The role of support leaders has undergone a huge metamorphosis. They are no longer just improving customer satisfaction and lowering support costs, but are now expected to grow business by retaining and growing their customers.
This impact is visible throughout the support organization, but as support leaders' roles have evolved, new challenges have surfaced too. They essentially boil down to two major aspirations: First, enterprises are trying to build personalized customer experience on self-service portals, and second, support teams are trying to upskill agents to meet customer expectations. Recently, AI and cognitive search are being seen as a possible solution to these problems, but there are a few things you should be aware of before you make the investment.
Why Cognitive Search?
Cognitive search is one of the first AI investments support organizations should make. However, it’s important to note that search is not just a search box; it’s an underlying technology that powers more apps. Businesses have observed a significant amount of payback and ROI on this investment, so it comes as no surprise that more of them are investing in it. The adoption of unified search has risen from a mere 7% to 19% from 2018-19, according to a survey by TSIA.
Cognitive search is one of the most crucial investments as it affects various facets of an organization directly—employees, partners, customers, users. However, only investing in technology is not enough; you need to have a well-thought-out strategy around the technology solution to make the most of it. You need a solution that understands your tech stack, content architecture, content strategy, audience while offering insights into how they can constantly improve.
6 Things to Look For Before You Make the Investment in Cognitive Search
In my experience, support leaders looking to invest in cognitive search solutions are usually told what the solution they are investing in can do, but are not guided well enough on how it directly affects their team, tech stack, and processes.
So, I want to walk you through a checklist that every support leader should be aware of before making this investment.
1. Connectivity
Why It’s Important: Growing enterprises have multiple siloed repositories with a huge amount of unstructured data, which makes it difficult for agents to find answers. This affects their productivity. To make all this information easily accessible, your enterprise search platform should be able to integrate all these silos.
What to Look For: The best fit for you will be the solution that offers out of the box and natively built connectors for the tech stack that you are using, rather than one that engages a third-party vendor for the integration and implementation. Go for a search engine that is smart enough to automatically extract information out of all the unstructured data files like e-mail messages, word processing documents, videos, photos, audio files, presentations, PDF documents, among others.
2. User Personalization and Experience
Why It’s Important: Customers expect personalized experiences and only displaying answers based on the keywords that users search for is no longer cutting it.
What to Look For: Your search engine should be capable of processing natural language and understanding the user intent based on the query. Not only should it serve content that is tailored to the user but also should allow the user to customize the interface according to their preferences—whether they want to see or not any particular elements on the search page like filters, recommendations, or related articles.
3. Analytics and Insights
Why It’s Important: As a support/business leader, you would want to measure the outcome of your support initiatives so you can make data driven decisions and improve your strategy. And for that you need an insights engine that enables you to measure case deflection accurately.
What to Look For: What differentiates a good insight engine from a great one is how deep and comprehensive those insights are. Also, it must use NLG (natural language generation), a technology that translates data into plain English and highlights what’s important for you.
Case deflection is a metric so important that you want to know what reports the platform offers to accurately measure it. I’ve seen various leaders measure deflection in different ways. You would expect your platform to measure it the way you define it.
4. Agent Augmentation
Why It’s Important: You want your support agents to stay motivated and focus on more important tasks instead of doing mundane tasks like answering L1 queries repeatedly. They can’t do so because they’re stuck with old technologies and processes. Also, you want to drive knowledge-centric support.
What to Look For: AI-powered tools that give them all the information they need to do their job efficiently and automate mundane tasks. So you need an agent helper, which clusters all your historical cases and creates a classification of all the issues. Once you get a new case, the agent can get to see the relevant similar cases which have been created in the past, the top expert agents and top articles which were used in the past to resolve those cases.
And you need to check whether your search solution is KCS-aligned or not. KCS-aligned technology enables pre-population of articles which the agent wants to create with the information they already have in the case or in the historical search queries. So at the end of the day, ML-driven auto-population ensures that agents just need to write snippets based on the case they are working on to create a new knowledge article, rather than writing the whole article from scratch.
5. Search-Powered Chatbot
Why It’s Important: Millennials love to chat! 63% of millennials prefer to have their basic customer support queries answered by live chat. They prefer asking questions to a chatbot over searching for them. So, you need to invest in a chatbot along with a search engine.
What to Look For: Chatbots are widely considered a standalone product but I’m a strong believer in the idea that chatbots need to be a part of the cognitive search framework.
When you implement chatbots as standalone products, it becomes problematic when you want to train your chatbots on L1 questions because you have a lot of queries like that on your support system and these L1 queries are constantly evolving. So, it becomes very difficult for your admin or your support manager to actually train your chatbot on these queries.
The beauty of having a chatbot that is powered by your search engine is that it can train on the basis of your historical queries. We don't want support managers to spend their precious time training the bots. The insights engine has sufficient information i.e. historical cases, search queries, the user intent, etc. all of which can be leveraged to train these bots.
6. Security
Why It’s Important: Like every potent technology, cognitive search also has its perils. While it makes data discovery and accessibility a breeze, we can not deny that it increases the risk of a data breach as well. Therefore, you want to be extra cautious about how secure your search engine is.
What to Look For: How and where the vendor stores your indices, what measures it takes and technologies it uses to protect the data in transit, how it respects the access level permissions. Do they provide a single-tenant or multi-tenant solution? While those audit reports SOC and certificates like HIPAA compliance are a good way to gauge how secure the solution is, you do want to read the contracts carefully to understand what information the vendor collects and how it uses it.
More Resources About Cognitive Search
To conclude, I’d like to emphasize that you should not look at cognitive search as a search box on your community or site—it’s much more than that! Look at it as a platform, an underlying technology that further powers a number of apps that augment your agents and delight your customers. That’s the future of cognitive search, and while making an investment in a search platform, make sure it’s future-ready.
To learn more about cognitive search, be sure to check out these additional resources: