In the online universe there are millions of avatars representing users. Undoubtedly you’ve come across them before – maybe while chatting in a forum or playing an online game. Maybe you’ve created one in the past to represent you in whatever online activity you were engaged in.

What these millions of images tell us is that we’re capable, as humans, of interacting with graphical representations of other humans. Businesses have cottoned on to this and use a combination of natural language, artificial intelligence (AI) and avatars as customer service tools on their websites – they’re available to deal with queries 24hrs a day and help enterprises reduce operator and training costs.

The use of on-site avatars is driven by e-commerce and they typically provide information on products along with customer support. Evidence cited by Clifford Nass, a professor of communication at Stanford University, suggests that, to the best of its ability, an avatar should establish sense of rapport and familiarity with website visitors, and try and mirror them in terms of appearance, posture, voice and language – it appears that the more natural an avatar’s appearance, the better we are at interacting with it (for more on marketing personas click here). An avatar’s ability to do this is restricted by developments in AI and web technology, but that hasn’t stopped some businesses from using them on their websites. Let’s take a look at the good and bad sides of online avatars.

Network Rail Enquiries – Lisa

The company took the avatar process further than most and had a good think about what the online assistant could do for them while helping their customers. The conclusion was that Lisa had to provide their customers with the best and most timely information but also give the Network Rail marketing team as much information on the customers using Lisa as possible. They decided to build Lisa as an app within Facebook so users could ‘Like’ their favourite answers which would then be automatically posted to National Rail Enquiries’ Facebook wall. At the same time the marketing team received details on the users’ locations, interests, ages etc. – invaluable information they use for further targeted campaigns.

IKEA – Anna

An earlier attempt at customer AI by home furnishings behemoth IKEA wasn’t 100 per cent successful. While Anna was always there to answer questions, the AI wasn’t quite up to the task back in early 2011, which left customers frustrated with the brand, and more engaged with customer service than the checkout. Click here for a video summary of the problem.

The science behind the marketing suggests that an avatar or embodied virtual agent (EVA) that mirrors your typical customer creates trust. That is one of the major hurdles in any purchasing decision – although, as these examples show, they’re not for every business and require research, resource and dedication to get right.


If you work in a marketing department and are interested in learning more, why not register here for one of our free seminars on the science of digital marketing?

Want to talk about your next project?

Contact us