Dreamforce 2019 - bringing the personal touch to vacations with Luxury Escapes

Profile picture for user slauchlan By Stuart Lauchlan November 20, 2019
Summary:
If you're planning a high-end vacation, Luxury Escapes aims to help deliver with an AI-enabled highly personal touch.

luxury escape

There are two things that Australians love, according to Jason Shugg - beer and travel. And as Chief Customer Officer of Australian holiday Luxury Escapes, he should be in a position to know.

Luxury Escapes has offices in Australia, Singapore, India and a recently opened arm in San Francisco. Over the 8 years it’s been in business it’s scaled up to the point where last year it sent over half-a-million travellers on holidays. Shugg says:

What we do is we try and inspire and connect people with luxurious travel experiences around the world, Maldives, Bali  Mexico, wherever those might be. We are for global travellers, so while the business did start 'down under', we are seeing our international business ramp up, with 20% of our business now coming from outside Australia.

There are a number of important differentiators between Luxury Escapes and other vacation platforms, like Expedia or Booking,com, he argues:

It's a membership-based program, but all that means is that you join up by email, it's free, and then we send you usually a daily email inspiring you about the amazing places you can travel in the world. At any one time our website only has around 70 offers on it. The way we work is that we go and we find amazing deals, we curate them, we handcraft them ourselves. We add a bunch of extras into those deals and then we put it up on our website or on the app for two weeks. We guarantee that we are the cheapest price in the market for those two weeks . And then the deal goes away and it's not seen again for another year or sometimes never again.

In addition, those vacations on offer have been tried and tested by Luxury Escapes staffers, he adds:

For every product you see on our website, we pay somebody to go there, jump on the beds, swim in the pools, drink the cocktails and really immerse themselves in the destination. We provide amazing value.What we do is we provide the right treat, but we also put inclusions in. That might be might be daily massages, cocktails, all-you-can-eat, we figure out what the best deal is.   So it's a great little business and we hope that it's going to become a great big business very soon.

Relevant 

Key to achieving that ambition are two words - personalization and relevance. But to help deliver a vision of using data to provide world class personalized experiences on every single customer interaction, there were some serious challenges to overcome, recalls Lead Developer Colin Pringle-Wood :

Our main challenge was that it didn't matter where you were, if you were a longtime customer or a first time visitor to the website,  if you're interested in adventure or you like city breaks or romantic breaks, when you came to the website you saw exactly the same thing as everybody else. When you opened up your in-box in the morning, you got exactly the same email as everybody else. And if you called up customer service, you’d probably have the same conversation that everybody else had that day.

Two technical challenges prevented the company from delivering a more personal service, he adds:

The first technical challenge was our platform. We had an old creaky website. It took a lot of tender loving care to keep this thing going. There was only a couple of people within the company that knew how to keep the thing running. We also had decentralized teams working across time zones so we were finding it really hard to collaborate. And most of the time, all we were doing was trying to solve problems instead of building that next feature that the customer wanted. When we finally did get around to building the next feature, we were stuck with so much technical debt that we were finding it harder and harder and harder to build those features.

Our second technical issue was we had siloed data. We had many different parts of our ecosystem, and none of these parts spoke to each other. So querying data across any of these systems in order to create anything personal was near to possible.

The solution was to rebuild the website and the mobile app using Salesforce’s Heroku. Pringle-Wood explains

Heroku solves a lot of issues for us. No longer do we need to maintain databases, keep servers up to date, deal with routing configuration - Heroku does all of that straight out-of-the-box. It's also about having a large ecosystem of add ons that we use these extensively to help us build the next feature for customers.

Now we can release multiple features in any given day. We've got our website and apps on Heroku and we sync Heroku, Service Cloud using Heroku Connect. That allows us to get the databases in Heroku and the tables in Service Cloud into sync.  We migrated away from our old email service provider and we integrated Salesforce Marketing Cloud. We now use Marketing Cloud Connect to keep Marketing Cloud and Service Cloud in sync. We use a range of Marketing Cloud APIs to read and write data between Heroku and Marketing Cloud. Now, within any node of our system, we can tell who viewed what, what phone calls were made to that person, and what sort of customer preferences that person has.

With the ability now in place to build new features, the next objective was to come up with the tools to enable personalization capabilities:

We needed to be able to track every single interaction that a user makes with us, really listen to them and once we found out what they wanted, we needed to be able to make recommendations at scale. And we needed to make sure that all these systems work in harmony. We didn't want to give the customer some sort of disjointed experience.

