First things first: AI is not coming for your job. However, AI tools are fast becoming part of daily workflows, and as customer service professionals, we will all be incorporating AI tools into our jobs.

As customers, we know the frustrations, like knowledge bases that are hard to navigate or slow reply times for tickets with easy answers. Focused AI tools are attacking these specific problems head on, helping customers find answers better and more efficiently and agents do their jobs quicker and easier. These tools will convey competitive advantages and unlock new revenue from happier customers.

That said, it’s a new and perhaps daunting area, owing to new terminology, workflows, and doom-and-gloom news articles. In this post, we’ll review the two main emerging categories of customer service automation tools, and what you should look for as a service professional investigating them.

The two categories are:

  1. Self-Service Products — help customers answer more of their questions, instantly
  2. Agent Augmentation Products — help agents do a better job with smart tools

At Myra Labs, we love self-service products as they can help customers before a ticket is even created, but many organizations will benefit from using agent augmentation products as well — they happily coexist as they tackle different parts of the service journey.

Making it easy

AI tools will eventually find their way into many areas of business. A Deloitte study found in 2017 that the chief obstacle of integrating AI initiatives into business was the difficulty of integrating “cognitive” projects into existing processes and systems. Self Service and Agent Augmentation products have emerged within customer service as the most exciting areas of AI tools because they pair clear business value (deflection, efficiency) with easy integration and maintenance.

Self-service products

Self-service products provide customers with automated assistance as they start their service journey. This helps more people than an augmentation platform. For each customer that submits a support ticket, there are likely many more customers who don’t bother to file a ticket and instead wander off dissatisfied.

Automated assistance here is particularly powerful as many thousands of tickets submitted each day are readily solvable by content that already exists in a knowledge base or can be taken care of with basic diagnostic workflows. Those customers either didn’t find, understand, or trust the information available, and self-service products help close this awareness gap.

Executed well, this can be very powerful, reducing contact volume by up to 40% and actually enhancing customer satisfaction.

Good self-service products have a few key components:

  • Easy integration with your existing knowledge base: if you’re like most companies, you’ve invested significant efforts into developing your knowledge base, and will want to re-use that content going forward. By using your existing knowledge base, it maximizes the use of that investment, maintains a single system of record, and minimizes process change.
  • Workflow support: A self-service platform that offers just knowledge base search is of limited utility. Instead:
    • Incorporate the ability to make API calls and do things that help users resolve their issues directly alongside your knowledge base content.
    • If a user is expressing a hard-to-solve but well-known issue, workflows should proactively ask them for useful information that an agent can use downstream.
    • Workflows should connect to external APIs, handle their credentials, pass custom information, and be able to take action based off what the API returns.
    • Connecting to well-known services like Salesforce and Zendesk is a great help.
  • Deep learning: Deep learning, a relatively new variant of natural language understanding, provides the ability to automatically recognize synonyms, remember context, pick up on subtleties in sentence syntax, and figure out which sentences in a longer text are meaningful. When a self-service platform doesn’t properly handle a user’s problem because it’s not using the latest technology, the user’s satisfaction suffers.
  • Complete analytics: Self-service AI products offer a new opportunity to finely instrument a customer’s journey and analyze how effective specific pieces of content are at answering given a category of questions. In a traditional knowledge base, the cause-effect between content and deflection is much murkier.

Agent augmentation products

Agent augmentation tools arm agents with efficiency tools within existing CS workflows. They listen in on your incoming text-based tickets and use machine learning to predict each ticket’s tags, metadata, and usually a suggested response.

You’ll be looking for a few key components:

  • Direct integration into your ticketing platform: in order to augment workflows, these tools need to be directly integrated into Zendesk, Salesforce Service Cloud, and or your system of choice.
  • Easy-to-use response choices and editing: Agents will want easy access to the results of the platform’s predictions, in case the agent needs to edit the proposed response or use one of the other response choices.
  • Deep learning: Yes, again! It’s a virtual requirement for agent augmentation to have significant business impact. Deep learning provides highly accurate tagging of tickets. Older approaches can be useful, too, but often have a high error rate — when the prediction is wrong, no time is saved.
  • Controllable auto-responses: When an augmentation platform is very good at recognizing certain types of tickets, it should be able to send appropriate responses automatically. However, CS Tools Managers need to have fine-grained control over what types of tickets will be responded to, at what level of recognition confidence they will be responded to, and what the response itself contains.

Conclusion

Customer service automation tools, both Agent Augmentation and Self-Service products, offer awesome possibilities for enhancing customer satisfaction, enhancing agent efficiency, and decreasing existing contacts and cost by as much as 40%. At Myra Labs, we’ve developed a unique approach to Self-Service focused on a unique blend of live diagnostic workflows and existing knowledge base content, but think that a variety of approaches can work for different companies with different needs, operations, and sizes.

What’s the same, though, is that the time to get serious about implementing customer service automation tools is upon us. Early movers will capture significant advantages.

%d bloggers like this: