So couple days ago i applied for a job opening at Traveloka. Im still working on my dissertation but reading the job description got me interested. Unfortunately i didn’t pass the analytic test. Got bad history with this kind of test. A little bit disappointed but on a bright side this probably a sign that i should gave 100% focus on finishing my thesis.
Since Ive been using their services for quite some time, i had several improvement ideas which i plan to develop inside the job application. I thought this might be useful if i get called for interview. However since it is not my luck this time rather waste the ideas, it might be better if elaborate it more here. Writing it can serve as means to stretch my “idea machine” while also a way to embrace my failure. Hopefully it will be useful in some ways for anybody who read this.
So here we go.
Product Integration / Cross-selling
From tabbed options on the home page, i can see that Traveloka sell flight, train, hotel, mobile phone top up and attraction tickets. Looking at this list, we can quickly deduce it might be better to integrate all or some of them together in one menu. By putting these product in one sales process, it might be easier to do cross-selling in order to increase Average Revenue per Order.
However as we can see in the picture above, the only one that is integrated is Flight and Hotel. Even that, it still has dedicated flight menu on the left. My assumption is flight tickets is the primary cash cow for Traveloka hence to the two separate menu for it.
So what is the rationale for this design choice ?
Integrating multiple product might be a way to increase revenue. But bear in mind, integrating multiple product will probably increase complexity to the sales process. User will be given more steps or prompted more options hence more hassle. More hassle means increased risk of cancelled order. Cant have that for your cash cow because too much revenue at risk. If lets say 70% of user came to the site to primarily search flight tickets, bothering them with hotel options would annoy them.
I noticed that Kayak also does this similar design while on the other hand, EasyJet cross sell everything if you buy directly through their site.
Despite the benefits of no-hassle approache, I believe there is some missed opportunity to not do integration with less risky product. Either with Attraction and Activities which i assume is the least bought product due to its menu placement or maybe even the Hotel one.
As you can see in the image above, within the Attraction and Activities sales menu, there is a page frame on Transportation and Venue info but the information is not that useful. I think this part is better replaced by info on nearest Hotel or available Flight within that date AND the options to book it.
Similarly, in the Hotel menu, there is even information on nearest Attraction however the info is just static text even though Traveloka also sell tickets to that attractions. Might as well put an option to put in the shopping cart.
Looking from the user perspective, i feel like the frame and the content is there already, hence to develop cross-selling approach shouldn’t cost that much. The challenge is to test this gradually using variety of approach within low-risk product both to ensure minimal annoyance to the user and optimal revenue gain.
The ability to cross-sell multiple products will be the stepping stone for my next idea, which is…
Since Traveloka sell flight, hotel and attractions and also data package for international roaming, i wonder why they didnt bundle it up into itinerary. Since Traveloka have the elements already, the next step after enabling cross selling different product is to combine multiple products into one bundled itinerary. User can pick and choose their flights, where they stay, what to do, add data roaming on the top and then charged it under one price.
Itinerary represents total expense of a traveler. I cannot find the data on Indonesia, but from the EU tourism statistics, transport and accommodation account for 60 percent of domestic trip and 74 percent international travel.
Traveloka already aimed for two-third of travelers budgets anyway so why dont make it easier for user to spend the budget 100 percent on the site. Convenience for the user and profitability for Traveloka.
I kept coming back for Easyjet reference because i used them recently. They had this kind of product called Holidays. It has nice filter options where you can choose what type of holiday product you want to get. However it only bundled Flight and Hotel which kind of the same like what Traveloka right now. Moreover, the combination of those two options is fixed.
Local startup TripVisto offer similar package-trip product. However their product mostly doesn’t include flight tickets and yet again, the itinerary options are fixed. With advantage in Flights and Hotels and developing menu where users can combine multiple product in one basket, Traveloka can complete the missing piece of this itinerary product. As far as i know havent been offered by other online travel provider hence it has potential to solidify Traveloka dominance in this space. Tiket and Nusatrip havent even combined Flight and Hotel at this point of writing.
The itinerary product should also leverage Traveloka apps. Once users create their custom itinerary, all of the tickets, schedule and other useful information will be pushed to the app. The app then become one-stop-travel companion with real time information on that itinerary. This is similar like what TripIt does.
