Mark Jennings (Lonely Planet)
31 Jan 2013
Not only is Lonely Planet an excellent online resource for travellers, they also take performance very, very seriously. In this week's podcast, Joshua talks with Mark Jennings, Technical Operations Manager at Lonely Planet Online, about how to sell performance within your own company, Lonely Planet's phenomenal growth in mobile traffic, and the secret of not just making your pages fast, but keeping them fast.
Joshua Bixby: Hi again. It’s me Joshua Bixby, President of Strangeloop and blogger at WebPerformanceToday.com. Thank you for joining me for another installment of the webperformancetoday.com podcast series. This one I’m excited about. I’ve been holding onto it. It’s my last interview from Velocity Europe with a gentleman that I really look up to – Mark Jennings who is the technical operations manager at Lonely Planet online. For those who are not familiar with lonely planet, you have probably not traveled to the far ends of the earth. As a traveler myself, I have relied on those books to get me across a number of very dodgy areas in West Africa among others. I was actually chastising Mark for the fact that Lonely Planet said there was a paved road across southern Niger that did not exist when I was driving about 15 years ago so they are a great resource for travel information and you know they also take performance really seriously which as you know is something I like. Mark and I talked about sell performance within a company, something Mark has done and the team have done exceptionally well, about mobile traffic and about the secret of not just making your pages fast but keeping them fast which I think is as important a topic. Enjoy.
Joshua Bixby: Tell us a bit about what you do at Lonely Planet.
Mark Jennings: So my job title is technical operations manager which basically means I work with a really talented development and operations team and right now our focus is on a massive transformation of our site. We’re changing technologies, we’re changing infrastructure. We are ultimately ripping apart our site and building it from scratch which is very challenging but also really exciting.
Joshua Bixby: And what are the business drivers for that? Why did the business buy I mean that sounds like a techie’s dream and a business person’s nightmare so help the people listening understand when they want to live the same dream you’re living. What is the logic?
Mark Jennings: I mean that’s a really I mean I’ve got the answer in my mind and the basic answer is that the previous business dream wasn’t successful
Joshua Bixby: Okay
Mark Jennings: So we’ve almost paid the piper and we’ve got another chance.
Joshua Bixby: Gotcha. And so a sense of renewal. There is a renewal phase.
Mark Jennings: Absolutely
Joshua Bixby: And with that comes a bit of latitude around technical thinking.
Mark Jennings: Yeah I mean actually it’s a massive shift from a previous we approached things previously to where we are now and a big change in that is the sort of the technology and really try to put technology at the absolute core of what we’re doing. A lot of what we’ve learned comes from a lot of people that we work with reading the link startup
Joshua Bixby: Yep
Mark Jennings: Focusing on that MVP, focusing on no waste, focusing on doing everything as lean as we can and that’s been a big driver so a huge part of that for us is metrics. You know if anyone’s seen our presentation at Velocity, Dave Nolan talked about this idea of facts vs. hippos. Hippos being the highest paid person’s opinion and we’ve been there in the past and we don’t want to be there again. We try to deal in facts in metrics in real data as often as we can and the only way we can do that is really putting technology at the heart of what we’re doing.
Joshua Bixby: So you have a dream job basically. Tell me something about your job that will dissuade people from thinking you have the greatest job in the world.
Mark Jennings: Well if you want to come work for us, then I have the greatest job in the world. The biggest challenge for us right now is delivery schedules
Joshua Bixby: Of course
Mark Jennings: And the reason we have this leeway is because we have a really demanding delivery schedule and we’ve sort of we’ve convinced people that the only way that we’ll give them the product in the time they want is to give us a long leash on the technology that we use.
Joshua Bixby: Yeah
Mark Jennings: I say convince people
Joshua Bixby: Sure
Mark Jennings: But I truly believe that’s the case
Joshua Bixby: For those who didn’t attend the presentation you did today; give me the two slides that matter.
Mark Jennings: Yeah, Ok.
Joshua Bixby: Because the rest is fluff, we all try to fill it in with fluff. You have to fill up 40 minutes like I did.
