Problems:
* Theme Park University is a leading themed entertainment blog covering primarily Disney and Universal attractions in Central Florida.
* Their website initially had a high bounce rate. Readers would access the site via a link to a single article (shared on social media) and wouldn't continue to read more of the site's content after finishing that article.
* Their website was designed only for one fixed, desktop-only screen width.
* There wasn't a clear way to make systematic, user-centered design decisions concerning a possible redesign of the site.
* At this time, we could not pursue a redesign. I also could not discuss a possible redesign with TPU readers, so this project needed to be a design concept done without contacting any of them.
Solution:
* Phase A of our project is featured in my Behance profile and consists of general usability improvements to the existing desktop site. It improved all of TPU's main metrics in Google Analytics.
* I received approval from the site's owner to create a redesign concept, provided that I did not contact the users (which would lead them to believe that it was really being built).
* My user research effort began by analyzing over 150 comments on the site's Facebook page, from about 100 different users. From here, I was able to determine the underlying goals and frustrations that many of these readers had.
* I used this to generate personas representing the people who would use the site.
At first, I classified the commenters based on the park chains, attractions, goals, and frustrations expressed in their comments.
I also went onto their public Facebook profiles to learn more about the demographics and goals of the people who use this site. An attractions-related website that is primarily for people who live in Central Florida would have very different content from one for out-of-town Florida visitors.
In each of my screenshots, I had the commenter's name, company, and biographical details for reference. (Again, I did not use any data that was not made available to the public.) This information is redacted from this portfolio piece.
Then, I ordered the comments by the name of the commenters to eliminate rework for the next step. Following goal-directed design, I mapped the users (as well as I knew how, from their publicly available information) onto several dozen behavioral variables. These included:
* Their interest in an attraction's theme, thrills, or both
* Their experience working in themed entertainment
* Their interest in newer or less common kinds of themed entertainment (e.g. escape rooms, immersive theater)
* Their frequency of sharing content on Facebook
Color-coding the behavioral variables via conditional formatting made it easier to detect patterns in the data.
Some of the behavioral variables, such as levels of interest in particular subjects, represented a sliding scale. To allow calculating an average of these for a particular set of users, I converted these variables' values to numbers.
This tab shows one of many filters I applied on the behavioral variables as I analyzed different users' cases.
Because there were 43 behavioral variables and over 150 comments to analyze, a full cross-case analysis would have taken far too much time. I chose to focus on these three variables for my preliminary cross-case analysis:
* Whether or not they have worked in themed entertainment: past, present, or future.
* How well they know the themed entertainment or amusement industry: This allows capturing nuances like people who work in amusement parks but not in a role that gives them much knowledge of their businesses or the rides from a technical perspective. It also allows me to differentiate between people who work outside the amusement industry and either understand the rides very well or do not.
* The main park chain or attraction type addressed in their comments.
This shows the interests and goals that one segment of the audience tended to have.
This shows how commenters with a high degree of amusement industry knowledge tended to score on other variables related to themed entertainment. Since escape rooms only recently became popular, their interest longer-term tended to skew low among all of the user groups. I expect that interest level to rise as more people experience escape rooms.
This shows some more personality-related behavioral variables for one of the segments in the audience.
For determining which hobbies personas tend to have, I often had insufficient data (again going by what people decided to disclose publicly on social media). So rather than deciding to filter for each one, I took a higher-level approach that merely would say later what hobbies people in this audience tended to have. It was fairly safe to say for this user group that they were casually interested in sports because of a moderate level of interest that I knew from over half of the users.
The personas were now ready to group into a spreadsheet. At this time, I had not yet fleshed out their goals and frustrations more specifically. But I could see patterns emerging. There are additional notes in many of these cells because I wanted to make sure that the persona set properly captured differences in the user base.
This is the original behavioral variable set for the initial persona set above.
I discovered quickly that the group of people commenting about Disney attractions, which represented over half of my initial data set, was definitely not a monolithic group. It was not possible to condense them into one persona.
I considered different ways of splitting this group. Ultimately, I chose to split it based on how much each group expressed that they wanted to spend on their trips to Disney attractions. Although I only had this data for 1/4 of the overall Disney commenters, the rest are accounted for in other personas.
My persona set has separate life goals, end goals, and experience goals, again following a best practice of goal-directed design. These directly stem from the behavioral variables but just capture the same information in a way that better addresses the main points about each group of users.
This next iteration just shows more of the ambiguities between different personas decided.
In determining the primary audience, goal-directed design then involves checking which groups of personas would get their needs met (or specifically not met) by an interface that addresses the needs of others. This involved comparing every pair of personas against each other. This screenshot shows one of these comparisons.
Due to how many variables there were, I color-coded differences that may or would cause one group's needs to go unmet in trying to meet the needs of another group.
This matrix shows the end result of all of the pairwise comparisons between personas.
This tab showed the personas grouped according to who might be the primary persona and who could not be. Until now, I had left the Disney Commenter persona in the spreadsheet.
Next, I drafted persona narratives for each persona that remained and added them to a Microsoft Word template.
The above personas were generated using the same type of process I use to generate personas for my clients. However, due to non-disclosure agreements, I may not post those personas in my online portfolios. Therefore, I have posted this project as a sample.
I believe that user experience design, done correctly, has its roots in user research techniques, such as the ones I used to develop these personas. The first step of creating an effective digital product that leads to a solid ROI is to understand your audience; then make sure you are solving the root cause of their problems. If this project resonated with you and you are considering having something like it done for your website, app, or software product, please feel free to contact me.
