Last week I was starting to design a new demonstration environment for our Search and Knowledge Discovery practice. It would simulate a search application for a high tech call center, searching across a customer support case repository. I was having difficultly coming up with a scenario that would be applicable to a wide audience and include all of the diverse features that I wanted to show off. I realized I was falling into a trap just like many of my own clients.
Ask most people what an enterprise search application looks like and they will probably describe something very traditional – a search box across the top with 10 search results down the middle of the page. The results will come from predictable sources such as the corporate intranet or knowledge management repositories. And the experience will be the same for all users, except for security trimming that might hide certain results based on document permissions.
This kind of enterprise search application can satisfy some employees some of the time, but it is far from perfect. It doesn’t take into account the diverse job functions within a company and the different search needs of each. It doesn’t learn from experience and get better over time. And it doesn’t promote the discovery of information by exposing the relationships between different content sources.
I realized that my demo design had the same problem. I was trying to jam not just a square peg, but a star shaped one and two triangles, all into one round hole. I needed to address the needs of different audiences, and by doing so, I could better serve each one. I needed a search application with different perspectives for different personas within the company.
Persona #1: Self-Service Customer
First, I should create a view for customers attempting to answer their own questions. It’s potentially the same data that an internal agent would search, but I could demonstrate different techniques for increasing relevancy and satisfaction. Let’s assume that a customer owns a particular product from a company. They frequently return to the website searching for new issues with the same product. The search application watches and learns. It learns that they are interested in that particular model of cell phone or photo editing software or toaster oven. The next time they visit, the website proactively shows them that a new firmware or software version or safety recall is available for their product. Or a search is automatically run showing any new (and relevant) documents available since their last visit, such as a new version of the user manual for their product or a new safety notice. We are all familiar with search engines like Google interjecting advertisements or that our spelling was incorrect. Imagine the self-service search application suggesting “I have noticed that you are interested in the developer API for our banking application. Here are 4 new documents I found on that subject since your last visit.”
Persona 2: Call Center Agent
Second, I should create a view for the call center agents answering calls or emails from customers. Speed and accuracy are critical here because they only have a few minutes to find the answer to each incoming question. Using data analysis, we can offer the call center agent several timesaving features in their search application. By analyzing the contents of incoming emails, we can automatically find and load pre-typed reply templates. By watching the notes being entered into a new case, we can automatically search and display existing cases that appear to be about the same issue. By exploiting the relationship between cases and the agents who resolve them, we can search and suggest other agents that have resolved a large number of similar cases and recommend them as an expert that can help with this new case.
Persona 3: Call Center Analyst
Third, call center analysts can use the valuable search traffic from the self-service portal and the agent console to spot issues or trends in near real-time. Imagine that a cell phone OS upgrade comes out on Thursday night and causes a flood of issues with the company’s application. On Friday morning the call center is hammered with calls about the issue. The call center analyst will have a dashboard showing clusters of new searches in real time. The OS issue will be spotted quickly and a bulletin can be issued. The analyst can also view daily or monthly search trends to improve the quality of the knowledge base over time. Reports of futile or fruitless searches can help spot missing articles or information.
Could I have satisfied all three types of users with a traditional search box and 10 search results? Definitely not. This does not imply that we should create completely separate and distinct search applications. In many cases, the source of the content is the same but each persona needs to see different relationships. And the search traffic from one persona can power reports for a different persona. By creating a tailored experience for each persona we can display the most relevant and useful information. One size does not fit all in enterprise search.
