If you’re looking to make customer experience (CX) a core part of your offering, there’s no way your strategy can leave out conversational AI. And why would it? Customers expect fast answers around the clock and a whopping 93% will make repeat purchases based on excellent service. Integrating a virtual agent into your CX can help you meet and even exceed those expectations, translating to higher retention and revenue.
The daunting question for many companies is: how do we get there? How do we transform fragmented processes into deep CX automation and highly personalized service delivered by bot and human agents working seamlessly together?
There’s no one-size-fits-all solution. And right now the number of options on the market is overwhelming. If you’re deep in the bot platform rabbit hole and feeling more confused than when you started, keep reading. I’ve compiled a list of the seven most important questions you should ask yourself when evaluating solutions.
Spoiler alert: Flexible is always best
Rather than sending you deeper into the rabbit hole, I’ll share the takeaway upfront. You need a solution that evolves with you. Thinking long-term about this investment (we’re talking years, not months) could save you loads of time, money and effort. That means choosing a flexible platform that allows you to pivot your approach as needed so you can keep driving revenue and giving customers the best experience possible.
With that big-picture consideration in mind, let’s break down everything you need to think about, end to end. There’s a lot to consider, but you’ll be glad you did—especially once you start seeing the benefits of the right bot platform.
1. Will the CX capabilities meet my business needs?
The use cases for CX automation are growing every day. Some brands use conversational AI as a first-stop virtual agent that resolves simple customer issues so human agents don’t have to. Others use it to deliver holistic CX that actually develops stronger relationships with customers.
What role will automation play in your CX today and years down the road? A free or cheap bot you can set up in a few minutes might be able to handle simple activities, like answering FAQs. But if you want to automate specialized tasks, provide deeper experiences and support, and take pressure off human agents, you’ll need a more robust platform.
Here are some things to consider when selecting a platform that meets your CX goals:
- Conversation focus: Whether you want to resolve customer issues, market a product or generate leads, determine what you want to achieve with conversational AI and whether the platform can deliver on that goal. Is the platform focused on marketing, sales or support? Does it let you do all of that and more?
- Horizontal or vertical application: You’ll also want to know if the AI is horizontal or vertical. Vertical AI is optimized to address specific use cases and tasks. It uses data from a particular industry or context to train its algorithms. Horizontal AI is more general. Designed to be good at a lot of things and able to solve a wide range of problems across contexts (e.g., HR, sales, IT, etc.), horizontal applications can sometimes falter when it comes to completing more specialized tasks.
- Workflow automation: Over the last few years workflow automation has infiltrated every corner of business, including CX, as companies realize the dramatic productivity boost it enables. Your bot should be able to take on the burden of time-consuming tasks and help you serve customers better by collaborating with employees. You’ll want to ensure human agents can trigger the bot to perform certain tasks and vice versa.
- Personalization: In our data-driven world, customers expect a high degree of personalization in their experience. According to Accenture, 91% are more likely to shop with brands who provide relevant offers and recommendations. Given those expectations, you’ll want to explore what level of personalization the platform can offer. Having access to authenticated users, for example, allows you to deliver highly personalized service. When customers are logged in, you can use all of the data you have stored about them, from what they last ordered to any previous issues or complaints, to tailor their experience.
2. Does the dev process fit my internal capabilities?
One of the biggest missteps businesses make in building bots is misjudging the amount of internal effort it can take to do it yourself. Some companies sink lots of resources into building a high-maintenance bot that doesn’t deliver. What they really needed all along was the help of an experienced provider with a flexible platform.
With that in mind, consider what the experience of building the bot will be like, including these factors:
- Customization: Is the platform fully programmable and is it possible to run custom code? Make sure you can build a complete solution that can be extended in every aspect, including custom triggers and elements to accommodate more complex steps in your workflow.
- Level of coding required: What level of coding suits your internal capabilities: no code, low code or deep code? And what does the platform you’re considering require? Can anyone use the app? Ideally the platform is customized to your comfort level and flexible enough for stakeholders to build and edit the app. Check to see if it supports a visual editor, which is more accessible for non-technical users.
- Programming language: What’s the programming language? Will it be easy for your team to learn? What resources are available to get them up to speed?
- Support: What support is available through the build process? Understand what channels the provider uses to communicate with your team, what professional services they offer, how they manage the project, whether there is an active roadmap with opportunities for feedback and what depth of documentation they provide.
3. Can it integrate with my existing systems, processes and brand?
Connecting to your existing ecosystem is one of the secrets to building a satisfying, frictionless and personalized experience for your customers and employees. Consider these key things:
- Depth of integration: Know which enterprise platforms and tools you need to connect with and evaluate whether the platform is capable. You’ll want to integrate with all aspects of your tech stack, including your backend system and existing tools, so data gets to where you need it.
