Chatbots for hotels is a hot topic right now, with more and more hoteliers realizing the high sales potential of artificial intelligence. A hotel bot is capable of making reservations and delivering 24/7 personalized service. I have been working with such chatbots for the last 4 years and sometimes hotel groups of a certain size ask me in confidence, Should we work with you or should we develop our own hotel chatbot?
That’s a smart question to ask. My answer is really clear: while the thought of developing your own chatbot might seem very tempting, it is a terrible idea.
To be fair, we’ll start with the smart arguments of groups that have decided to make their own chatbot from scratch. Then, I’ll explain why I think they have already, or will eventually fail. Or, be doomed to deliver a retarded chatbot because of realities that they are not considering. This is my opinion and I might be wrong (but I don’t think so).
The solution that I would recommend is chatbot as a service.
Building your own hotel bot isn’t so hard, right?
- Chatbot building tools are available for non-developers. There are tools more or less advanced that allow you to create a chatbot on your own. My favorites are IBM Watson and DialogFlow. So basically, you can make your own dialogs and not worry too much about the code behind. How hard can it be, right?
- These tools are not very expensive. So, you don’t take a huge risk. In fact, it might make sense to try it on your own to see what you can do. What you absolutely shouldn’t do is work with a consultant that is going to cost you tens of thousands of Euros for a chatbot that is never going to work. Bear in mind that a chatbot is a living project. It is NOT something that is delivered overnight, which then stops evolving.
- You want to offer a unique experience. When you develop your own hotel bot, you get more control over its evolution. I have also heard that some groups hope to develop a sense of ownership and engagement among the team around an exciting project. Why not?
These are all valid points and there is nothing wrong with them. However, there are some operational realities that most hoteliers are not aware of. And these massively outweigh the arguments in favor of a DIY* chatbot (Do It Yourself).
So, why would a DIY hotel bot fail?
I sell hotel chatbots. So, I have a vested interest in you going to people like me, instead of making your own bot. I am also a recognized expert in the field. At the end of the day, exercise caution and do what you think is best. Of course, I must warn you about certain things that you may not have considered, and which are going to bite you in the bottom. I have had to go through this myself. Even though I have crossed the river and our chatbot has reached maturity, I still have a stomach ache when I remember how hard it was to develop it in the last 4 years. It was so tough to get there that in all honesty, I wouldn’t be able to go through it all over again.
- Terrible time to market. When dealing with innovation, timing is crucial. How long do you think it will take you to develop a decent chatbot? Take that time, triple it and then triple it again, again and again and you might get close to reality. In addition, you’ll spend a year developing a chatbot to fulfil a need that you have today. The thing is, that in 2020, your customers’ expectations will have evolved and you’ll keep running behind forever.
- The chatbot is not the end game. What you’d like right now is a simple reservation chatbot that can handle a few frequently asked questions. In fact, chatbots are not a simple feature that you add to your tech stack. They are a rare opportunity to redefine your whole e-commerce and customer relationship strategy. Customers are mobile and the DNA of mobile is interaction. So your chatbot will need to encompass more and more of your e-commerce strategy.
What matters is not the small thing you’ll build at first. It’s how your hotel bot is going to evolve, interact and transform your tech ecosystem over time and how you’re going to help it grow. In short, the real cost of a chatbot project is not in the initial design budget. Rather, it is in the maintenance and constant R&D that you’ll need, to keep up with customer expectations. And here we’re talking about large investments in R&D over long periods of time. This is not particularly compatible with the short-term orientation of most groups.
- Machine learning is harder than you think. I know your team is smarter than the others, but a 6-month project is not going to cut it. To achieve something, you’d need a team of at least 7-10 people, working for years. In our case, we started by offering live chat services for 2 years and built our chatbot against real conversations. What we really did was accumulate a 360° experience. When working with a partner, you gain immediate access to all its accumulated experience for a fraction of the cost of developing it internally without closing the door on personalization. There are 2 areas of expertise that a good chatbot provider should bring to the table:
— NLU: The ability of the chatbot to understand hotel-related requests in multiple languages. This is hard and painful to develop and justifies the need for a chatbot partner.
— Copywriting: Ability to make sure that the chatbot dialogs efficiently answer customer requests (easier said than done). Content is customizable so the chatbot your get is unique.
Hotel chatbot as a service – the key to success?
All technologies are like living cells. They start as stem cells with enormous potential, grow into maturity and eventually decay into cancer. Technology life cycle is accelerating, so it makes less and less sense for a single hotel group to own its technology as it ages so fast. What is the point of reinventing the wheel? That’s why Software as a Service (Saas) is so popular right now.
The key benefit of adopting a chatbot as a service is to have the perfect time to market. You get a mature product immediately that gives you instant ROI for a fraction of the cost. Also, you do not pay for continuous product upgrades. If/when the current technology you’re using completes its life cycle, it’s easy to switch to the next that will keep you moving forward.
For years you have been stuck with your PMS that was good at first and then grew old. However, you kept it because the down payment that you made when you purchased it was really high. Now, many hotels are finally switching PMS. Do not repeat the same mistake, this time with chatbots. If you find a provider who can cover at least 80% of your need, go for it.
|Build your own chatbot||Work with a chatbot as a service|
|Time to market||Late||Instant|
|Ability to evolve/Switch||Low||High|