For the listening to customers requirement, the answer was found in Interaction Studio and Salesforce’s Einstein AI tech:

This allows us to listen to each customer on each touchpoint. Then we can make a recommendation for that customer. Einstein gives us the 'wisdom of the crowd'. User browsing behavior is used to recommend products for that next time the user interacts with us. Equally as important, as we didn't have the time to build our recommendation engine, or the skill set, this came  straight out-of-the-box with Einstein.

With the addition of “a little bit of custom tech”, Luxury Escapes was now able to build out what Pringle-Wood calls a “Personalization Architecture”:

We've got our custom tech and we put it on top of our Heroku server.  Both our apps and our website communicate with Heroku and can pull up a complete data picture of what a user is looking at. We then send that data on to Einstein and Interaction Studio. We can then easily use those products in order to maybe change the look and feel of the homepage for that user, provide content that we're going to put in an email or provide talking cues to our customer service representatives, so the next time that customer phones up, she can have that first class conversation.

That last point is seen as particularly important:

We take the history of the  browsing data and we show it to customer service agents. What sort of content you seem interested in and what sort of price categories you've been interested in and what's the type of holiday -  family friendly or romantic or something along those lines. So instead of a user just phoning customer service saying, 'Hey I'm interested in going on holiday' and the customer service agent going, 'Tell me a bit about what you like', the agents can say, 'Well, how about Bali?  I’ve got two price categories that maybe you'd be interested in, and here's a really luxurious family friendly resort. How does that sound?' And I think that's a that's a stronger message than, 'Tell me what you want'.

Jess

Pringle-Wood illustrates how all this works in practice using a customer he calls Jess:

Jess is a longtime customer of ours and she's thinking about going on her next trip. She really likes the idea of going to Europe as she has never been before. So she opens up her app and starts to have a look around. She keeps browsing, clicks on one or two deals, but she doesn't see anything that she likes, so she navigates away.

But what's happening at the same time is that we  are sending this data on to Einstein and Interaction Studio. We can see all the events as Jess browsed our app and we can see all the data we collected on Jess's last customer journey. We can use this data, anywhere in our system.

Every single customer that browses our apps or websites has their own Interaction Studio profile and those profiles are as unique as the person browsing. In her Einstein profile, we can see that Jess has a lot of affinity for  Europe, especially Italy and Greece. Einstein uses these affinities to build up recommendations and really uses those recommendations to change the look and feel of the homepage, just for Jess.

This results in recommendations being built up that can be served to Jess the next time she signs in or which will inform the contents of the next email that’s sent to her. Content and recommendations are determined upon and weighted via a scoring processL

Behind the scenes what we also collecting is that all of the data aggregated across all the sessions with us. Some of the things that we collect are the destinations that Jess is interested in, the holiday types that she's interested in and budget categories. And every time she interacts with something within any three of those categories, we ‘count back’. So for example, if she interacted a lot with us for, for example, a deal in the Maldives and continued to come back to our website to look at Maldives deal, the score for the Maldives would increase, much higher than other destinations. If she stops being interested in the Maldives three or four weeks later, that score will go to zero.

Pringle-Wood is a convert to the effectiveness of this way of operating it seems, despite being “a little bit skeptical” prior to implementationL

It was actually a huge increase for us, much bigger than I expected. We can directly correlate that we got a 32% increase in conversion rate, when we started implementing personalization. That came with a 21% increase in revenue and a 35% increase in the number of transactions that are happening on our website, and our apps. We also have a 24% open rate for marketing emails, compared to the industry average of 21% so we're kind of blowing that figure out of the water.

We are sending close to 1.5 million emails a day to email subscriber. We've got a very good subscriber base, and each one of those emails is personalized.  We've also got hundreds and thousands of people browsing our website every day and we are providing personalized content to a majority of those people. 

We have populations that we test against each other and we are continually striving to see what's working for us and what's not working. We are using Google to tag users and certain population groups. That's how we track conversion. 

As for lessons learned to date, Pringle-Wood boils these down to some essential truths:

Having a single customer view is imperative to personalization.  We couldn't have done what we did if we still had that old legacy platform. That was too hard for us to build the next feature or to see what those customers actually want. We also needed the right tools in order to provide personalization and we found those tools within Einstein and Interaction Studio. And we’ve never stopped learning. We take what we've learned, and we feed it back into the system as we continue to try and provide a better and a better experience for our customers.