There has been a lot of hype around Artificial Intelligence (AI) in recent years, with organisations deploying various technologies within their process to deliver efficiencies and improved experiences. However, with so many businesses quick to take advantage of these solutions, some are now struggling to move the projects past proof of concept (POC) stage or scale across the business.
This blog post looks to outline some of the common challenges organisations face when implementing AI projects and how to overcome them.
AI and automation technologies have become a top investment priority for enterprises. This has meant that the need for AI leads the project objectives rather than the business case itself.
For example, a customer service team has been given a budget to investigate automate customer care. They have identified that a chatbot will enable them to do this, however, upon implementation have discovered that the technology is not returning the efficiencies they were hoping for. Consequently, the project fails to scale or move past POC. Yet when it comes addressing the reason for the failure, the technology is working as intended, rather it is the internal company processes that have created a barrier. On closer examination, the challenges that the customer service team faced are not an inability to handle call volumes, but rather the internal processes and systems that do not allow them to do it efficiently.
The problem within this scenario, is that the team have not understood what problem they are trying to solve. In order to ensure the success of any project, organisations and teams must truly understand and identify the business case. In doing so they can apply AI or automation technologies in the correct way to solve that specific challenge and deliver the desired efficiencies and savings.
There are a myriad of AI solutions in the market, which are all applied in different ways to solve specific business challenges. A misunderstanding the technology application can cause severe disappointments when expectations have not been set correctly, so selecting the most appropriate technology for your use case is key.
Within the previous example, perhaps a Conversational AI Assistant would have been more appropriate because of its ability to handle more naturalistic customer conversations, but more importantly because of integrations with line of business systems – enabling the platform to retrieve specific customer or business information from systems. This would have enabled the customer service team to speed up call handling through reducing the amount time and effort required to systems in order to find the information required. Delivering efficiencies through increasing the number of calls that could be handled, but also improved customer experiences, retention and reputation.
Before implementing any AI or Intelligent Automation project, spend time researching and speaking to vendors to really understand the technology applications and whether they are the right fit for the problem you are trying to solve.
Data is a key part of any AI project, particularly if there is some element of Machine Learning or analytics. The principle here is laudable: the collection of data to understand user behaviour is essential, as is clear understanding of what the technology success metrics are. However, the desire to keep these technologies in stasis during a trial period is perhaps a misunderstanding what role applications such as Conversational AI must play.
There are two reasons for this. For example, Conversational AI will not have been trained correctly, have the right information or the experience of dealing with real people. All of which mean that within the first few days, weeks, and months the Assistant will need considerable care and attention. Providing this level of support, is perhaps not what a team was expecting and therefore can immediately reduce adoption and usage, particularly if some technical knowledge is required.
Hence, iteration is crucial, and the success metrics need to be based on the rate of improvement as much as the overall accuracy of the service itself.
The final reason that projects fail is that, regardless of the success of a technology application, organisations must be honest about whether the solution they are building is better than the alternatives. This is not a question of whether one chatbot is better than the next chatbot, but a question of whether the technology solution is better than the customer service agent, better than email or the web site. Businesses need to ask themselves “if we are not building better experiences, then why should we expect the users to adopt the technology?”
The answers to this is not difficult, it is just a case of entering into an AI project with eyes open to some of the challenges that lie ahead, and being honest about the drivers and benefits of the project.
If you would like to find out more how to get started with Conversational AI get in touch today.