Reinventing the guidebook
Eighteen months, six experiments and a mission to build the ultimate list of things to see and do in the world.
Way back in March 2012 I joined Lonely Planet’s Online Editorial team as the Digital Editor looking after the “destinations” content online. Basically, if it was in a guidebook and now on the website, it was my job to keep it up to date.
Apart from being a dream job, the task seemed like a good one to sink my teeth into. However, within weeks it became clear that there was a much bigger project to be worked on.
How to make money online
Lonely Planet has set up one of the web’s first travel forums back in 1996 giving it an early start as a digital innovator. They put their content online for free and by the time I joined even the legal department were using Agile to streamline their processes.
Yet, they still hadn’t found a way to make it pay.
For Lonely Planet to complete its transition to digital it needed a business model that could fund expert authors in a world of UGC competitors and Google Adwords.
Getting experimental
I was seconded into a small six-person multi-discipline team and set the challenge of creating a Minimum Viable Product that would multiply our online revenue by around a factor of five.
As the only content person on the team it was my job to make sure the team’s ideas continued to represent the Lonely Planet voice and ethos and that the user was still getting the quality advice they expect from the world’s leading guidebook publisher.
Working with a UX designer and someone from online marketing we would test concepts while making sure the quality of advice remained the same and could be built from pre-existing content. I would deliberately search out edge-cases where the ideas fell down so we could design a product that worked as well for Bamako as it did for Berlin.
Because each person in the ‘Experiments team’ came from a different discipline we needed something to rally around — a shared mission that would help us find common ground. We found this in the statement:
To find and curate the best sights and experiences in the world and make them bookable.
By blending editorial reviews with online booking, Lonely Planet could use its editorial strength as its USP in attracting customers who will book.
The end of editorial independence?
What’s the point of listing a hotel you can’t book? Is it ok to include a bookable tour that hasn’t been reviewed? By focusing on the user even these thorny questions became simple. In most cases, a more commercial product provided a better experience and where it didn’t we held the line.
Through careful work (you wouldn’t believe how long it takes to manually match tens of thousands of hotel reviews with their bookable counterpart) we were able to draw the editorial and commercial offerings together without compromising Lonely Planet’s editorial integrity or risking the future business model.
Scaling up
Once a workable hypothesis had been tested and validated we started to think about how it would apply to the whole site.
Destination guides on lonelyplanet.com present an interesting content challenge, varying from major world cities like San Francisco with enormous amounts of content to tiny villages like Kep in Cambodia which have just a few details on how to get there.
Not only did the amount of content vary significantly but the types ranged from opening hours to large image galleries, bespoke video guides and zeitgeisty articles on the latest trends.
Defining a content strategy that would work across the entire site, providing an experience for every destination while prioritising the top places that were key to our business goals was essential.
Luckily Pareto was on our side. 95% of people who visited the site looked at one of 650 destinations and 50% of people who visited the site looked at one of 100 top destinations. I used these numbers to sort Lonely Planet’s destinations into 3 categories: Gold, Silver and Bronze.
Bronze content was anything outside the top 650 destinations. All the content would be arranged on the page algorithmically using rules that that the team and I had defined.
This covered the vast majority of the site’s pages but actually a relatively small percentage of its visits.
Silver content was anything that was in the top 650 apart from the top 100. These destinations got hand curation of pre-existing content into two collections “Top things to do in x” and “The best places to stay in x”.
User testing showed that these two categories were consistently the most looked for across the site and allowed us a quick way to bring the bookable elements available in that destination to the fore while showcasing the best that Lonely Planet has to offer.
Both of the categories also stood as a clear ‘flag in the ground’ — Lonely Planet is famous for having opinions on a destination and these categories allowed us to showcase that voice to whole new audiences arriving by search that may never have heard of it.
Both headlines were selected because of their strong performance in search but identified user needs through social sharing and user testing made sure the value proposition was clear.
Gold destinations (the top 100) also got the same “Top things to do in x” and “Best places to stay in x” collections plus a flexible number of thematic collections based on the popularity of the page.
These thematic collections said as much about a place by their existence as the content within them. For example, the fact that Berlin has a collection detailing its nightlife whereas Rome has one about historic churches tells you something about both destinations at a glance.
Each of these ‘gold collections’ was created through a process where a one of our guidebook authors or editors would come up with ideas and then we would whittle away the ones with no potential search traffic until only those meeting a demonstrated user need were left.
All told, just over 2000 collections we commissioned and curated over a period of about six months. Almost all of these re-used existing content in one form or another allowing significant savings over a straightforward commissioning project.
Creating the collections
Building collections required a lot of co-operation from different teams. Although our new grid-based design was completely flexible we decided for launch that items of content would follow a standard pattern within the grid across the site and then we would iterate as performance data became available.
One of the simple tools we used was to draw up the grid on a whiteboard and then have anyone who was interested crowd around and discuss what they thought should be included and in what position. Even seasoned editors found this process invigorating as it challenged them to see their destinations from a whole new perspective.
Within these collections we asked whoever was curating to include everything that was editorially relevant but to try and aim for no more than 16 cards so the information would still feel managable on when viewed on the mobile site.
One of the trickiest tasks in the project was finding appropriate images for all the key sights in the collections. This involved significant image research and a very heavy reliance on Flickr Creative Commons — thank you to all those that listed their photos under that license by the way.
Not only did we have to find images with the right visual dynamic but we had to fact check that they were of the thing the review actually mentioned — not an easy task when you’re talking about a remote hostel in the Himalayas.
As eagle-eyed readers might notice, the text on the individual reviews cuts off after a few words and you have to click on the ‘card’ to read the full review. Originally we planned to rewrite this text so it would form a neat sentence but decided the resource requirements were too high for the minimal increase in click-throughs shown in testing.
Instead, instructions to write the review in a way where the first sentence can be pulled through nicely have been passed to the guidebook authors and the changes will take place over the length of the book update cycle.
Simplifying and standardising
Even though curation allowed us to serve 95% of users pretty comprehensively, we decided that there was more we could do to improve our long tail offering and that these benefits would feed back into our more popular destinations.
A good example of this is the way that Lonely Planet categorises its reviews. For example it might call a restaurant “pan-european” in one book and a “european” in another.
We decided to combine down these categories into a list of around 200 that could then be used to automatically generate collections containing enough content that they would be able attract long-tail search traffic.
Going live and measuring sucess
During the rebuild we took a deliberately lean approach, treating every feature and piece of content as a hypothesis to be validated. That meant publishing early and iterating often.
When the collections first went live only the “Best places to stay in x” ones were initially available. This allowed us to test our hypothesis that curating editorial content and bookable items into the same list would provide a meaningful revenue stream for Lonely Planet. Only once this was validated did the serious editorial work on creating “Top things to do in x” and the ‘gold collections’ begin.
The ‘gold collections’ were again commissioned in a way that was designed to be the minimum needed to see whether thematic content would provide booking opportunities and revenue in that destination while also serving user needs and meeting brand expectations.
Where editorial work demonstrated its worth e.g. image research, it was rolled out at scale and where it didn’t e.g. review ‘standfirsts’ it wasn’t.
Overall Lonely Planet estimates the project has led to around a 40% increase in revenue generated through its online booking partners while providing a better editorial experience and new opportunities to gain traffic. Not bad for eighteen months.
Want to know more? Say hi @tomhewitson on Twitter or at www.tomhewitson.com
