Our client is a large company in the United States. Their customer-facing team deals with millions of users every month trying to get the best deal they can based on their personal circumstances. It’s a billion dollar business that deals with very complex queries, forms, and customer interactions.
In 2015, our client started looking into chatbot technology. As the idea of turning their complex forms into simple conversational chatbots appealed to their sales and marketing directors, a proper plan was put into place. The team picked a chatbot platform and got to work. Soon enough, they hit a wall… (much like any company who tries an off-the-shelf solution).
The chatbot DIY platform limitations
At ubisend, we are big fans of DIY chatbot platforms. They’re amazing to draft ideas, test the waters, even roll out a super-basic MVP. Beyond that, though, we always advise going custom. This case study is testament to that advice.
Our client had made great process with their platform chatbot. It worked sufficiently enough to understand the flow users would go through, draft language, and even perform some early demos to key stakeholders. Unfortunately, this is all the chatbot could do for them.
To make the chatbot worthwhile, it needed to connect to multiple internal APIs. This, by itself, should take you away from using a platform chatbot. On top that, they needed 24/7 control of the bot (to guarantee uptime), retain complete control of all the data parsing through it, and customise it's personality.
Migrating a free chatbot into a professional build
ubisend was brought in to do something we’re often tasked to do: migrate a free DIY platform-based chatbot into a professional conversational interface. The platform chatbot had served its purpose and proved unable to assist our client any further. It was time for the big boys to step in.
The first thing we did was to study what had been done so far. We certainly didn’t want to dismiss everything. There is a lot a company can learn from using a platform, and we want to make sure we capture this knowledge. We found one conversational flow in particular that worked quite well within the original free chatbot.
We dissected this particular flow, laid it out on a whiteboard (as we would during a chatbot user story exercise), and rebuilt an MVP version of it from the ground up. Once that was done, we started to make parts of the conversational inputs available via API. Instead of requiring the user to input something our client already knows about them, we were now grabbing it in real time from their database. Win.
Finally, we built them a lightweight content management system. With it, they gained full control of the bot’s language, from small talk to deflections and human takeovers.
Building on the solid groundwork
Today, we have a solid chatbot to start growing and optimising. Since the main conversational flow our client wanted is complete and performing well, we focus on making their chatbot smarter and more efficient. We pinpointed a few ways their customers tend to find and first interact with the chatbot. This is a perfect opportunity to offer a more customised experience and get better results.
Only high-end, custom chatbot development can allow this sort of improvement. This is the power we bring to the table.
Due to the confidential data we help our clients manage, we often work under hefty NDAs. If you're interested to know more about this project, get in contact, as we may be able to introduce you to the client for a chat.