welcome folks. We are here for Ezoic’s first podcast and we have Jacqui McGuire here to ask questions on behalf of the publishers. These are the most frequent questions that we get asked and for those people who don’t like to do lots of research and read reams and reams of pages on the website, this is for you. So Jacqui, hit me!

 

Jacqui: Okay John. Just briefly to start with, what is Ezoic and how does it work?

 

John: Okay, so Ezoic is a tech platform. How it works is it takes the content of any site that’s been written – pretty much in any language – and reconstitutes it into layout experiments. Now that sounds quite difficult to understand, but the whole idea behind it is: it’s the same content in a new layout. And then the new layout means different navigation features, and it also means different ad positions, and the result of testing different layouts means more money and better usability. Are you with me?

 

Jacqui: Well how does that affect my existing set up? Do I need to be worried about anything?

 

John: Okay, so if you’re a site owner, obviously everybody is concerned. They’ve quite often run sites for many years, sometimes 15 years. So no, you don’t need to worry about it. It really does work with any site and the reason for that is because there are two ways you can integrate. One is using a javascript tag. We don’t recommend that one for everyone and the reason for that is it’s slightly slower – it means that the site loads a bit slower. Name server integration, which is when Ezoic sits between existing hosts and the user and what it does is it acts like a proxy, and we can do all sorts of great stuff like code compression, lazy loading on mobile. We use Amazon web services for loading the site fast around the world. We’ve got four data centers. We use a CDN called Cloudfront. So there are lots of layers of benefits of using the name server integration. And what the system does is it pulls in the existing DNS, which is all the settings that control things like MX records, email and so on. Pulls those in – you’re always in control. The good thing about name servers is that you can turn it on and off using our system at any time without changing your code.

 

Jacqui: Right. Okay.

 

John: Did that go over your head?

 

Jacqui: I’m still with you so far. That’s a good sign.

 

John: Alright.

 

Jacqui: I guess lots of us might be worried about: there’s got to be a catch somewhere. How do you guys make money from this?

 

John: How do we make money? Well, I suppose what I should do, I’ll give you the quick answer. The quick answer is after 30 days of a free trial, we put an extra ad on the page and because we’re partners with Google, we are certified Google AdSense partners, and we also have a partnership with them for their ad exchange product. That means that if certain pages are big enough we can put five ads on a page instead of three and we make money by putting an extra ad on that we keep the income from. So effectively we are diluting the income of the owner, but because after dilution everybody seems to be making about double what they were using the system versus just plain AdSense, most people look at it like it’s paying for itself and it’s free. The long answer to that is a really long answer, but it’s really about where Ezoic comes from. The idea behind Ezoic was to improve websites that we already owned. Dwayne LeFleur founded the business. I came in quite a bit later, but we both come from the position of owning sites that we were monetizing. So we built this system for ourselves, which is how do we improve income and how do we protect usability – because anybody can spam out a site and put loads of ads in it – but how do you do it scientifically so that you can get the bounce rate going down, which is when somebody comes and looks at one page and leaves straight away. How do you get the time on site going up and get the page views per visitor going up? If that user experience is protected and you’re putting the ads a little more front and center, effectively you’re getting more money from advertising because it’s being seen and people are leaving your site through an ad rather than just hitting the back button. Of course Google – the Google algorithm – no one really knows how it affects you and how you get up in rankings. However, overall if you’re improving usability and user experience metrics and you’ve got great content, you’ve got great rankings and we never set out our stool to be an SEO business or whatever. That’s one of the byproducts of layout testing is that you improve user experience every time, particularly on mobile. That means more money, more money because you’re getting better rankings and better rankings means you’re getting more visitors.

 

Jacqui: You mentioned doubling revenue. I’ve read a few things about that. How realistic is that and how quickly can we expect to see it?

 

John: So it’s not instant and it’s like with all testing. I was talking to a site owner the other day and he was saying running a website is testing because you’re always trying new things. An average size site, I’m talking about 200,000 visits per month, takes six to eight weeks and the reason for that is we’re at the mercy of the mathematics behind everything. You need to be statistically confident that the results you’re getting from particular layouts are not just a bunch of outlying kinds of users. Let’s say someone spent 20 minutes on your site. You don’t want that to be affecting the overall dependability of the results. You need to show a certain number of views per layout for the results to be good and you can say “we can rely on that going forward”. The smaller sites do take longer because of that, and larger sites in this sort of several million to tens of millions per month you can see results in a couple of weeks. I mean, they can pile through tests and we’ve seen it sometimes when they’ve gone through 40 different desktop layouts, for example, in a few days because it gets them to a statistically relevant point and then it says: “okay that’s not beating our current winner. Get rid of that.” Overall, what the system is, it is quite counter intuitive. Most people are used to a “look” of a website being the website, but you’ve got to look at a site like Google looks at a website, which is not design, it’s usage metrics, it’s page yield.