- Messaging channels: Does the provider support the customer-facing channels you already use? Do they have their own live chat? Is there on-page or in-app support? And do these options align with your processes and workflows? Ensure you can be present 24/7 on all the customer-facing channels you need, on both mobile and web.
- User interface (UI): When it comes to what users see and interact with, can you create a look, feel and experience that aligns with your brand personality? Is there a rich UI toolkit? Look for sophisticated UI elements that help make conversations with customers easier and faster.
4. Is it flexible enough to evolve with my business?
Customers’ needs can change and so can the world we live in—rather drastically, as the coronavirus pandemic has shown. The best bot platforms give you the flexibility to adapt your CX program over time as your business context evolves.
When considering the flexibility of the solution, make sure it meets these criteria:
- Commitment level: Some companies shy away from platform offerings because they’re worried about getting locked into a particular solution. To avoid that, you’ll want a platform that gives you ownership of your solution, data and natural language understanding (NLU) models, so you have the flexibility to walk away at any time if you decide you need to pivot. (More on that in the next section.)
- UX enhancements: Does the platform expand and scale to support user experience enhancements? Over time, can you introduce new ways to improve their journey? Consider features like proactive messaging, which helps them along the way before they ever have to reach out to support.
- Customer volumes: Can the platform support an increasingly higher volume of customers as you grow? Make sure there are no limits to how much you can take on—or how successful your CX program can be. You need a provider that supports your growth and high performance for the long haul.
5. Does it use natural language understanding and how does it work?
While a simple rule-based bot works for some applications, more advanced interactions use NLU as well, which allows bots to make sense of words, intent and even sentiment. Intelligent bots built on NLU correctly analyze, process and respond in a natural way, even if the input contains slang, jargon or spelling and grammatical errors.
Every bot platform will vary in terms of its NLU. Here are some things to consider:
- Ownership: Be aware of who owns the NLU model in case you need to switch platforms. If the provider owns it, then you won’t be able to export the model to a new platform and you’ll lose all of the time spent training it—amounting to thousands of phrases and significant amounts of data.
- Type: Is the NLU from a third party or proprietary? Typically third-party NLU, such as Google’s best-in-class Dialogflow, is the stronger option. Some platforms use white-label Dialogflow or a subpar version that’s less accurate and has a lower F-score.
- Languages: Think about what languages your customers speak. Does the platform support all of them?
6. What are the deployment options?
Whether you host the bot yourself or use a cloud service provider, you’ll need a stable hosting environment that gives you the flexibility to expand if and when you need to. When weighing bot deployment options, think about these factors:
- Hosting: Depending on your business, you may want the provider to take care of the hosting and scaling. Decide whether you want hosting on prem or in the cloud, and evaluate what the platform can offer. Look for secure, scalable hosting in any region.
- Speed and responsiveness: When it comes to delivering good customer service, 73% of customers say valuing their time is a brand’s most important differentiator. To support your CX goals, conversational AI needs to respond quickly, resolve issues efficiently and be available 24/7. Make sure the bot’s latency is a top priority for the platform. Delays are barely noticeable when they’re between 100 and 300 milliseconds. A delay of over 300 milliseconds will feel like the bot is “thinking”, but anything over 1,000 milliseconds could lead the customer to switch tasks.
- Compliance and security: Is the platform an enterprise-level solution that meets your requirements for information access, permissions, compliance and security? In order to keep customer and enterprise data safe, you’ll need full support for encryption of sensitive and personally identifiable information (PII). Enterprise-grade permissions ensure you only give access to team members in accordance with your company's rules.
7. What performance insights are available?
I think we've shown that CX automation can transform your business over many years, helping you achieve higher levels of customer service, satisfaction and retention. To optimize those results, you need access to rich data insights that let you learn, grow and continuously improve.
Consider how well the platform stands up to performance-related requirements, including:
- Analytics: Can you analyze bot performance? Does it have built-in or external analytics? Look for out-of-the-box analytics and a reporting dashboard for the most important metrics, as well as full performance and event logs with customizable retention periods. Consider full event tracking of your users, including transcripts to know how they interacted with your bot so you can identify areas for improvement. To keep your data scientists happy, make sure the platform lets you track or export bot data to your data centers.
- Continuous learning: Does the bot improve over time? Conversational AI should become more sophisticated as it processes more data about how customers communicate and what information they need. The best platforms allow you to understand your missed intents, flow funnels and if your bot is meeting its goals.
Building a bot is like most things in life: you get out what you put in. And while the fix you need isn’t a quick one, that doesn’t mean it has to be complicated, resource-intensive or expensive.
It should, however, be flexible. Your CX strategy evolves as you go, incorporating new customer insights and offerings. And, since the purpose of conversational AI is to support that strategy, your bot needs to be able to adapt too. The answer? A bot platform delivered by experienced partners who give you the flexibility to expand and the support to modify your CX—whenever and however makes the most sense for your business.