The boardroom guide: Creating a successful conversation AI strategy for your business
If you’re reading this, the chances are you already know what conversation AI is and that your business can make huge gains from its use. You may have departmental managers trying to persuade you that chatbots and virtual assistants will automate support, reduce response times, or improve customer satisfaction for your business, and they’re right.
What you may not know, is where to start to pull together your own conversation AI strategy or the key things to consider. So here are the steps we’d take to implement such a strategy.
Step 1. Define your business goals for conversation AI
It largely splits into two main camps for most organizations – improvement of:
- Automation goals – making communication with audiences faster, more integrated, and more efficient
- Customer experience goals – delighting customers and building brand ambassadors
Conversation AI can unify both – automating conversations and processes whilst also improving customer experience.
The more conversations we have, the better they should be, so it’s important to balance your ‘scope’ with your customer experience. That is, balance your investment into technology, integrations, and customer journeys with the need to deliver a solution that results in widespread adoption and improves customer experience. Failure to get the balance right means you’re either investing in novelty experiences or technology that doesn’t improve customer experience and won’t be adopted.
Simply put: scope + experience = value
Scope is developed through use case selection and investment in technology. Design standards drive customer experience – design for function, trust, behavior, and delight. As the use cases become more advanced, the design will also be more advanced. For example, a simple answer to a simple question might suffice - you can just design for basic functionality. However, as journeys become multi-turn and personalized, it becomes important to design for trust through consistent tone of voice, personality content, and variety.
Step 2. Who is your audience and what is your use case for conversation AI?
It might seem obvious, but the inability to understand your audience whether they’re customers, patients, employees, or more widely, users, will witness the failure of your strategy.
- Who are they? Create detailed audience personas.
- What conversations are you having with them? What conversations do you want to have with them?
- Are they internal or external to the organization?
- How do they communicate with you – via what channels? Mainly through your website, email, voice, or via third-party apps like WhatsApp or Facebook Messenger for example?
- Are you engaging with them in a relatable way to them – through the correct channels, in the right tone of voice within an expected timeframe?
- What is your use case – to increase productivity and revenues, reduce costs, or grow reputation? And what does that look like practically - sales, support, or general visit chatbots? Be clear about where you can make the biggest gains for conversation AI.
Audience personas should be at the heart of your approach – work out what people want, what motivates them, and how they feel. It will shape the scope, design, and execution of your conversation AI strategy.
Step 3. Build or buy the platform?
Do you design and build the platform you need inhouse over time, or do you buy it to get to market faster? It depends upon your organization’s maturity, skills, budget, time, and requirement for speed to market. Many businesses invest millions, build everything, take lots of time and make mistakes – it’s a journey. Others benefit from years of experience via partners, allowing them to focus on value creation.
- Building it means you will manage all aspects of chatbot creation, including the use cases, managing IT infrastructure, integrations, building NLP models, performance, training, measurement analytics, conversation design, tone of voice, channels, and brand.
- Buying it allows you to work with partners to help with best practices, channels, tone of voice and design and speeding up time to market while providing time to experiment in parallel. Enterprise solutions can also deliver NLP performance, integrations, security, and hosting ready to roll out at scale.
Step 4. Design for customer experience
Based upon the scope of the AI assistant, there are five altitudes of development that all come with their own design patterns. Based on your scope, you should follow the right design patterns accordingly. For example, when you have a simple chatbot, functional design is sufficient. However, if your AI assistant handles personalized and transactional journeys, then design patterns that contribute to trust become more important as well.
- StepPotential - founding value lock-in. Being inclusive, human-centric, empathetic, and goal driven.
- Function – mainly FAQ bots with basic integration. Design patterns include acknowledgements, confirmations, clear prompts, discourse markers, active language, never a dead-end-street, and others. This helps people get basic value from the assistant.
- Trust - integrating your bot through multiple channels. Detailed persona, consistency, personality, and onboarding content, explainability, and accountability. This helps people feel more comfortable around the assistant.
- Behavior – proactive contextual and omnichannel capabilities. Reciprocity, commitment, social proof, authority, liking and scarcity. These psychological design patterns help the assistant truly influence people’s behavior, whether it’s helping a customer trying to buy, managing emotions in customer support, or guiding patients in behavioral therapy.
- Delight - the bot that reaches the delight stage can be the first point of contact and can help with almost everything. It has visual and sonic design, custom voices, virtual beings and is a seamless and consolidated experience. People will reach out to this assistant first because they know it will get the job done. This is the holy grail of conversational AI.
Step 5. Maintain the balance within the integral strategy framework
No matter how you reach your goals, there will always be four elements that are integral to success. They are the key ingredients in the conversation AI cake – remove one, the recipe falls apart and the cake is inedible.
Everything you implement should nurture and develop these four elements, so they mature over time and deliver conversation AI excellence.
- Mindset – how individuals on your team perceive conversation AI work in your organization. Do you have the right personalities on board to deliver tone of voice, editorial strategy, or train the language models for each level of maturity of the project?
- Skillset – the ability to execute the work. Do you have the depth of roles and skill sets to deliver the plan? Each altitude requires a more advanced skill set.
- Culture – the context in which people work. Does your organization have the culture and strategy to support conversation AI – do you have conversational AI ambassadors?
- Systems – the tools and structure with which people do the work. You can have an amazing team, but do you have the right technology?
As your AI strategy matures, these four integral elements mature too. For example, a modest inhouse team that tests a basic proof of concept in a small organization, experiments more - it doesn’t have the technology or skills found in an enterprise organization. Likewise, a developed team in an enterprise can’t readily break things to try new ideas in the same way. But be aware, maturity in each of the four elements takes time and can frustrate culture.
Step 6. Be patient
Rome wasn’t built in a day, so don’t over, (or under!) invest in one of the four elements in the hope you’ll get there faster.
Building conversation AI in your organization is like human life stages - you can’t take a baby and make it a fully functioning adult overnight. You simply have to go through each stage, teaching and refining your parenting skills as you go so that you deliver the best version of a human being.
The same rule applies to conversation AI - ultimately you want the best AI assistant that you can create and you have to go through all the phases in step 4 whilst developing the conversational mindset, skillset, culture, and systems in step 5. The four quadrants have to be developed in harmony - or there will be an imbalance in the scope and customer experience chart (step 1) - and you’ll struggle to deliver value.
You don't want to fall into the ‘squandering’ quadrant in table 1 and be the organization that doubles down on technology, but underestimates how long it takes for culture and skillset to develop - so keep all four elements on your radar, and be patient.
Click here for more details on the conversation design courses CDI provide