Why & How to Start With Comprehensive User Research
Author: Rain Michaels
Comprehensive User Research can significantly improve the quality of your website and the value you provide to your target audience. User research results in products that favor people over technology, or isolated organizational structures – products, which are capable of delighting their users, no matter how banal the tools may be.
This initial step in creating or re-architecting a web application is frequently skipped for a number of reasons. A few common justifications for bypassing User Research include:
- “It’s too expensive and I can predict what my users need.”
- “My site structure is very straightforward.”
- “It has to be organized this way because that’s what x stakeholder wants.”
The primary fallacy of all of these arguments, however, is that you are not the intended user, but rather the application owner. Therefore, your understanding of the product, vocabulary, and needs, vary from those of your target audience. In some instances, you may have multiple audiences who each have a different reason for accessing your product, and a different method or workflow for doing so.
League of Women Voters: My League Online (MyLO)
Let’s look at the League of Women Voters My League Online product.
Local League’s serve three primary proposes:
- Voter Education
- Voter Advocacy
- Providing Action Materials For League Members
This also means that they are targeting different user groups (four to be exact):
- Voter Education: Users researching how they will vote during their downtime, primarily using mobile devices and seeking information quickly; or
- Voter Education 2: Users who may have minimal access to resources and online tools, or who may be seeking information in multiple languages. They may be gaining access through publicly available computers, such as those at libraries, which may have imposed time-limits, and therefore are looking to print information and move on.
- Voter Advocacy: Users who are politically engaged, and wish to share content on social media and print resources for local groups.
- Action Materials: Users who are looking for meetings, interest groups, and ways they can volunteer.
Without researching these users, what they want, and how they will engage with the final application, it is highly likely that the tool you build will not effectively serve these groups.
While it is best to engage experts for comprehensive research, even doing some of this work on your own before diving into requirements will yield a huge advantage in building a successful product.
At Stauffer, we are flexible about how we handle User Research engagements based on the actual needs and unique circumstances (and budget) of the client. Ideally, however, there is a full-suite process that we would love to follow.
The Ideal User Research Process:
Step 1. Determine The Context
Our first step would be to identify the context of your existing site user needs (assuming that an existing product exists). This includes a full analysis of:
- What currently exists?
- How is it currently organized?
- What do analytics show about the fidelity and findability of the content?
Whether or not there is a pre-existing product, we also need to ensure that we have a clear understanding of the culture, within which the organization and/or existing site and its content exist:
- What unique vocabularies are used within the organization?
- What arbitrary expectations and/or structures are considered standard within the organization?
Finally, we need to understand all the possible user types for whom the web application will be designed (as we briefly looked at with My League Online above). This way, we can ensure all their unique perspectives are considered. This comes in the form of accessing any pre-existing user personas and/or creating new ones as needed.
Personas need to include both obvious users (e.g., in the case of My League Online, a voter looking for ballot measure details) and the less obvious ones (e.g., a candidate looking to gauge the community temperature surrounding a ballot measure, or a local League Webmaster looking to post community events).
Step 2. Gather Input
Ideally, this would include a combination of two input models: Generative Input and Evaluative Input. In many cases, clients are only able to follow one model. If this is your circumstance, choose whichever one makes the most sense given your resources and desired result.
In this model, we evaluate how people find data based on the structures available to them. This may include a combination of the pre-existing site structure (if one exists), and examples of an initial plan that you think you may use before initiating User Research.
A key component of this model includes interviews:
Identify a selection of individuals who reflect an effective sampling of the types determined by personas and ask them for feedback on the existing site.
- Find out what they like.
- Find out what they come to the site for often.
- Find out what they wish they could access.
- Find out where they are frustrated.
Ideally, interviews will not be conducted or attended by anyone who was or is perceived as being involved with the production of the existing site so that no one modifies their feedback to avoid hurting feelings. This process can be done via remote web conference calls or in-person interviews.
If a low budget alternative is required, web surveys are a cheap alternative to actual interviews. If you use this method alone, however, be aware that strictly using a survey method is likely to result in skewed data, as only the most aggrieved participants are likely to provide substantial responses.
The idea behind this model is that we identify the most intuitive, but perhaps not so obvious, structure for the targeted users. We gather input from representative sample users through either Open Card Sorting, Closed Card Sorting, or a combination there-in.
- Open Card Sorting: We provide users with a collection of cards with categories, and ask them to group the cards in patterns that make sense to them.
- Closed Card Sorting: We present the users with a pre-determined top-level set of categories, and ask the users to group the cards we provide under those pre-set groupings.
Open Card Sorting can deliver a more organic site structure, but it can also generate a significant amount of information that must be analyzed, which may be overwhelming.
In-Person / Remote
All these techniques can be run both in-person or remote, using web-conference and cloud-based card sorting applications. Both practices have their pros and cons, so the primary factor in determination is based on access and resources.
Step 3: Analyze & Evaluate
Once interviews have been completed and any card sorting processes have been run, we must look at the results and identify the clusters in which user tendencies go the same way. Especially valuable are those results that we do not expect.
Preferably, we would then take this data and develop a categorization and site structure from it, and then prepare a next level of testing: Tree-Testing. In this model, we use a cloud-based application to have users run through a list of tasks and see if they can complete them solely using an outline-style navigation. If it is possible to do this as part of the process, it will help us identify potential mis-steps prior to investing time in wireframes and/or interactive prototypes.
Step 4: Lo-Fi Prototypes & Usability Testing
The final step in the Ideal User Research Process would be to develop Lo-Fi Prototypes (wireframes/gray boxes) and present them in a “clickable” environment such as InVision. We would then run representatives of the sample target users through this interactive prototype, and find out where the interface may not be as successful as desired. This would be an iterative process, with potentially 3-4 rounds of testing and revision.
Another option, if budget and timeline allow, would be to perform A/B Testing. We would set up two different versions of the interface, and review analytics and metrics regarding performance to determine which version is more successful with the target audience.
Usability Testing can be run in-person, using an application such as Silverback (Guerrilla Usability Testing), or web-conference (with screen-sharing). As with interviews and card sorting, all three options have different pros and cons, and which approach we take should depend most on what will best serve the user-personas determined in Step 1.
The Level of Importance User Research Plays
When all else is said and done, your application is only as good as its ability to serve its users and meet your business goals. If your application does not feel organic to and delight your users on some level, they will find other resources that do. Without knowing and understanding your users — and accepting that their natural flow may not match the one you expect — you will create a subpar product, and you will either fail, or end up having to re-invest more resources to fix something that was not built well initially.
Think of your favorite web experiences. These applications are ones that do not necessarily follow any preset patterns. They may not be the glossiest ones. But, they serve your needs because the product owner(s) understood your needs from the very beginning, and built a tool specifically for you, rather than for any preconceived notions of what should be created.
Looking for assistance with your User Research and Product Planning? Let Stauffer help!
Contact Us today to learn more about our suite of services and to get started.