sense of direction vector id165669747 copy

Businesses are going through a transformation of adopting Conversational AI at pace. The speed of adoption has never been quicker given the current economic climate, the impact of COVID-19 and the appetite of consumers for instant self-service. This has had both a positive and negative impact on the market.

On the one hand, Conversational AI is viewed as the holy grail for its potential to support the ability to scale. On the other hand, such rapid adoption can cause solutions to be deployed without an underlying business case or value proposition. By moving too quickly without clear goals, businesses are at risk of falling short of expectations, being unable to prove the benefit of the solution and then having to decommission the service.

To bring the importance of this topic to life, here is a real-life example of what can happen when an organisation ignores the business value topic. We once reviewed a solution built by a large US health insurance provider. It was a pilot Web Chat use case for their customers’ employees. The only direction provided to the team at the outset of the project was to ‘prove the technology worked in our ecosystem’. That’s it. There was no consideration of business benefit, value or user experience. The development team were extremely distracted with technical purism and methods that had no impact whatsoever on the user experience or (undefined) business goals. In fact, the user experience and business benefit didn’t even make the bottom of the priority list. After 6 months and $1.2m of investment in the pilot project it launched to a selection of customers. 2 weeks after launch it was shut down and the project killed.

If you cannot articulate the business case, save yourself the heartache – don’t start a Conversational AI project. But if you can define the business case, use it as your compass to steer you away from the shores of disillusionment.

The articulation of business value has many dimensions but is critically important in all phases of a Conversational AI initiative. You need to consider:

1. Initiation: Stakeholder commitment

It is essential for executive stakeholders to provide organisational commitment, investment and definition of the value a program should deliver. This need not be ‘all-in’ from the outset. Perhaps the project is delivered in phases with proof-points required along the way to release more investment. Clear business objectives and programme commitment is important for steering every phase of the programme towards success.

2. Project approach & scoping

The potential landscape of a conversational digital transformation programme is gigantic. You have the channels (e.g. Webchat, IVR, etc) and knowledge coverage to consider, as well as the depth to which you automate a response to the users intents. Typically, a programme will be phased to cover the higher volume areas to maximise the return early on. But how do you prioritise the features and knowledge coverage into phases? You do this by triangulating (1) analysis of existing data, (2) user experience and (3) budget, all in the context of your business objectives.

By knowing the value, and subsequently the features that are necessary to deliver the value, only then can you optimally shape the approach and scope of an implementation.

3. Design

It is critical to ensure the design is focused on meeting clear business objectives. Otherwise, you may be distracted by cool but perhaps unnecessary things. By design, we do not just mean the Conversational brain of the solution, but the entire customer experience from the introduction of the experience, the look and feel, through to the Conversational experience and conclusion of the interaction. Whether it be colours, or words, or the use of integrated data, there are many aspects that will impact the ability to meet the business objectives. Importantly, the design should also be considered from a Conversational viewpoint based on primary user intents, not historical lists of keyword data or convoluted internal processes. All areas and each design decision should ensure the optimal experience and maximum return – even if it means changing business processes to deliver the most awesome experience in line with the business objectives.

4. Build

There are many Conversational techniques to consider when building the ideal experience. A Conversational AI platform has many out of the box or custom developed opportunities to ensure the dialogue is accurately tuned to meet the needs of the user. This may be the use of memory, context, multi-intent recognition, sentence segmentation, lexical resources, machine learning algorithms, natural language generation, integrated data, sentiment analysis, etc, – all the things humans take for granted when communicating. However, if you try to use every trick in the book you may find the solution complex and not scalable. Careful consideration is required to use the right techniques in the right context for the most efficient experience towards the business goal. This really requires the experience of a Conversational AI developer who is focusing on the business objectives.

5. Benchmarking & ongoing improvement

There is a misconception that Machine Learning is a silver bullet. That a flick of a switch will enable the cogs to work independently and any initial investment will be dwarfed by the opportunity once the machine is fully exposed to customer intents and expectations. For now that’s not a world that exists and we’ve had lessons along the way to prove this (Microsoft’s ill-fated ‘Tay’ for example). The point is, Conversational AI is an investment, not for a single static solution, but a continuous improvement that is constantly evolving to the needs of users. It is important that this ongoing investment in benchmarking and improvement is understood and planned early on.

Each interaction has a value: it could be the monetary value generated through sales or the total cost of ownership to provide a service. There will also be other metrics at play to determine the success of a solution, for example customer satisfaction, containment, knowledge coverage, or engagement rate. All appropriate metrics should be included in a model to provide an expected ROI in line with the business case. Regularly assess performance against the benchmark and targets. Prioritise developments or fixes to those areas that will have the most impact and you will see a solution improve over time, every deployment adding more value.



The use of Conversational AI is likely a new interaction paradigm for your business. Your path forward will be an iterative process based on real user interaction and metric analysis. You will learn much on the way. Use the business case as your true north, a shining beacon, to guide you towards business success and to help keep the team focussed on the right goals when venturing into the unknown.