Key takeaways

  • Chatbots and virtual assistants are now being used by many customer-facing organisations.
  • Conversational experiences with chatbots will transform the relationship between brand and customer.
  • Organisations need to have a clear strategy so they can design experiences that provide value.


It’s not every day that you get to talk to a drunk George Washington.

Well, it wasn’t the real George Washington. I sensed as much when I asked how the Revolution was going and received the unpresidential response: “u wanna join my army or wut?” In fact, it was a chatbot created to promote an American TV series on the Kik messaging app.

Drunk George is one of the less impressive figures in a burgeoning community of intelligent agents – AI-powered virtual assistants and customer service bots embedded in smart devices like your mobile phone or smart home-enabled speakers.

Virtual assistants facilitate conversational experiences with customers, enabling them to ask questions, receive answers and complete tasks through natural dialogue. And they have already arrived: 42% of American consumers use digital assistants, as do 72% of business executives and 53% of millennials¹.

Conversational experiences are still in their infancy. Most disappoint customers because they are hard to use and have limited functionality. Many organisations don’t have the underlying capabilities to do conversational experiences well.

As with any new technology, organisations need to go back to basics and plan their strategy. It’s easy to get caught up in the bubble and fizz of artificial intelligence only to end up with a drunk George Washington.

Here are five rules organisations should follow to create valuable conversational experiences:

1. Define your chatbot’s purpose

Like any tech project, it’s necessary to identify the problem your chatbot will solve. Domino’s bot is called Dom. He has a clear purpose: helping hungry customers order pizza. He won’t help you refinance your loan or manage your calendar. If you ask him about the football results, he’ll just say, “I’m sorry, I didn’t get that”. That’s fine, because customers know that isn’t Dom’s job.

Organisations should educate customers on the role of their virtual helper. When someone interacts with a website or a catalogue, they can see what services are available. However, in conversation with a chatbot or virtual assistant, they don’t have the same context and can be left feeling unsure of what it does or how it can help. To avoid disappointing customers, invest in educating them around the purpose of your intelligent agents.

2. Pick the right model for your business

There isn’t just one way to implement a bot or virtual assistant. The right approach will depend on your organisation’s strengths. Forrester has identified three models to implementing intelligent agents².

  1. Make your data open to a third party that can access and deliver your product via a chatbot according to its algorithms. Think of an online marketplace that offers your goods(and those of others) via a searchbot.
  2. Integrate your products into a partner brands ecosystem. For example, hotel rooms that only show in the search of partnered travel websites.
  3. Go it alone and create your own bot. An example of this is financial services group Capital One’s ‘Eno’. Unlike many other banking chatbots, Eno is not available on Facebook Messenger or other third party platforms but instead responds to customers directly via text message.

Picking a model isn’t easy. Jumping onto a platform or into a partnership has the obvious benefits of leveraging scale and reducing investment costs. However, going it alone avoids the danger of becoming distanced from your customers. Are you comfortable with your products being just another option in a virtual assistant’s mind, completely at the mercy of unknown algorithms?

3. Bring together the creative and the rational in your organisation

No prizes for guessing that building a chatbot will need the smarts of developers and data analysts. But would you hire a comedy writer to give your bot a sense of humour? How about a behavioural scientist like the P2P insurer Lemonade did to nudge customers in the right direction? Or even a poet, as Microsoft did, to help Cortana wax lyrical? Leading companies are combining creative talent with data and analytics capabilities to develop intelligent agents that are simple, intuitive and human.

A role that is becoming increasingly important is the in-house ethicist. A leading virtual assistant on the market recently added a camera to its product so the company can provide personality-based fashion advice. This comes with moral issues related to collecting photos of barely-dressed consumers³.

An ethicist isn’t the first role that comes to mind when thinking of delivering cutting edge AI but it’s a good example of how organisations need to think holistically about the capabilities they need to deliver conversational experiences.

4. Align your chatbot’s personality with your brand

 As intelligent agents become more advanced, they will adopt realistic, human-like personalities. Have you heard the old line that customer service reps are the face of an organisation? In the future, chatbots will be the new brand ambassador. The technology may be new, but the same rules will apply.

The best chatbots will positively differentiate their brand through the way they speak, what they look like and the values they embody.

For example, one company has created a digital avatar that interacts with patients as a medical advisor. The avatar has access to the patient’s medical records and is responsible for sharing highly sensitive information. As you can imagine, the company has tried to create a sympathetic and caring personality, as demonstrated by the bot’s regular check-ins to ask the patient how they are feeling4

5. Give the customer power over their data

 Conversational experiences offer organisations the ability to continuously personalise and refine experiences by collecting customer insights. A key to successful conversational experiences will be how companies gather and use consumer data5. There is a fine line, however, between anticipating customer needs and crossing into the realm of ‘creepy’ privacy intrusions.

Many consumers would be more comfortable if they could control when and for how long the data they give to intelligent agents exists. For example, the data from a conversation with a chatbot being available to its owner for a limited period, and then erased at a time of the customer’s choosing¹.

Having the
right conversations

Conversational experiences may be one of the most momentous change to the relationship between brands and customers since the birth of the internet. However, organisations still have to get the basics right. The best conversational experiences are simple. The challenge for organisations trying to facilitate them is much more complex.

Now if you’ll excuse me, I have a chat with an intoxicated president to get back to.

In August 2017, PwC released Changing the game: The new rules of Customer Experience in the ‘Intelligent Experience Economy’, a report that talks about the rise of conversation in customer interactions. Click here to download your copy.



Sean Colvin

Sean leads PwC’s Customer Transformation offering.

More About Sean Colvin


Will Kingston

Will Kingston was a management consultant in PwC Australia’s Customer Strategy team.

More About Will Kingston