Identifying potential use cases for bots

Knowing how to build and use a bot is great, but completely useless if we don’t know where we can use them.

To begin identifying use cases, we need to first know what it is that we do. Majority of organisation’s work load involves answering queries. If you have a product that people use, the clients will often ask repeated questions, like when do you open, where are you located, etc. Despite such information easily available, people prefer to ask again. We can build a hybrid bot using FAQ, and contextual bots to answer. Suppose if the client is still not satisfied, we can then seamlessly transfer the conversation over to a human agent.

With the traditional approach where all queries are answered by humans, a significantly smaller number of queries can be answered. Utilising bots also enables to have customer support 24x7, since there are no off days. Automatically storing unanswered questions for an agent to review later and respond if required.

In larger organisations, a lot of queries are internal as well. For example an employee is heading to another office location on work. So they would contact the immigration team to help get the visa ready. In said process they might have multiple queries as to what documents need to be submitted, what might apply to them, etc. Hence, we can reduce the workload of the immigration team by answering majority of the queries by a bot.

Some organisations would even like to showcase their use of technology by the use of bots and automation. Some key points that are achieved by use of bots:

• Personalised response — Each user gets a more personal feel to their responses and feels important. As we have a slew of information on the user, the content can be customised to what they prefer instead of giving a generalised response with a huge block of information.
• Faster response — This one is fairly obvious. Since the bot’s capability to answer to users simultaneously scales with your backend fairly easily saving you on cost.
• Improve response rate — Most users get frustrated or angry when they are on hold for a long time. As such potential business and good will of the users are lost. We can make each person feel special.
• Automate those pesky simply questions — As we discussed, users like to ask questions to which the answer is obvious. They do this not of malicious intent but rather due to lack of confidence. They want to confirm their answers as they are unsure. Giving them the confirmation is key to gaining their confidence and trust.

How to go about actually having it built?

If you’re now intrigued and want one in your organisation. Here are some key milestones that need to be met or thought about as you go along:

1. Defining goals — Decide exactly what you expect the bot to accomplish. Be as specific as you can. A goal such as answer customer queries is very vague, instead something like answering frequent customer queries and having it fallback to an agent if it fails is a more concrete goal.
2. Decide on the platform and channel — Find out where you want the bot to respond and how should it respond. Twitter, Facebook, Google Assistant, Cortana, Alexa, or any other platform like web. What kind of platform do you want to build on might be restricted by the channel. Check out the previous post on platforms to learn more.
3. Decide on method of creation — Will an in-house team build it? Hire an external agency? Or are you going to use a prebuilt bot and customise the content?
4. Launch — After deciding on all of these, actually have it built and go for the launch. Receive the feedback and iterate changes as required.

Build from scratch or use prebuilt tool?

Points to consider while using a prebuilt tool:

• Cost — It’s much cheaper.
• Speed — It’s much faster to build.
• Customisation — Very little customisation can be done.
• Data security — Control of data flow will be done by the platform.
• Dependency — You’ll have to depend upon the provider for maintaining your bot and data.

Some popular platforms for building chatbots are:

• Chatfuel,
• Beep Boop,
• ChattyPeople,
• Botsify,
• Smooch.io.

Points to consider while building from scratch:

• Cost — Usually higher.
• Speed — Usually slower since all the components need to be built.
• Customisation — Extreme levels are possible since you’re build the entire bot.
• Data security — You’re in control of the data flow and are free to manage it as you see fit.
• Dependency — Since it’s your in house team, you are only reliant on them.

So how to decide?

You will need to weigh the various pro’s and con’s for your use case and decide accordingly. This technology is still at an early stage and many business are trying to utilise them for improving productivity. You’d also need to consider the cost savings that it would enable your organisation to make. In extremely rare cases, it doesn’t make financial sense to have a bot.

We’d need to ensure that the bot solves a problem without creating a new one. Sometimes, the hassle of accessing a bot maybe too much for some clients. Also, not all bots that perform customer support is required. As your customers may perhaps never ask you any questions, or you’re a contractor and all the questions that are asked are different in nature due to the client.

Conclusion

In essence, we need to go thoroughly think this through. A good step would be to begin the conversation with your colleagues or management to create an analysis report. We should also remember that these bots don’t get created in a day or even a month. It takes time to have a good working bot. We need to ensure that our expectations are correct and reasonable. Since aiming for the moon may lead you to be disappointed with the results. Overall, the case for having a bot is overwhelmingly positive. Hence, do not miss the opportunity to start integrating bots into your life.