 

Jacqui: That can be tricky for most of us to get our heads around, especially when you mentioned earlier about many publishers have spent years building their site and tweaking it to make it look exactly how they want it to look.

 

John: Yes and that is a very, very good point because most people spend a lot of their time trying to make the site look better and trying to improve the usability, but they are constrained by their own preference. Now that may sound a bit weird as well, but for some sites and all sites strangely enough do optimize differently according to the user behavior because how Ezoic works is it takes these behaviors and the results of the user data is influencing what gets tested. So one publisher might be really, really accomplished in their chosen field of interest and that means they are experts in writing what they’re writing about, Medieval History let’s say, but they might not know about how to make the site work really well on Opera Mini Browser or Amazon Silk, or have it work on the new iPhone. They might not know about how to put the ads in in the right position so you’re not upsetting user experience, but also where to take advantage of putting ads. And we’re not saying we have the answers, because it’s not us. It’s a system we’re testing and it’s actually the users that have the answers because they vote with their mice, if you like.

 

Jacqui: How much control would a publisher have if there was say, some ads they didn’t want to display, or certain layouts that they didn’t want to use?

 

John: Like I said, we built the system for ourselves. We’ve given all the control we can think of to the webmaster / the website owner, so you can block ads in the same way that you have them blocked at the moment in things like AdSense. You can remove specific ad locations if you have a personal preference and you don’t want the system to work out the positions for you. You can only test mobile, only test tablet, only test desktop. So you can turn the whole system off at any time and of course you can control the percentage of your old site traffic that is being monetized by your old methodology and your old layout and the proportion you want to divert for testing.

 

Jacqui: So I guess if I were a publisher then it would give me A) a bit of control, but also an opportunity to have some fun with it and see what was working and what wasn’t.

 

John: You can. Everybody has to be honest about what they want to get out of testing. They all say you need to come up with a hypothesis first. You say “my goal is to…” and if it’s user experience metrics, you tell the system to go get it, and that’s what it will test for and it will weight the results of user experience testing above revenue. And if you want revenue more than you want user experience, let’s say you’re prepared to see the bounce rate go up by a couple percent, you can say “okay let’s weight revenue instead of user experience”. But I would say the vast majority of our publishers go for a balance of the two because, I think by now, everybody knows search engines do reward sites that improve user experience.

 

Jacqui: With all of this testing and all of the different layouts and ad placements, what impact does that have on traffic and speed?

 

John: Okay so as I said before, we use lots of different methods to load this site fast wherever the user is in the world. That’s number one. User experience, particular on mobile, it is very important to load fast and that means having lazy loading. Lazy loading sounds like it’s going to be so slow, but actually you’re loading content first and you’re lazy loading other things – ads as well – later so the users are getting what they came to the site for first, so that they’re not going to bounce back and go “this is taking too long”. We’re always focused on user experience anyway, even if the system is trying to get income yield and the other thing was, I’m sorry. What did you ask?

 

Jacqui: Speed and traffic, number of visitors.

 

John: Okay and how it might affect it. When you’re getting most of your traffic from Google – let’s be honest, Google knows pretty much everything from the data that they get from Google.com, Google.uk, wherever you are using Google – they’ll know a search return and whatever you’re searching for, like I said Medieval history. You go to a site that’s ranking for that keyword and if you land and you don’t like what you see, you’re going to bounce back. Google sees you again. They say “oh you didn’t spend very long on that site. You bounced out”. Now high bounce rate isn’t necessarily bad because – particularly for short bits of information – you just want to get information and go. For example, you capture more ads in the high bounce rate site. That’s what we’ve seen by the testing that’s been done by the system. Overall if you’re able to improve your bounce rate even by a few percent, you can start to win more traffic from the competitors for those key words. So overall we see sites grow and grow. Some of them who haven’t had mobile sites, who have had a slow site around the world because as you know their servers are in Germany and most of their users are in the US, there’s some latency problems that they didn’t even know about because most website owners will look at their own site and go “loads fast for me”. That’s always the problem. This is the – I said counterintuitive before – but the thing that is weird is you’re realizing that the content and the source that you created isn’t the same for everybody. Everybody’s seeing different layouts in an experiment. They’re not seeing a different one every time they come – they’re getting cookied and it’s all being done in a way that improves user experience. We don’t want to confuse the users, but we do want to get the data about what they like and what they don’t like. It does require a mindset change, which is to say it’s not what I might like, it’s not what you might like.; it’s actually what the majority of my users like. That is the most important thing. That’s what testing achieves.

