Large companies have set their investment eyes and dollars toward artificial intelligence as the next digital opportunity. However, to date, only one in five companies has incorporated Artificial Intelligence in its' offerings or processes. And while companies with at least 100,000 employees are the most likely to have an AI strategy, a recent study from MIT Sloan suggests that only half of them actually do. In this blog, we will uncover today's market leading chatbot platform providers and provide you criteria that will allow you to approach the introduction of chatbots in your business like a pro.
Today's Market Leading Chatbot Platforms and their use cases
There are no doubts that Facebook is the dominant player in messaging services, therefore it has invested heavily in chatbots and in its own chatbot platform. In fact, Facebook is trying to attract companies to use its bot technology by offering new services through its platform, such as the ability to order an Uber... or a pizza. Currently Facebook is the worlds largest messaging platform with 1.2 billion people and over 300,000 monthly active bots. It's platform may not fit all Enterprise needs, but in the context of chatbots - it’s worth to mention.
The IBM chatbot platform (called IBM Watson) is the first choice as a chatbot platform for 61% of enterprises according to a 2017 market survey conducted by MindBowser. One of the Watson’s most important services is the Conversation Service which is built on a neural network and capable of understanding intent, interpreting entities and dialogues. This is a nearly unique feature when compared to other chatbot platforms, where developers must manually define intents and entities for the chatbot to understand the conversation (meaning, plenty of manual training is required for the chatbot to be effectively functional).
Microsoft has also developed and launched its chatbot platform called Bot Framework in early 2016. Its chatbot platform is natively integrated with Cortana, where developers can produce their own chatbots and it has support for Azure to scale out-of-the-box. Currently, Microsoft reports over 200,000 developers registered with their chatbot platform and this number is rapidly growing.
Even if they started only in 2016, Google started big in the chatbot platform world. Google bought the existing platform API.AI (now called Dialogflow), a tool for developing chatbots which incorporate Google’s machine learning products. Like IBM Watson, Dialogflow can leverage a neural network and it can understand conversation and get questions answered. Thanks to its massive user base on their products such as Gmail and GSuite, we see Google soon incorporating bots within its communications tools.
Amazon released its conversational interface tools, Amazon Lex in 2017. Using the same technology behind Alexa, users can easily build text or voice bots on Amazon Web Services. It’s interesting to note that Amazon is not pushing to develop an equivalent for example of Facebook’s “M” for Facebook Messenger. It looks like that Amazon believes more in Echo and voice interfaces rather than chat interfaces. That’s easy to understand, because with over 10 million Echo devices sold, they are trying to keep up their momentum and focus on the area where they could have higher returns.
Finally, even SAP has entered the chatbot platform world. In 2018, SAP acquired the former Recast.AI. SAP is offering their customers a platform where developers can develop their own chatbots and, most interesting for SAP customers, leverage pre-configured integrations with SAP and non-SAP technologies to improve customer experience. Since this is a rather recent acquisition, we will see in the next months what the SAP strategy will be and the directions they will give to their chatbot platform.
Considerations for Choosing a Chatbot Platform for your Enterprise
Despite a broad and valid range of +20 chatbot platforms on the market to date, there exists no universal chatbot platform to satisfy all AI needs of all enterprises. Therefore, I recommend that when you are selecting a platform for your needs, you evaluate the platform based on the following criteria: Integration, Security, Pricing and Metrics.
Most of today's chatbot platforms are focused on enabling easy use cases, such as customer support, lead generation, and FAQ bots. However, the chatbot platforms we just mentioned have been designed to satisfy the requirements of the large and medium enterprises, as well.
At Techedge we are leveraging independently Dialogflow and Watson, each of these platform has difference advantages when enabling integration to backend systems using RESTful APIs and web services, so that chatbots can be quickly architected, developed and used to work in conjunction with core Enterprise systems like SAP, ServiceNOW, CRM, RPA, etc..
Leveraging the integration of chatbots platform with internal systems gives an option to enterprises to explore more interesting use cases and obtain equal cost reductions and operational efficiencies on a larger scale.
The pricing is always a decision maker when evaluating a new enterprise AI chatbot platform. Generally, Enterprise Chatbot platforms use one of the following pricing models:
Subscription: Build limited or unlimited bots for a flat monthly or annual fee. This is probably the most suitable model for Enterprise.
Pay per usage: Chatbots platform cost is based on chatbot usage and number of API calls. This model can give higher flexibility to pay-as-you-go and increase gradually.
Pay based on performance: Chatbot platform cost is based on the performance of the chatbot that enterprise can establish and agree upon the platform vendor. For example, successful interaction between the bot and the client is considered as a paid conversation, regardless the length of the chat. A good optimization of the chatbot conversations will attain certain performance metrics. With this model, enterprises will be charged accordingly to the goals achieved.
While there are pro and cons to each of these pricing models, we think that an investment in any chatbot platform should be combined with a strategic partnership. Techedge helps enterprises to define goals, project potential ROI, tie your business objectives to cost of implementation and management. The costs of creating chatbots will change widely depending on the scope of functionality, the technology used, integration and data storage requirements, etc. We believe that only matching chatbot functionality to the exact business requirements will make possible significant ROI from chatbot implementation.
Chatbots need to provide a level of security and assurance with regards to the information received, exchanged, and delivered via chatbot. This includes appropriate rules that take into consideration how the information is exchanged, as well as how it’s saved during and following transport.
Basic requirements should be that any communication via chatbot must be encrypted end-to-end and the ability to interact with the chatbot must be regulated with a robust, role-based security framework.
Other requirements especially for the EU and UK market, with regulations like GDPR, is that data privacy should be a key design aspect to ensure data privacy and security. Basically, the recommendations here are to ensure that by design, the chatbot platform and the chatbot itself does not store any customer sensitive data and to provide full transparency to the enterprise through an auditable interface of all the data points stored and used for analysis.
After the launch of a new chatbot, it’s fundamental to collect metrics to improve its efficiency. The metrics depend on the type of business and the nature of the chatbot, but some metrics to consider tracking for all type of chatbots include:
1. In-message: the number of messages sent by the user. Understanding the number of interactions there are between the user and the bot is fundamental to understand the level of engagement on one side. On the other side, can hide issues with the bot not understanding the conversation and the user is forced to repeat its questions.
2. Retention rate: the percentage of users returning on the chatbot in a given period. This metric can help you understand the level of engagement and can give you more insight regarding customer’s preferences.
3. Goal Completion Rate: the percentage of successful engagements through chatbot. (for example: how many people scheduled a ride after contacting a Uber-like chatbot)
4. User Satisfaction: Whether or not the user would recommend the chatbot to a friend or colleague (example: the customer will rate the chatbot experience responding through exit survey)
Metrics are the key aspect to consider when you are thinking to adopt a pricing model based on chatbot performance.