pitching_hr_marketing_chatbots

When we talk with a partner going for their first chatbot pitch, we often end up covering this topic. Is it best to focus on pitching an internal or a customer-facing solution?

Of course, there are nuances to each. Every client you meet will have different requirements, degrees of risk aversion, and ideas on the future of chatbots.

Below, I will share the typical work we do in both areas. Use this article as inspiration for your next pitch. If it's your first, come tell us how you did!

 

Pitching an internal chatbot

We will predominantly focus on HR chatbots here.

HR chatbots are, by far, the most popular type of bots. About 83% of all the enquiries we get are around HR. This often surprises clients, as most people think of chatbots as a fun, marketing-y tool.

In fact, within the HR space, there is a LOT that can be done from a chatbot. In medium to large organisations, HR is often an after-thought. Resources are focused on attracting new clients, growing the business, making more money. As a result, HR is often understaffed and overworked.

Chatbots can do a lot in the HR space. With most of our clients, we start from a highly focused, simple approach before expanding. You will see this is an approach we use for all our builds.

 

A 3-phase approach to HR chatbots

In HR, our first order of business is usually to scoop all the low hanging fruits. HR is bombarded with queries every day, often the same queries, often at the same time every month. Our first target is to find what these repetitive queries are and build the first instance of the chatbot to answer them.

Through a high-level discovery process with our client, we extract two to four HR areas (i.e. pension, payroll, recruitment, parenting leave, etc.) to focus on. Then, we map out the specific areas employees typically enquire about. For instance, working on payroll, the questions tend to always be around 'why is my paycheck less than last month?' or 'what does this acronym mean?'.

Our goal is not only to answer these questions 24/7 and bypass the need for human intervention. It is also to start building the first layer of artificial intelligence, and start feeding the machine with real data.

hr_chatbots_approach_min

We found that through this simple first process, we are able to save HR staff up to 40% of time.

The second phase after that is focused on expanding the chatbot's field of knowledge. In this phase, we improve the bot's language understanding, add more areas of expertise, and start forking some of the conversations into deeper actions.

For instance, in phase two, if you were to ask the bot: 'How many days off can we get to attend a funeral?', instead of simply answering with the textbook 'All employees are allowed three days off' sentence, it would go a few steps deeper.

'I'm sorry for your loss. Was that person a friend, relative, or close relative?'

'Relative'

'Thank you. For the loss of a relative, you are allowed three days off.'

Apologies for the grim example, but it does represent the goal of phase two pretty well: give a better answer to any important question.

Phase three is where things get interesting. In phase three, we tend to start looking at integrations. Most companies use HR software internally. These software are a huge mine of valuable information which, if leveraged properly, can truly skyrocket the usefulness of the bot.

In phase three, employees can now ask 'how many holidays do I have left?' or 'who is MY manager?'. It isn't about general enquiries anymore, it is focused on this particular user.

Some resources either Dean or I wrote on HR chatbots, which you might find interesting:

What can an AI HR Chatbot do for you?

Solving the Top 3 HR Challenges with Chatbots

How to Build a Chatbot for HR Department Business Case

 

Pitching a customer-facing chatbot

When we talk about customer-facing bots, we could be talking about a customer service bot that answers questions about a restaurant's menu ('where does the meat come from?', 'do you have vegan options?', etc.). We could be talking about a table booking chatbot.

We could be talking about a purely marketing/branding chatbot where Willy the wine mascot (made that up, by the way!), takes you on a journey to find the perfect red wine to marry with the steak you are having tonight.

There are almost no limits to what we can do in this environment, so instead I will give you a few thoughts to keep in mind when pitching them.

 

Marketing: focus on a journey

Anything marketing related from a chatbot point of view will have to be heavily journey-oriented. Anything gimmicky or limited will quickly annoy users. It's not easy to make clients understand this, you will have to be the voice of reason here.

A good example is the 'Willy the wine chatbot' I gave above. This is great because you can create a backstory for Willy, put together a journey for the user, and give great value to them -- all at the same time.

 

Operational: straight to the point

On the other hand, anything operational such as table booking has to be straight to the point. I always advise not to do anything clever in these instances. If you are building a table booking bot, make sure it books table, answers a few questions about the business, maybe makes a couple of jokes, but that's it.

 

Best of both worlds

My personal favourite is a best of both world situation. If you take the phase approach I laid out with the HR bot above and apply it to a customer-facing chatbot, you can get to something really amazing quickly.

I would advise picking one of the two above and then gradually add more. The trick in this instance is to manage the user expectations and educate them on what can be done through the bot.

 

So, which one should you focus on?

Good question. There is no right or wrong answer here. If you are selling high-end, custom chatbots (like we are), you will have to listen to the client. They know what is wrong with their organisation, what they'd like to improve.

Listen and pick an approach from the two above that make the most sense to this particular business.