5 steps to avoid pitfalls in Conversational AI

New technology can rapidly create business and customer value. Still, a fair share of projects tend to fail. In this article we recommend five key components to include in your Conversational AI roadmap that increases the likelihood of success.


Conversational AI is an overarching term for technology that enables humans to interact with machines and systems through natural language and dialog. 

The area is expanding rapidly and is projected to reach a value of 40 billion USD by 2030 with a CAGR of 24% per year. 

The interface of this technology is usually a digital assistant (an advanced, AI-powered chatbot or voicebot) that can perform tasks – either standalone or as a hybrid solution where the digital assistant handles parts of a process and cooperates with human agents for instance in customer service for more complex tasks. 

Value-adding use cases range from creating faster, more efficient and scalable customer service, to generating increased revenue by providing for example personalized conversational product guides and recommendations. 

The need for a strategic approach and roadmap

The technology opens up a wide range of possibilities. But there are also pitfalls to avoid. Some of these are not focusing enough on design aspects in the development process or moving forward with too much functionality or too broad use cases from the start. Another pitfall could be to approach these development projects as you would for instance any web design and development project.

In order to achieve scalability quickly and to realize the potential customer and business value that Conversational AI can generate, the following is crucial: 

  1. A strategic approach. Risks involved with conducting only short term planning include creating bottlenecks when attempting to scale and lack of coordination between initiatives will create non-consistent experiences that can lead to brand erosion. 
  2. Agile and incremental development. The landscape is developing rapidly and the technology is new to large parts of organizations which will inevitably lead to new requirements developing over time. In addition, data collected from live users will influence upcoming steps of development and improvement.

In order to succeed, an agile roadmap should be developed. This roadmap should have its starting point in your business’ current state of maturity and should clearly describe prioritized steps towards fast realization of business value. 

The road map needs to address the following main components:  

Each component is described more in detail below. 

  1. Where are we heading? 

Description of current state,  maturity and Desired State.

The necessary transformation and development can be facilitated by formulating an attractive vision for automated customer touchpoints and Conversational AI activities – a “Desired State”,  including goals and principles for how you want your customer interactions to work within 3-5 years, how the organization should drive development within the Conversational AI area, how value will be created and how the journey towards the goal will be managed.

  1. Use case prioritization. 

Map and prioritize use cases and applications. 

What does your current customer journeys look like? What pain points and situations can be addressed through automation and Conversational AI? Where can you create efficiencies?

Conducting workshops and interviews with cross-functional stakeholders will generate a long list of potential improvements and ideas. Based on this list, and best practices and trends in the industry, use cases should be prioritized based on criteria such as customer value, business value, strategic relevance and feasibility in order to form the basis for your roadmap.  

  1. Controlling the experience. 

A plan for developing the customer experience in automated conversational touchpoints.

Perhaps there are already several parallel ongoing initiatives within your organization related to automation of customer interfaces. It could be a product team that is developing a conversational interface or looking into voice interaction or the customer service team developing a solution for automating incoming calls through voicebot technology – or perhaps Sales and Marketing are developing solutions within Conversational Commerce.

As the automation of customer interfaces gradually increases, it becomes more crucial to develop a plan and guidelines for this new customer interface. These guidelines need to be aligned with your existing brand principles.  What is our ‘tone of conversation’? ‘How do we sound?’, ‘What artificial voices do we use?’, ‘Should we be represented by an avatar or digital human and what does that imply?’, ‘What type of personality does our brand represent?’, ‘What ethical considerations do we need to address’?

By addressing these types of questions early on, the business can avoid a fragmented customer experience and unnecessary double work later on, as the need for alignment becomes a critical business issue.

  1. Preferred tech stack. 

An understanding of the ecosystem and platforms available, and the requirements going forward. 

Just like in other growing areas of digitalisation and automation both end-to-end platforms as well as their components are developing at a very rapid pace. 

Based on your Desired State, your approach and your core use cases you can make well informed choices around tech and platforms. These choices will be crucial in order to scale and utilize the technology across channels, devices and modalities (web, chat, telephony, social media, smart home etc).  

  1. Competence needed. 

A plan for developing internal capabilities, knowledge, roles and responsibilities. 

In order to generate maximum ROI and the best customer experience, Conversational AI solutions need to be trained, optimized and developed continuously. This requires access to relevant data and knowledge on how to conduct this ongoing development. This, in turn, requires new capabilities, roles and competencies – such as Conversational Designers and AI-trainers. 

These types of roles and teams can gradually be built in-house. One successful way of building the inhouse knowledge in our experience is to establish a cross-functional Centre of Excellence for Conversational AI where knowledge and learning can be shared on an ongoing basis.  


By addressing these five core areas and developing a strategic and agile roadmap, the efficiency of implementation will increase radically and the road to improved customer experience and ROI will be significantly shortened. 


About the author

Mattias Falkendal is the CEO and founder of Talking to me, a specialized professional services firm focused on providing business and customer value through Conversational AI. Mattias provides senior advice to leading enterprises on the strategic implications of adapting new technology and runs the implementation of several Conversational AI programs for leading brands in the Nordics.