Guided journeys with decision trees for GovCMS
Sometimes the page metaphor for navigation is too cumbersome to get your users to the answers they need. Interactive decision trees offer a way to guide users on a journey of questions and answers, arriving at an outcome which gives them the information and resources they need.
Hypertext. The web was built on it. What would you do without the humble, yet powerful, link?
The web is based around the page metaphor. Users click on links and navigate a graph of links which takes them from one page to the next. This metaphor has started to break down giving rise to web applications and single page websites. Users expect to get to the information they need quickly. Page reloads are seen as a distraction as information needs to be surfaced quickly.
The page metaphor is also less useful when the required user context is complex. Static pages provide one kind of answer for the subject at hand. How does a user find this page when there may be a series of steps required to work out just what resource needs to be shown.
Interactive decision trees provide a solution to these issues.
- An interactive decision tree is able to immediately serve the next most relevant piece of content. There is no need for a page reload and the user stays focussed on the task at hand.
- Decision trees are helpful for quickly getting the correct context from the user. The questions asked will be relevant to the current circumstances and the answers given will move the user through the process.
Take the demo for a test drive
At Morpht we have recently had several clients request decision tree functionality. Use cases have ranged from the simple to the complex and have covered areas such as regulatory compliance, information gathering, question answering, and product decision making. As this seemed a more common requirement for GovCMS sites, we decided to develop the customised work we had done in the past to a more general solution.
To see the above in action, please see our Test decision tree on the Convivial for GovCMS site. The scenario there is for a user who may be interested in the GovCMS platform. It guides them through a series of questions and leads to an outcome with some advice. This is a simple example demonstrating how decision trees work.
How does it work?
The decision tree logic and structure has been implemented into Drupal 9 and releases as a new feature on Convivial for GovCMS as well as Convivial CMS. Behind the scenes there is a Decision Tree content type which holds several Steps. Each Step has several Answers which link to other Steps. A journey is complete when a Step with no Answers is reached. We call this an outcome.
Graphs vs Trees
Earlier in this article we noted that the web is built around links which define a giant graph. Each link joins different resources into a giant network. This network is not necessarily hierarchical in nature. Links can point anywhere, including back to already visited content.
The thinking behind the architecture of Decision trees follows these principles. It is a graph rather than a tree. The following schema shows how we've designed the system.
It is, therefore, somewhat a misnomer. Decision trees do not force users down deeper and deeper. It is possible to define them so that users can navigate to Steps which are also used by other parts of the decision tree. It is possible for users to end at the same step by different routes. This structure allows more complex trees to be built - a handy feature for when there may be uncertainty or different paths leading to the same outcome.
We naturally want to know what users are doing. This is even more the case for users answering questions about themselves or what their intentions are. This is vital information that can help site owner understand their audience and their needs.
The system has therefore been designed with analytics in mind. If Google Analytics is installed each Answer and Step will be tracked as a pageview event, making it easy for the site owner to review. URLs are defined in a hierarchical nature for easy filtering. Further, with some sensible naming conventions on the outcomes, it should be easy to create summary statistics for the outcomes (the conclusion of the tree) over time.
You can think of the analytics data derived from Decision Trees as having Xray vision into your site and users. The lens shifts from knowing what content is popular to understanding just what is driving the decision-making process. This should help you build better content for the future by addressing the most common paths for your users.
Personalisation for the win
Morpht has been working in the content personalisation space for some time and has developed a set of tools for GovCMS sites. You may have seen the recent article on Personalisation in GovCMS. The underpinnings of the personalisation solution there is through user context stored in localstorage. As a user traverses a site it is possible to use certain events to indicate “intent”. For example, if a user visits an audience page, or comes in on a certain campaign, it is reasonable to infer a particular intent and then to personalise content around that intent.
The Decision tree implementation integrates with this approach. It is possible to assign certain “user attributes” to a user when they reach a Step in the tree. Generally this will be when they reach an outcome, although it could be any Step.
Taking the Test decision tree as an example, there are two Steps which assign user attributes.
- If a user lands on the Migrate to Drupal 9 step, they are assigned the audience.upgrade intent.
- If a user lands on the Enhance Drupal, they are assigned the audience.personalisation intent.
Once this intent has been assigned, it is stored in localstorage and is then used to drive personalisation on the site. When a user returns to the homepage, they will see personalisation promotions, articles, events and calls to action.
The integration between personalisation and decision trees is incredibly powerful. At Morpht, this has been a goal of ours for some time to release this functionality as we believe it has the capacity to deliver much more engaging and relevant content to users. The decision tree can be used to infer user attributes in a very specific and targeted manner. Personalisation can then be configured to make the most of the increased knowledge of the user.
The release of Decision trees on Convivial for GovCMS and to the wider Convivial CMS platform represents another step in our efforts to build features which make Drupal relevant in the modern web. We are looking to build simple solutions which deliver value to customers and users.
Decision trees take the oldest idea on the web, hypertext, and update it to make it more interactive and capable of delivering relevant content to users more quickly. The added benefits of the analytics and personalization integrations place it at the centre of a solution, which can provide better insights and more relevant content delivery.
Morpht is a supplier on the Drupal Services Panel and can be engaged to implement this solution, or a similar one for your next GovCMS project.