In the current app of Traveloka, i can view all of my tickets but its separated by product (flights, hotels, attractions, internet). If the various tickets from flights, hotels and attractions are grouped in a trip, arranged by time, its already pretty similar to TripIt . Hence i assume the development cost wouldnt be too high because all the elements are there. I believe by doing this, it will increase the app usage and stickiness since currently the app is geared more to conduct sales. Having complete itinerary information, user will constantly open the app while travelling not just when they try to plan their trip.
Yet again, the challenge is to ensure the new features developed in the app is essential for users. Putting too many functions will risk making the app bloated.
The third ideas is the culmination of the first and the second one, which is…
Apart from revenue increase, bundling multiple product into itinerary could bring crucial benefits which is inventory correlation data. By enabling users customise itinerary, Traveloka could unlock interesting data points such as:
- Average duration and spending per each customer trip (e.g Single traveller who goes to Jogja typically spends 4 mio and 2 nights)
- Correlation between flights and hotel ( e.g User who fly using AirAsia to KL is more likely to stay in Tune hotel, Business class travellers are most likely to stay in hotel close to airport)
- What kind of activities that different users are most likely to do (e.g Families who go to Singapore for more than 2 days are most likely to buy Universal studios tickets, Single traveller staying in a less than 3 stars hotel in Vietnam are most likely to go to Ha Long Bay )
- Multi-mode transport and hotel connections (e.g Traveler who arrive in Cengkareng and depart to Bali in a 7 day interval is more likely to buy Jakarta – Surabaya train tickets)
- Propensity to use data roaming (e.g Single traveler who buy ticket to Disneyland Paris is more likely to buy higher data package for Instagram stories, Business class traveler is more likely to buy pulsa for calling back home)
This is what i can think of the top of my head right now, im sure there is more interesting stuff can be explored.
With their current scale of inventories, i assume some of this correlation can already be done inside Traveloka. However by offering customised itinerary product, the user will do the item clustering themselves. Hence Traveloka can capture not only customer preference on one product but multiple combination of product within one purchase in a specific time.
This inventory correlation data can be used in a simple way by providing Amazon-like “User who buy X also buy Y” recommendations both within the purchase process or through email newsletter. I also believe that this correlation data could enable something more sophisticated such as intelligent recommendation.
Imagine having the urge to go for holiday but you’re not sure where to go. At this point of time one thing you’re sure of is your budget. You have around 5 to 7 million budget and cant afford to spend more than that. So what do you do ? you consult with Traveloka.
As depicted in the image above, I imagine users can type their budget (e.g “7 million” or “7 mio”) added by specific destination keyword such as “Tokyo” or even just “domestic” or “international” into text box. Traveloka then recommend relevant product such as attraction, flight and also the most common itineraries.
Well, the design could be so much better than this though.
To judge the value of recommendation engine, theres no better example than Amazon. Back in 2011, Fortune reported that successful implementation of integrated item-to-item recommendation increased 30% of the company sales year-on-year. On the other hand, in this classic marketing / forecasting paper by MIT’s Ben-Akiva and Gershenfeld, user-customised-bundling can potentially increased total demand of products significantly given that the bundled-price is reasonable. From both this empirical and theoretical reference, i believe there is big potential in developing itinerary which further improved by intelligent recommendation.
However one could also argue that total development cost both in time and resource will be significant. In this prominent paper developed by Amazon’s Personalization Group, even though item-to-item recommendation brings lot of benefits, the cost of computation is also expensive. It also correlated by the order of magnitude with number of inventories as well.
Hence the key to develop this is by spreading the investments and development into multiple stages (as proposed by this blog post). By doing this, risk will be lower and development team can have gradual feedback which spark further improvement when developing the more advance stage. Ideally, the benefits from the first stage can be harvested while the next stage development starts hence the development cost and revenue gain can be balanced nicely.
Considering the current scale of Traveloka users and inventories, there is huge opportunities in offering personalised travel recommendation services. I believe this services not only could increase average order value but also become an unique value proposition to further solidify Traveloka dominance in Indonesia online travel space.
To lower the risk and spread the investment, this services should be build in stages starting with simple cross-selling mechanism at the start, improved by offering custom itinerary and then developed further into full blown recommendation services.