When do chatbots become mainstream? With Mark Billingham
Episode 2, Aug 11, 2021, 11:54 AM
Terminator. Blade Runner. Ex Machina. iRobot. While Emma’s chat today with Mark Billingham of the Very Group isn’t quite as dramatic as some of those blockbusters, the topic of AI has been around for a long time.
Mark gives us some amazing insight into his learnings, practical tips and overwhelmingly positive results of building a bot to handle a huge amount of customer queries each month - 250,000 to be exact.
They dive into what capability you need sitting behind the bot, how they got customers involved using their UX Lab (which is as cool as it sounds), the opportunities it creates for your internal team and why the job is never done even after you have built your bot.
If you are looking to build an AI chatbot in your organisation, here is a 45-minute practical guide from someone who has done just that. So please enjoy the show and don’t forget to subscribe to the podcast.
Intro - 1.20
- 250,000 contacts handled per month at The Very Group
- AI frees up advisers for more complex and emotive questions for customers
- Contacts have reduced by 23% and nearly 50% over two years
- The bot is getting ⅘ stars for ease of use from customers
The catalyst for building AI functionality - 2.37
- Previously navigated trying out a bot in Vodafone called Toby
- Learnt how NOT to do it
- Need to understand every single reason a customer contacts you
- There were 10,000 different reasons they got in touch
- Acts as a concierge service and fix it, otherwise will send to a human
- Training a bot is like training a person - takes time
What was the goal? - 7.29
- Don’t upset our customers
- Put something on the web that got users to the right place and answers quickly
- Ideally, people can self-serve which benefits both us and customers
- Start with ease and simplicity
The Capability that sat behind the AI - 9.43
- Used our own web development team to create a look and feel
- Then use tech dev team to build but use IBM Watson technology in the background
- But all knowledge comes from the customer care team
Customer Involvement - 11.24
- We have a UX lab where we get customers to come in and try out the bot
- Kept it going during the pandemic virtually, but not the same as the face to face
- Feedback can often be based on how comfortable they are with technology
- ‘I wouldn’t have used this bot before, but now I absolutely would’
- 40% of customers use our bot, so plenty of work to do
UX Lab Investment - 14.14
- Need two rooms - can be really basic
- Just need customers using the tech and giving you feedback
- Doesn’t have to be Big Brother /MI5
Intentional Actions Delivered During Change Project - 17.07
- Identify key reasons why they contact us
- Understand what we need to build
- Build the most important questions first
- Make sure its useful straight away means the customers are on board
- Where does the bot need guidance?
- The balance between it being too ‘botty’ but also not too human and missing the tone
What are the next steps to perfection? - 20.51
- Target is to go from 40% to 70% using the bot
- Need to expose the bot to more people - only available for people who are logged in
- The next step will be using them on the outside of the site
- SEO needs to work for us to point to the bot
- Teach it to answer more queries
- 25% have to ring contact centre after - so need to focus on them, get it below 10%
Biggest Challenges - 25.19
- Hygiene - we are a digital business so web changes
- Finding the right resource and prioritisation
- Making sure the language is absolutely spot on
- How far through the maturity curve is AI?
- Have to make sure using the bot is optional for users
Managing the internal feeling that the robots are taking our jobs? - 29.11
- The organisation has been continuously evolving for the last 90 years
- We have a high tolerance for change
- Give lots of notice and move onto other campaigns
- Re-investing savings made and into your people
Biggest benefits for your employees - 32.21
- The new set of skills and roles that have been developed internally
- Continuous improvement, change management culture
- Created specialism that people can grow entire careers from
Where is the future headed? - 36.50
- Pandemic has swung it back to the preference for human touch
- Bots will become more mainstream and better
- You will have a divide of people based on consumer demographics
- My 8-year-old son uses Alexa for everything
- Bots could be superseded by voice assistants
- Video is up for debate. It’s expensive and Zoom fatigue hasn’t helped
Customer Feedback - 42.24
- Yes, positive reaction and easy to navigate
- Use NPS and TrustPilot feedback
- Negative is when there’s a handover between channels
Advice before an organisation kicks off a bot build? - 43.19
- Understand why customers contact you.
- Don’t shoe-horn assumptions into the bot
- Don’t lie to people that its a human
- No dead ends
- It’s a consistent journey and needs continual management