Modern life demands speed, right? We eat, speak and walk faster than we used to. We are disappointed if our favourite online store does not offer next day delivery (even if, pragmatically, we could very well wait). And we want 24/7 customer service so that we can get instant replies to our queries.
However, on some occasions, offering slower service can actually improve customer satisfaction. For instance, depending on the purpose of the activity, such as shopping vs searching for information. Or, when the task seems complex and accuracy is an important attribute for customers.
But what about interactions with chatbots?
Businesses have long adopted some form of automation to handle common customer queries, for instance, via FAQ pages on websites, or interactive voice recognition (IVR) options on inbound phone calls. Conversational chatbots are an obvious evolution in this area. Natural language processing capabilities mean that the chatbot can interact with users in a manner than feels more natural than IV, and is more interactive than FAQ pages. Hence, many organisations are experimenting with using chatbots to handle common queries (e.g., store opening hours), or to facilitate certain transactions (e.g., booking an appointment).
So, it is pertinent to ask: Would consumers value speed of response when interacting with chatbots? Or is this one of those situations where slower is actually better, as far as customers are concerned?
According to research carried out by Ulrich Gnewuch, Stefan Morana, Marc T. P. Adam, and Alexander Maedche, faster is not necessarily better, as far as chatbots are concerned. In the paper “Faster Is Not Always Better: Understanding the Effect of Dynamic Response Delays in Human-Chatbot Interaction”, available here, shows that introducing a delay in the response actually increases customer satisfaction.
In their study, Gnewuch and colleagues introduced a delay that reflected the complexity of the customer’s query and of the chatbot’s response, based on sentence length and number of syllables. They found that, when compared with a near-instant response, participants in the study found the delayed responses produced higher overall user satisfaction with the chatbot interaction. The research participants in the control group (i.e., which received near-instant replies), reported feeling that the quick response speeds were confusing.
The researchers also found that the research participants perceived the chatbot in the delayed scenario to be more human like and to have a higher social presence than the near-instant scenario. Perceived humanness and social presence have been suggested as important factors in driving user acceptance of AI solutions, as I wrote here.
It’s an interesting example of the need to temper technology capabilities to meet customer expectations.
Are you happy to interact with AI-powered chatbots, in customer service?
Ana Canhoto 