Tips for Drafting Guided User Interviews
As technology becomes more and more accessible and intuitive in the Access to Justice space, many legal service providers are taking the leap to online triage. With the Louisiana Civil Legal Navigator, we’re doing just that. This article outlines some tips and tricks for drafting your own triage interview questions.
In my previous blog post, I talked about how we looked to data to make decisions about what areas of the law we should focus on in order to impact the most people. Having made those decisions, it was time to start really focusing on content – after all, triage is virtually useless without information that helps guide the user to their next destination.
For our project, there are essentially three ways individuals in need can access content on the site: by using the search function, by manually reviewing different categories of information, and by using the “Guided Assistant” feature. The Guided Assistant drives the user through a series of questions that helps narrow the scope of their issue(s), which, in the end, narrows the scope of information they receive.
Most legal service providers that have robust triage systems built into their websites are no stranger to this concept of delivering targeted information. And as most know, getting the questions for triage just right is no simple undertaking. It takes careful planning, an empathetic mindset, and most importantly, a lot of patience.
I can’t give you empathy, and I can’t monitor your patience meter, but I can help you plan! Hopefully some of the tricks I’ve learned as my incredibly empathetic and patient team put their energy into perfecting the Louisiana Civil Legal Navigator triage interviews will come in handy if you’re embarking on a new triage project.
Tip #0 – Find the Right Tools
This is ground zero for any successful attempt at creating complicated decision trees. I spent hours in agony drafting and redrafting trees and searching for the right tools that would help me properly manage this process. Some of the more important considerations for me were:
Ease of use and sharing (for engaging with attorney volunteers and working groups),Depth of mapping capabilities (for a clear picture of where and what type of information was needed and how the interview and user would behave in a more dynamic environment),Note-taking and commenting abilities (to pose questions to the working group, and use as a temporary storage environment for associating responsive information), andCost (always!)
For this exercise, I tried Visio, Mindmeister, SmartDraw, and Draw.io, but ultimately landed on LucidChart. It is slightly more cumbersome to use than, say, Mindmeister, but it accommodated most of my functionality needs, and is fairly inexpensive – it’s worth the monthly fee to have unlimited files and storage.
Version one began in Mindmeister, but things really started coming together when we made the switch to LucidChart.
Divorce Guided Interview v1 – a heated mess.
Divorce Guided Interview v2 – such beauty and purpose.
I also found that flat decision trees were just no substitute a wireframe or mock up – there’s just something more tangible about the interactivity. I think it’s the dynamic movement of conditional questions and clicking “continue” to view the next page that makes it more “real,” giving you the opportunity to put yourself in the actual user’s shoes. It’s also much easier for stakeholders and working group members to get a handle on progress and identify other places the interview might need to go.
My saving grace for this task was a new tool unveiled at this year’s LSC Innovations in Technology Conference – Community.Lawyer. If you haven’t gotten the chance to play around with their software yet, I highly encourage it – it’s free(mium)! I think their priority is building easy-to-use, user-facing “apps” that automate documents / complete forms, but it worked very well for my Guided Interview workflows.
Now, with the right tools in place, let’s think about how to best draft the interviews from a 30,000-foot perspective.
Tip #1 – Start with the Big Picture
I’m sure this is obvious for most, but I’m trying to outline the process, so it’s only fair to start here. I suppose though, I’m not talking about your big picture, but rather the user’s big picture. What do they want to accomplish? What situation do they find themselves in, and what do you need to know to help them accomplish this?
In my working group, I had the attorneys think about what information they’d need to know to give the most targeted information. This resulted, of course, in a lot of factual issues to consider. For example in divorce, are there children involved? do you agree on dividing property? will you need spousal support? But it also led to a great number of procedural questions – has service been made? has the case been abandoned? has the other party waited long enough for X?
Triage (as *not* legal advice) has an interesting dynamic. It aims to sort information in great detail within a topic but allowing the user to “choose their own destiny” with the tool. Question and answer options that focus on a user’s desired action – what they want to accomplish with their session – are the best way to touch on all of the possible situations they might find themselves in while giving them a sense of control over the outcome.
