Meet Lucena – a Smart Answer to Your Search

By Dominika Gallot and Jeroen Furst in Development
·7 min read

In our digital era, the world is flooded with information. Making it available is one thing, but even if your content is concise and your website has a great structure, how do you make sure people can find what they’re looking for? This is not another blog post about content personalization. This article will give some ideas to those who cannot personalize content on their website.

Respective Perspective

Let’s consider the case of a patient suffering from arthritis (Let’s call him Archie). He’s in his 50s, and he can browse the Internet fairly effectively. However, when confronted with a hospital website, he’s suddenly at a loss. He needs to find out how long he can expect to wait for his examination, but the website is overflowing with information about various departments, services offered, details about, hospital maps, visiting hours, content about all kinds of afflictions, etc. This leaves Archie overwhelmed and frustrated. There must be many patients interested in the same thing... Why doesn’t the hospital make it easier to find information like this? Why is this so difficult?

Now, look at this issue from the hospitals’ perspective. There is so much information they need to relay to the public and structuring a website to fit one group of people is not an option. They need to provide information not only to patients but also researchers, visitors, internal and external doctors, etc. Hospitals don't have armies of marketers and content editors who would be able to curate content to all these target groups. How would they even do that? In order to display personalized content, you need data about your visitors. Marketing departments store data aiming to build visitors’ profiles to be able to market to them better—facilitating the purchase of services or products. That stands in stark contrast to what hospital websites do: provide content that helps patients, doctors, and other communities, without any sales agenda. Finding the right content often comes at the price of providing personal information. Hospitals understand that their visitors may not be comfortable doing this due to the sensitivity of such information, which in turn makes it more difficult to direct them towards the content they need. The website needs to be generic enough to cover all bases across the main target audiences. The information is there, but it is up to the visitors to find it.

This is the gap we need to bridge. Let’s talk about what we can do to help people find what they need.

Building Bridges 101

One of the keys to providing relevant information is to know whom it is for in the first place. You don’t need to think twice to know that Archie needs different content than, say, a doctor referring him to the hospital. One way of doing this is to identify personas and serve them personalized content. Here you will run into problems. In order to correctly match website visitors with their personas, you need to have gathered information about them and their behavior. We have already established why this may not be an option for hospitals. Besides, how much time do visitors need to spend on your website for you to be able to decide which personas they are? How many pages or actions do they need to go through? There’s an easier way to do this. Come up with your top 3-5 target audiences and provide your visitors with a way to self-identify to which audience they belong. Your website could feature sections for those who identify themselves as patients, doctors, researchers, and other audiences.

The next step towards bridging the gap is to identify which pieces of information visitors look for most often, or the most common tasks they might perform on the website. This helps you to focus on relevant content and make it stand out. By doing so, you will inevitably make it easier for your website visitors to find what they’re after. There are several ways of identifying these top tasks, beyond simply referencing website statistics. To be thorough, you might want to identify all tasks that a visitor might need to perform—which can be done using a combination of website behavior analysis, surveys, search analysis, corporate philosophy and strategy, etc. The process results in a long list, which needs to be edited down and segmented according to the target groups you have identified previously. At the end of this process, you will have established top tasks for each of your target groups. Now you can get creative with how you present them.

Trying to Be Helpful

You need to provide a way for content retrieval. You could argue that visitors can simply use the search function to find what they need and you can avoid all the above. Right. Many websites have this nifty function that is supposed to supply the visitors with answers to their questions if they cannot find them easily enough by browsing the website, but is it really that helpful? .NET CMSs use technologies like lucene.net to provide “good enough” website search functionality. The results of such a search are ranked by relevance. However, which pieces of content are most relevant are determined by keywords and their frequency. This could be problematic—two texts with 100 and 500 words both feature the same keyword three times. The one with 100 words is rated as more relevant due to the higher ratio of keyword to text, but it doesn’t necessarily mean it is a better match.

Meet Lucena

You can make looking for information almost effortless by introducing a chatbot. It will allow you to take shortcuts in identifying which target audience a particular visitor belongs to by simply asking them. A chatbot inevitably improves visitors’ experience with your website. Instead of being left to cope on their own, they get a personalized and much friendlier experience. A chatbot allows you to be proactive in helping your website visitors from the first interaction by searching the website content for them.

Introducing a chatbot to the website can be easy; especially if the website is a microservice-based model. In this case, having a chatbot is just a matter of introducing another service to the package that you already have. To do so, you will need a technology that understands language and is capable of extracting visitors’ intents from their queries, but also technology that can clarify the intent. Chatbots are great, but the problem with using them on hospital websites is that people may not realize they are talking to a chatbot. This leads to confusion and frustration when the bot returns inaccurate results. In our case, we realized that identifying audiences and their top tasks is not enough. We wanted to create a solution capable of reaching other communication channels as well, and enable people using AI devices. AI even allows you to train the solution you use to retrieve content in order to learn from its mistakes. To bridge the gap between hospitals and their audiences, we introduced Lucena.

Lucena is a cross between a search and chatbot. We could say she’s an extension of the search function. She looks like a website search, but is capable of asking for additional information in order to provide relevant results. We modeled data based on the target groups and top tasks. Lucena only stores information within the context of the query as to avoid all the challenges of storing personal data.

Bright Future Ahead—Bring Your Sunglasses

After investing time and energy into creating Lucena, we don’t want it to end there. Just like a chatbot or solutions similar to Lucena, you don’t need to end with a website. Once you have created a microservice solution, you can introduce other services and use your chatbot as a baseline for an omnichannel solution, allowing people to use mobile and AI devices (such as Siri, Alexa, or Google Home) to help them with their queries. Imagine how much easier it would be for Archie to simply ask, “Alexa, how long do I need to wait for a rheumatologist appointment at St. George hospital?” instead of laboriously trying to find this piece of content on the hospital website.

There are several things that we would like to see in the future. For instance, we would like Lucena or any similar solution to learn from the context. If Archie asks about the opening times for the Rheumatology Department and subsequently asks for a phone number, the solution should be able to recognize that he’s asking for the rheumatology phone number. In addition to that, it would be interesting to see AI-generated content. As for Lucena, the texts she uses to ask for additional information are based on the top tasks for each target audience. They are hardcoded now, but this is not possible to do for all questions website visitors may ask. We would love to see Lucena generate some answers of her own. When talking about AI-generated content, imagine a service that automatically retrieves tags. Using alt texts for images is one of the website accessibility basics—they make it possible for screen readers to understand what is on the page, they provide context for search engine crawlers, etc. So why not have AI add metadata automatically—similarly to Facebook’s automatic alt text?

That’s our vision of Lucena’s future. If you want to find out more, let us know. We’ll introduce you to her.

By Dominika Gallot and Jeroen Furst in Development
search
Gartner report
We`re named a Challenger in the 2018
Gartner Magic Quadrant for WCM!
×