Mark Jennings: 40 minutes is a lot of fluff.
Joshua Bixby: Give me the two money slides.
Mark Jennings: OK. The two main points for me and Dave was around selling site performance in your organization and making it matter to everyone how you do that and then making damn sure that once you’ve done the hard work to make your site faster that it stays fast. We’ve seen in the past where we’ve done a lot of work and made the site faster and you look at it three months later and the great work people did deteriorated because we’ve added 16 new features.
Joshua Bixby: Sure. So what’s the secret to that because I spend a lot of time as you know talking to CTOs, CMOs, CEOs, and I can get them excited about performance, particularly when there’s a problem, an outage I can get energy and the when the daily grind of we’re a few kilobytes larger we’re a few roundtrips more or you know this week to last week and you know 20 milliseconds it wanes and this garbage sort of the it’s kind of like how glaciers form there’s a layer of snow no one cares that much then another layer and another layer and then you look up and you’ve got a kilometer of snow above you. Have you guys found tricks to or have you even gotten to the point where you can see you have a strategy that overcomes that continual sort of you know slow demise slowdown.
Mark Jennings: Yeah so this morning Dave talked a lot about how we’re using metrics. It’s really about having those metrics in everything we do and having everyone in the organization bought into those metrics. Not bought into but believing in those metrics and we do really track them every day and every week we’re having conversations with not just inside the technology team with marketing with editorial about how those metrics are moving, what that means for us. Look, there’s no real secret. It really is about making sure you understand what is happening on your site making sure people are vigilant and passionate about what’s going on on your site and sharing with other people.
Joshua Bixby: Have you seen clear evidence that when you make your site faster, your business is better?
Mark Jennings: Oh absolutely, absolutely.
Joshua Bixby: So you’re a complete convert?
Mark Jennings: Absolutely. I don’t think there’s anyone left at lonely planet that’s no longer a complete convert. And it’s been a long and difficult journey to get here.
Joshua Bixby: Yep
Mark Jennings: You know I’m talking two year two year journey.
Joshua Bixby: Yeah, well we started talking two years ago.
Mark Jennings: Yeah, yeah exactly. So you know it really it mattered to lonely planet in just after Velocity 2010 when we’d seen Shopzilla and everything they’d said it’s easy. We knew it all then and there but the big challenge that we faced was getting other people on board and I think you know that as a technology group we didn’t do that in the best way and there were some things that we did that ultimately hurt us.
Joshua Bixby: Sure
Mark Jennings: And it wasn’t until a more recent experiment that was actually part of a series of experiments that was a collaboration between all parts of it on my team that we really got people to understand how important performance is.
Joshua Bixby: When I think of lonely planet, I think about books, I think about travelling to exotic places off the beaten path. That’s probably another guide book I should mention. That’s a guidebook name isn’t it? Isn’t there something along those?
Mark Jennings: If it’s not it should be.
Joshua Bixby: It should be but the idea of going places other don’t on a budget that isn’t 5 star hotel. How does that translate on the web into more than just book sales. What is the interaction you guys are looking for on your site?
Mark Jennings: Yeah, truly we do cater to that really exotic travel that really you know we call it travelling on a shoestring. But right through to it, if you want to travel 5 star you can travel 5 star with lonely planet as well. On our site, it’s really about having rich content that’s both relevant up to date and then the new part to that that takes it more than just book sales is bookable. So we want people to come to us we want them to read our content if you want to go and you want to visit the gorillas, you can do that. We can tell you who to talk to, where to stay and then sure you should go buy our book but we can also help you book the hotel, we can help you book the tour. We have no figures on this but I’ve got to imagine after the last sort of forty years of book sales we’ve facilitated billions of dollars of well maybe not billions but
Joshua Bixby: Oh I’m sure, I’m sure
Mark Jennings: A lot of transactions, whether it’s hotels or tours and what we’re looking to do now with online is it enables us to say yeah we’re going to facilitate this but we also want to be involved. Come to us, see that content, make your decisions and we’ll help you book it as well.
Joshua Bixby: One of the things that was interesting from one of the presentations yesterday that Tim did. Did you catch Tim Morrow’s presentation, the Betfair one?
Mark Jennings: Yes
Joshua Bixby: Where he was talking about the fact that they had you know a certain SLA, they hit it and RUM looked great but when they actually dug into the interaction the customer had, it was negative. It was not as good as they’d expected. And it’s got me thinking about this idea that we track you know it’s getting easy to track from when you clicked to when some random onload event fires, but that doesn’t track a big chunk of the interaction that the customer’s having putting things into a cart, doing a discussion, whatever it is. Things that don’t necessarily play into this clear paradigm of click to click right? How do you guys measure you know interaction do you look at for example time between onload events, are people spending too much time trying to find that next page or how are you guys looking at the sort of the dark time that isn’t tracked today or do you?
Mark Jennings: The answer is we should. So yeah we don’t at the moment. We don’t have that kind of data. That’s a really interesting one, it would be really hard to try and measure how long is too long. I mean we want people a lot of our pages are really about providing a lot of content to users, and we experiment a lot with how we provide different content, and we provide content in different ways to users but we don’t really track at the moment how long people are spending on those pages.
Joshua Bixby: Yeah, it would be an interesting concept to figure out because as you say it might be a great thing that somebody spends 3 or 4 minutes, maybe that is a good sign, maybe that is a bad sign. It is difficult to know, it is interesting as we think about tracking how those, I mean it is much more about usability question I guess in many ways and a content question but it is interesting nonetheless. Have you learnt anything at Velocity this year?
Mark Jennings: Oh absolutely.
Joshua Bixby: And what, like, anything that you sort of are going to sort of take back and evangelize?
Mark Jennings: One thing is hooking yslow into jenkins and I cannot wait to do that. Actually we already a colleague back in Australia twitting it that he loves that.
Joshua Bixby: Ok why, what would that do for you that you don’t have, what benefit that you don’t have today?
Mark Jennings: Well actually at the end of our presentation today we had a question for Steve Souders which was what metrics do you have pre-release that tell you what the performance is going to before the release.
Joshua Bixby: Good question.
Mark Jennings: And the answers was, we don’t really.
Joshua Bixby: Yeah
Mark Jennings: We rely on releasing to small amounts of traffic and measuring that, the user interaction at the moment and to be honest I kind of hope we don’t change this. We don’t have any kind of pre-prod environments, everything goes straight from you know a commit to production and I love that. So hopefully this will give us that little bit more insight, just to make sure, one extra check of should we be deploying this. The other big one, the other presentation that I loved is about Google, the dev tools and I did not know Chrome dev tools could do that because I sure, I did not know Chrome dev tools could do all that.
Joshua Bixby: Yeah, it is amazing, isn’t it? The depth of which they are developing those tools, and where they are going. I did not see that presentation, but I got a chance to see a similar one and I was just amazed.
Mark Jennings: On the surface, the tools are amazing and some of the things it can do
Joshua Bixby: And then you watched a power user get in there
Mark Jennings: Wow yeah, yeah absolutely.
Joshua Bixby: Have you been to Borough market?
Mark Jennings: No
Joshua Bixby: OK. So I took off for an hour, hour and a half, I decided it was time to leave the hotel. I went to Borough market; this is a place to go in this town.
Mark Jennings: Ok
Joshua Bixby: So I sat and had beautiful fresh oysters, gorgeous, and then a pork belly sandwich with some nice crackling. It is worth, and then you can get your glass of Prosecco and walk around. It is a good date type of experience. There you go, that’s my London tip, tip for Londoners, I was quite impressed. My brother told me about it, he was here a month ago and he loved it, so there you go that’s a tip for you.
Mark Jennings: I should make sure it exists on our website.
Joshua Bixby: Yeah, I am going to check it out. I am going to check it out.
Mark Jennings: Fingers crossed.
Joshua Bixby: Velocity, how many people from Lonely Planet come? Would you send more if you could?
Mark Jennings: So in terms of numbers, we sent one person to Santa Clara in 2010, two in 2011, and three in 2012. But we’ve got, coming all the way from Australia everyone going to Santa Clara is
Joshua Bixby: A big expensive
Mark Jennings: Yeah is big. I think we’ve got about 8 or 9 people here this week. We should have had a couple more but they couldn’t it make so, we sent as many people from our products, from our dev and our ops teams down here as possible.
Joshua Bixby: That’s amazing. How big is the team, like in round numbers?
Mark Jennings: It is relatively small, our entire online team is about 45 people.
Joshua Bixby: OK wow
Mark Jennings: But the sort our delivery team is only about less than half of that, so we have about roughly speaking 12 developers and a couple of ops people. That 45 includes editorial, marketing, finance.
Joshua Bixby: Ah ok, everything.
Mark Jennings:So I think for what we do right now, we are a really small team.
Joshua Bixby: You’re tiny.
Mark Jennings: Yeah.
Joshua Bixby: I did not know you were that small.
Mark Jennings: No, no that surprises a lot of people.
Joshua Bixby: That’s surprising to me.
Mark Jennings: Yeah. Seriously what we can manage to do with that few people over last year has been phenomenal.
Joshua Bixby: What are your biggest pain points today as a business? Like what are the things if you could have a session or attend something that says I solve this problem, is there something which sticks out that you guys are really struggling with? That you can share. I am sure there are a lot of things that you struggle with that you can’t share, but any that you are struggling with that you can share.
Mark Jennings: I am trying to think, what are really the true pain points for us. I mean I think in many cases we are so early in what we are developing right now that we haven’t hit those true pain points. We are going through so much change, we are in the fairly early phase of some of that that’s really hard to say this is what I think is going to be a real killer for us.
Joshua Bixby: It is like everything is a pain and everything is not a pain. In a way, you are sort of
Mark Jennings: Yeah, that’s right. I mean, I think I some point there are going to be things that we ignored
Joshua Bixby: Yeah
Mark Jennings: Or we have taken a short cut to things, which are going to kill us later on, but I think at the moment, we are comfortable with these shortcuts.
Joshua Bixby: Use of the cloud. I know you guys are heavy cloud users, and increasingly so. What has been your experience you know, moving an organization into being a I think you guys are a pretty cloud friendly organization.
Mark Jennings: yeah, so at the moment most of our site runs from a colo, we manage the hybrid ourselves but we are in a transition to move everything to the cloud. Our experience so far has been alright. We read the stories of other people that been burnt by the cloud but it is working well for us. The thing, the real killer thing for us is two things: one is how quickly we can develop, we can get products in front of users that we couldn’t do before and the other one is just a rich tool set that using a cloud provider and that is their bread and butter and we are never going to be to cap that so why try. And those two things go hand in hand.
Joshua Bixby: And cost wise, though that’s probably not your department, it feels accessible and reasonable when you look at this over a 5 years horizon?
Mark Jennings: Absolutely, we just came off doing a cost analysis of moving from our own, basically running it ourselves to running it in the cloud and it pays itself off in less than 3 years and that’s ignoring all of the other benefits
Joshua Bixby: Yeah the benefits just talked about. This is pure cost side. How about multi-cloud. Do you see a world, where you know, EC2, Rackspace and Azure all at the same time. Is that a world you are preparing for or is it single honed.
Mark Jennings: I think at the moment for us it is really a single origin and potentially multi CDN.
Joshua Bixby: Ok
Mark Jennings: Really for us, it is at the moment looking at the moment one single core provider in multiple regions.
Joshua Bixby: Ok
Mark Jennings: Whether it is AWS or Rackspace. If we take AWS as an example, theoretically if we are running in the US and the UK, then there is very little which can bring down our site completely for a significant period.
Joshua Bixby: You should knock on wood, is there wood anywhere here?
Mark Jennings: Yeah there is, right there.
Joshua Bixby: Phew, because this is being recorded, you know.
Mark Jennings: I did preface it theoretically.
Joshua Bixby: Cause there’s going to be a point where this is played back to you.
Mark Jennings: Oh God.
Joshua Bixby: How difficult would it be, the reason I am asking is because I was having this conversation with a friend of mine about whether the world giving the commoditization of the cloud providers whether the world is going to be to a multi-honed cloud environment where you could load balance between clouds in the same way people have been talking about load balancing between CDNs.
Mark Jennings: Yeah
Joshua Bixby: How difficult would that be as a technical problem for your business to be active in multiple clouds at the same time?
Mark Jennings: I think it would be a significant challenge.
Joshua Bixby: More on the database side is that where you imagine or just across every tier.
Mark Jennings: You know to be able to manage the config and the different api’s, particularly if you are looking at open stack versus AWS, it would create a fair bit of overhead in trying to do that. Yeah but you’ve got companies like says, not exactly the same as a Cedexis who helps you manage multiple CDNs, maybe there is an opportunity in there for someone to come in and play that. But I was reading something the other day about the biggest risk with cloud is vendor locking. I think to an extend that is true but it is not weighing on us too much at the moment.
Joshua Bixby: Last topic for you, mobile.
Mark Jennings: Oh yeah.
Joshua Bixby: Tell me, I mean I would imagine that as a traveler, someone who travels a lot, I spend a lot of my time in foreign cities, countries looking for information, looking for that small restaurant that is going to serve me up the local delicacy or that hotel, although me of my hotel is booked by other people, how is that interaction how is mobile as a part of your business, important, growing. Tell me about how it is thought of internally?
Mark Jennings: It is both important and growing. We look at our analytics and the number of requests we see coming in from mobile devices has grown phenomenally over the last couple of years. There was a stat on this the other day about, this is going back almost 12 months now, I think tripled in the number of months. So yeah it is massive. So I guess in managing this is kind of three ways: we’ve got a team of people based out of the US who build our mobile apps, primarily iPhone but you can get a lot of cities on there, you can get a lot of countries on there, for the online team which is us here in London mobile is a core requirement, we expect that alongside this idea of screamingly fast website, mobile accessible and friendly website is core to what we do. There is a third part to this which is the kind of social mobile, which we are trying to tackle, called wenzani.
Joshua Bixby: Which is
Mark Jennings: it is in the Apple store right now, it is actually an off-shoot kind of small start-up, which is really about trying to take travel social, I am definitely not the right person to explain what it is but the point I am trying to make is at least 3
Joshua Bixby: There is a couple of different facets to the strategy
Mark Jennings: Yeah there is at least 3 different parts at Lonely Planet which are really focused on mobile and understands that it is a key part of what we are doing.
Joshua Bixby: And do you guys have testing capabilities? Do you have, are you prepared as you are in the desktop environment to test and understand who users use it. Is that as developed as your desktop world?
Mark Jennings: Definitely not, definitely not. It is going to be, become a challenge for us. The metrics infrastructure that we are building will allow us to capture those stats. I think the real challenge is going to be how do we make sure that we, in the same way we have Webpagetest now and I have seen Webpagetest has extended to mobile but look at that waterfall, it is really well understood, you can glance at it and you can get it, I think the challenge is taking that to the mobile world as well.
Joshua Bixby: Thanks for sitting down with us.
Mark Jennings: My pleasure.
Joshua Bixby: And let’s keep in touch.
Mark Jennings: Alright Cheers.
Joshua Bixby: Take care
Mark Jennings: You too.
Joshua Bixby: I really enjoyed that chat, I hope you did too. Big thanks to Mark for joining me. If you want to check out any of the links we talked about you will see them on the side of the page at Webperformancetoday.com/podcasts, where I encourage to check the interviews that we’ve done with some other guests. If you want to drop me a line, you have a comment or some feedback, please do so atJoshua@webperformance.com . I did get some great feedback that somebody told me that I was like Anderson Cooper, the team here said that I was more like a homeless Anderson Cooper given the beard and the hoodie but I took that as a great compliment so more compliments can come, Anderson Cooper is much better looking than me I must acknowledge but it is nice to hear feedback positive and negative. Have a great week.