Tip #2 – What are your Non-Negotiables?
Anytime you are building a new interview or questionnaire, always start with your non-negotiables. These are the things that you’ve decided the system will absolutely not be able to help with. Work closely with the stakeholder group to identify these things and lay them out up front.
I’ll use an example from our project in the divorce context. In Louisiana, our working group determined that, at least initially, this tool would not be able to provide detailed steps and information for non-standard divorces (i.e., divorces based on domestic violence, adultery, or felony conviction of a spouse). The risk of providing information and it not actually being responsive to their circumstances was too high. In those cases, it’s best to screen those individuals out from the get-go and give them general information and a referral to the legal services organization(s) and/or the bar referral service.
Divorces based on DV, Adultery, and Felony Conviction of a Spouse are off limits for now.
These types of projects are a delicate balance of fulfilling a desire to help as many people as possible, but also realizing that at least a consultation with an attorney goes a long way towards saving filing fees and reducing unnecessary volume in the courts.
Tip #3 – Break Down the Individual Issues
For us, this was about reducing the number of dependencies that lines of questioning would have. Of course in certain cases, questions are dependent on those that came before it. For example, questions about how long its been since the spouse was served in a divorce (which will determine what type of information you give on waiting period for filing default, etc.) are dependent on whether or not service had to be made on that spouse out of state.
In contrast, questions about whether or not a user might need interim spousal support is not dependent at all on a specific divorce type, or whether or not the spouses agree on dividing property or debt. Those types of questions can move up a level because they will be asked regardless of how the user answers the previous question. So, the flowchart will look wider than it will deep in those situations.
The primary benefit for this structure is going to come later in content association. For this project, information is assembled dynamically on the back end as the user walks through the questions. Breaking it down to specific issues without dependencies lets you create discrete content that is exactly responsive to how the user answers a particular question. And that is great because it means you can zero in on what content needs to be produced in order to satisfy the user’s knowledge gap, but it doesn’t have to be integrated into X number of versions of an article (like divorce, divorce with kids, divorce with kids and property, and so on). Instead, there’s one source of truth, and that information can be reused in other places.
This is especially useful for procedural information that does not really change. For example in child custody, there are many different lines of questioning and user scenarios that will incorporate questions about whether / how the other parent can be served – filing a new suit, modifying an existing order, enforcing one, etc. If we don’t create dependencies, that service information can be used in all of those places. They are also easily used in the filing of a divorce.
All of this also ultimately helps you reduce the number of questions in your overall interview, and that’s a win for everyone.
Tip #4 – Draft Questions with Intention
Again, I’m sure I’m preaching to the choir on this one, but it’s very important to be intentional in the drafting of interview questions. In the Access to Justice space, the most important issue is readability – if the user cannot understand the question, they may find themselves frustrated and/or unable to derive any utility out of the tool.
Most would like to see a Flesch-Kincaid (FK) reading level of 5th grade. And most will also agree its extremely difficult to meet this threshold. This is largely due to the fact that we’re trying our hardest to make our questions and content understandable, but also legally accurate – and the legally accurate part throws a kink into all of the best laid plans. Alas, we do our best.
Another tip I picked up in this exercise is give the user clues about where they are. I mean this in the traditional sense like “you’re almost done” type clues, but also from the standpoint of making sure they are in the right line of questioning. For example, just entering the divorce interview gives a user a panoply of options – filing for a new divorce, answering one filed against you, enforcement, etc. On the off-chance the user selected the wrong action or scenario, thoughtfully drafted questions like “was the divorce you filed a X divorce or an X divorce? This might clue them in to the fact that they are in the wrong place if they are looking for different information on, say, answering a divorce instead.
These are just a few of the things I would recommend if you’re just getting started with building a triage system. Of course, every case and every set of stakeholders have their own idiosyncrasies that may take you in a different direction, but hopefully these are some helpful tips.
I also welcome any others and will happily amend this list to create a crowdsourced guide to drafting interview questions. To do that, email me at firstname.lastname@example.org.