The use of conversational commerce also has a number of business and organizational consequences.
The brand experience within the conversation channel largely reduces text perception. So word use and conversation techniques are decisive in this. This requires other disciplines, such as conversation design and conversation protocols.
Another aspect is that the product range within the conversation can only be very limited. On the basis of the dialogue, the consumer expects a targeted and appropriate offer of probably a maximum of 3 to 5 alternatives. This requires personalization and a strong product suggestion functionality.
In addition, the need for data and knowledge from various sources and business functions will also require cross-departmental collaborations and agreements. Companies must allocate resources and develop competences in this area.
Online chat with real employees of companies has a lot of added value and is considered valuable by consumers (see figure 2). For example, BCC (Dutch electronics retailer) helps the online customer via a chat service with experienced product specialists from the store. The chats are mostly about the individual needs of the customer with regard to a product and added service possibilities. Thanks to the specialist product knowledge, customers are immediately assisted in the online funnel and this contributes positively to the customer satisfaction and the conversion of the online customer. This requires a tight organization and planning of these employees.
While further automation with the help of chat bots takes place, the importance of data, artificial intelligence (AI) and emotion analysis grows. The need for data scientists, psychologists and AI developers will increase further. And last but not least, trust and investment space in chat bots and other artificial intelligence technology is required. This investment concerns not only the cash-out for the purchase of technology, but also the investment of time to make the chat bot intelligent for specific companies.
To develop a good understanding of the IT architecture behind the chat function, it is important to distinguish a number of issues in general terms. These are shown in Figure 10.
- The outer layer is that of the conversation user interface such as Whatsapp, WeChat, KIK or a voice system like Siri with integrated speech recognition;
- The chat bot is an automated chat solution that is often active within such a platform;
- Through so-called APIs, these chat bots integrate with artificially intelligent virtual assistants, who, on the basis of Natural Language Processing (NLP, language comprehension), can converse with consumers and 'transfer' to underlying services;
- These services (web services in the cloud) provide the bot with relevant data and information to serve the consumer; these services can be offered by own systems or third parties, but can also be connected to the Internet.
A resident of Amsterdam asks Alexa what the weather forecast is. Alexa understands this question and understands that it revolves around the weather forecast for Amsterdam. Alexa then uses an API to connect to Buienradar to retrieve the forecast for Amsterdam, and returns this in a spoken dialogue "The next hour it will remain dry in Amsterdam, after which rain is expected. The temperature is 15 degrees Celsius and there is little wind." In this example, the temperature could also be based on an Internet-connected thermometer that is present in the vicinity of the consumer.
The chat bot is organized on the basis of 3 important aspects:
- Intentions: what is the purpose of the user? For example: making a purchase;
- Entities: the issues that matter. For example: products, stores, etc.;
- Dialogue: how does a conversation work? For example: greeting followed by an opening question.
The virtual assistant makes the chat bot smarter. This is done in the following ways:
Figure 11 shows how important it is to continuously make the chat bot smarter by analyzing conversations and processing the feedback in the system. This takes time! Professional software solutions support this learning process with analyses on content, but also on suspected emotions (recognized from language use, emotions and from voice pitch and voice volume). Finally, a good integration with other systems ensures that the chat bot can increase considerably in knowledge.