2. Design and align on your goals
Set clear expectations, goals, and strategies by deciding on KPIs. Then, map out existing workflows and prioritize which you’d like to improve. This will determine which products or channels to start with for most effective results.
Check out this free event on how to measure automation success.
“Make the virtual agent a member of your team. Don't think of it as external software that you need to get the hang of, but a tool that you can teach to follow your processes.”
- Maeve Condell, Customer Success Lead at Ultimate
We recommend starting by launching a generative AI-powered knowledge base bot. This is the fastest and easiest place to begin — and will deliver near-instant time to value. You won't need to invest a load of resources upfront in training an AI model. Simply:
- Connect a gen AI bot to your public knowledge base
- Let the bot analyze your existing help center articles
- Start deflecting FAQs and resolving customer queries
Learn how to build an AI chatbot using ChatGPT.
Once your generative bot is up and running, you can start thinking about adding workflows, building API integrations, and mapping out dialogues to ramp up automation rates.
3. Set up your team for success
Once you move beyond knowledge base bots, it can get a little more complicated. Depending on the complexity of your automation project, you might want to assign a project lead, a bot builder, a CRM expert, and potentially specialist tech resources to build backend APIs. Having the right team in place will help you to start building out a more holistic automation solution quickly. As with all things automation, your mileage may vary, but we recommend carving out more time to work on your initial setup — designing conversation flows and optimizing the end-user experience.
Learn more about what resources you need to allocate for implementing automation.
4. Set up your existing technology for success
Look for automation providers that can seamlessly integrate into your existing tech stack, including your backoffice systems like your CRM or your order management system. While many automated customer service platforms are no-code, you may need additional IT support for backend integrations.
5. Obtain your main customer pain points
Many automation providers offer tools to discover your most popular customer intents. Testing them on your unique support data is a win-win for you: If you already have an idea of your top intents and their tool confirms that result, you know the AI is good. If you end up finding additional intents, then you have an even more precise idea of what to automate first.
Or (with the help of generative AI) you can do away with intents altogether. The beauty of gen AI-powered knowledge base bots is that customers can ask about anything — and as long as you have a help center article on this topic, the bot will be able to give an accurate response. Where intents and dialogue flows are still useful is with carrying out specific processes related to a topic: like creating a returns label, or pulling subscription details from your back office systems to offer personalized answers.
Learn more about how generative AI and conversation design work together to elevate CX.
Want to find your most common customer intents and discover your automation potential? Get your free, customized report from the CS Automation Explorer.
6. Prep for launch success
You can launch a generative AI bot (like UltimateGPT) in just minutes. But for chat and ticket automation that relies on pre-designed dialogue flows, it takes a little longer to get up and running.
Most automation providers will help you go live within 2-4 weeks, depending on the channels you’d like to automate first. (Generally, ticket automation can go live a little more quickly than chat automation.) Before your official launch, it’s time to put your new virtual agent to the test and throw everything at it! For example, at Ultimate, we usually guide customers through a ‘break-the-bot’ session to test their AI’s performance and improve chat conversation flows.
7. Achieve long-term success
Now’s the time to monitor and iterate, and assess your performance against your KPIs. You’ll want to raise your automation rate further by training your AI and expanding the number of automated intents. Then, it’s time to think about a long-term growth strategy with added KPIs around 24/7 support, additional languages, or channels.
Learn how to get started with support automation from TransferGo's CCO
So the steps to getting started sound simple enough in theory, but what about in practice? We sat down with Povilas Čiuplys, Chief Customer Officer of TransferGo to chat about:
- What pain points led TransferGo to consider automation
- What the onboarding and launch processes were like
- What results they saw immediately after launch as well as further down the line
Watch the free on-demand event and hear what the getting started process was like from someone who’s gone through it.