 

Jacqui: How does the publisher get started with it?

 

John: It’s quite easy. Like I said, we try to make it easy. You get started by just signing up on the site. Now it sounds very sales-y, but most people are preparing themselves, steeling themselves for some enormous technical hurdle. “Yeah, this sounds really great and I want to use it, but it’s going to take six months integration”. No it doesn’t. Its a few days. You sign up, you create the account and that pulls in most of the data automatically. The only thing the publisher has to do is change their name servers, which allows us to show the experiments. Then after a few days of manual checks, most sites go straight through, some of them can be ready within a few hours, but we have to wait for the DNS, the name server to propagate around the internet. It’s really no work to integrate because the system takes everything that’s already there and just creates new layouts. That’s it. It sounds easy, right?

 

Jacqui: Simple.

 

John: If you were going to try to do this yourself, you would have to construct every AB test. Is the ad better on the left? Is the ad better on the right? What about the menu? Should the menu be on the left or should the menu be on the top? What about on mobile? What about double drop down mobile? What about sidebar in mobile? What about on tablet? There’s so many permutations of content and ads, usability concerns. You’ve got multiple angles to come at it from, and automated testing, which is basically what we do, is really complicated to do if you’re going to try to do it manually, but that’s why we’ve got the automated system.

 

Jacqui: So that is Ezoic?

 

John: That is Ezoic. Is that the really long answer to the short question?

 

Jacqui: I asked for a few words about Ezoic.

 

John: Oh. I do talk a lot. I know that. Okay, c’mon and hit me.

 

Jacqui: You’ve answered all the questions.

 

John: Okay, alright.

 

Jacqui: Is there anything I haven’t asked that you think is really important and publishers need to know?

 

John: I would say overall patience. The people who do best from the system are, just somebody off the top of my head, I was just looking at the results, they signed up and it’s a pretty big site. They were getting about 60,000 visits a day. I looked at their stats and they’re up to 100,000 a day now so day by day, week by week with this sort of undulations of the weekly cycle that traffic is improving terrifically. They used to make $170 a day from the site. They owned it for years and years and years. Now over $700 a day and within the first week, it was ridiculous. I’m not sure about this and I would say to everybody who tries testing whether you’re doing it yourself or using an automated service like ours is give it time, go with the numbers. If the user experience is going up – so the bounce rate is going down, the time on site is improving and page views per visitor are going up – if the user experience is going up, the users are liking it. Now you might be very nervous about how all of this changed. It feels like a lot to take in, but stick with it and you’ll get the results. I’d say patience is the biggest factor in success when people are looking at this kind of thing.

 

Jacqui: So the example you’ve just given – what period of time did it take to get to that point?

 

John: It was pretty quick actually because it was a big site, and when you’re at sort of two million visitors a month, which that site is, plus it’s growing, that was three weeks. So if you’ve got a more modest site, you need to give it four to six weeks. Now that might sound like a big deal to wait that long, but if you think about the way you’ve been working on the site in the past, you’ve been doing testing, you just haven’t done it scientifically until now. It’s just accelerating what you already do when you run your site.

 

Jacqui: So is the key to be patient and dispassionate?

 

John: Brave. Ballsy. I think you’ve got to give it a really good go if you want to get the results because you’ve been getting the results you’ve been getting because of what you’ve been doing. Now try something new, like this. It does take a lot of getting your head around, but that’s why we’re here. That’s why we’re recording this thing. Anybody who is listening to this, thank you for listening. We would like to get more questions from you and particularly really difficult ones because we love answering those. Send them over to [email protected] and we look forward to hearing from you. Thanks very much. Thanks Jacqui.

 

Jacqui: Nice job.