Using a chatbot template

How pre-made content drastically cuts chatbot assembly time

Written by Maja Olsen and Øyvind Sund.

Before embarking on a chatbot project, everyone involved needs to consider the following: What do users want to know, and do we have time for all this work? In negotiations, prospective customers might consider these questions and disregard the idea of a chatbot altogether, thinking the task is too extensive. Once we let them know that they can “cheat” by using our pre-made chatbot templates instead, there is far less concern in the meeting room.

There are different approaches to building a chatbot. Many customers prefer a 50/50 solution, where some of their employees — often from customer service — work alongside professional bot builders on assembling the chatbot from scratch. This allows the customer to more easily get a say in what questions the bot should handle, and they feel a sense of ownership to the chatbot they are about to adopt.

But, chatbot building can be a time-consuming process and not all companies can afford to assign manpower to the bot project for the necessary amount of time — not to mention how swamped with work these employees already are. For chatbots that are supposed to cover a wide range of questions, using pre-made content can save valuable time and resources.

When they just don’t have the time

There are several perks of implementing pre-made content. Whether AI companies call them templates, modules, plugins or so forth, the idea is the same. They offer a finished layout, already trained with sentence examples, a set-up for internal and external links and suggested answers — ready to be applied to any chatbot.

Focusing on a specific sector or purpose, a template will handle recurring enquiries in that area. For instance, the hotel template from Convertelligence is able to answer a wide range of enquiries a hotel visitor might have, including questions on bookings, customer service, facilities, and membership details.

This template was essential when building our largest chatbot yet, the Thon Hotels front-page bot. With all the enquiries the chatbot needed to answer, building it from scratch during a limited project period simply wasn’t an option.

Aside from hotels, the opportunities are endless. Any industry that sees recurring questions from customers, or internally from employees, could benefit from a chatbot template. EHS, retail, transport, and education or information industries — wherever there are common routines, a bot template can automate the chatbot building process. Why make chatbot content several times when you could make it once?

By implementing a pre-made template, the customer is free to focus on fine-tuning the answers and perfecting the tone-of-voice. The time often spent on building the bot can instead be invested in testing and polishing.

Does that mean I have to make up intents?

Pre-made content is not based on guesswork. Before creating a template, days go into studying real user data and uncovering the most frequently asked questions in the field. There are two main methods of gaining this insight: data analyses and field surveys.

A data analysis is conducted by a developer from Convertelligence and is based on collected data from previous customer communication, such as chat logs and emails.

The field survey is performed by copywriters from Convertelligence who visit companies over a certain period of time, talking to relevant staff, conducting surveys and listening in on conversations, among other things.

Combining both a data analysis and a field survey will provide the best overview. When building a chatbot template, it is essential that we collect information that can be applied to the entire field. However, this also makes the content generic.

Using pre-made templates will provide the bot with far more content than building it from scratch in a limited amount of time would have done. But, the answers still need to be tailored to fit the specific customer.

Can I use these intents again?

One of the advantages of having customer employees collaborate in the building of the bot is that they become so-called superusers, trained in bot building and able to maintain the bot internally. With bots built on pre-existing content, a superuser will have to be trained in other ways.

Without helping to build it from scratch, superusers might lack the same sense of ownership. Convertelligence will in these cases train superusers by involving them in the planning and testing phases, as well as arranging workshops and other bot events, geared towards turning customer service agents, marketers, web designers, and nurses into bona fide chatbot builders.

Provided that the transfer of ownership and training of superusers are handled properly, using a template as the starting point for a chatbot will save massive amounts of time. Chatbot templates are the most intelligent way of automating a bot builder’s copy output, because training a bot is meticulous and time-consuming work.

Always think “Can I use this chatbot content again later?” If the answer is yes, feed the content into a template instead of straight into a bot, and you will have the content already built and trained the next time a bot requires these intents.

Pre-made content is, in other words, an excellent method of drastically cutting chatbot assembly time!