MultiCare Provider Search

Client
MultiCare Health System

Role
Lead UX Designer / Co-Researcher

MultiCare-Foundations-Color 1 (1).png

Project Overview

MultiCare had implemented a feedback collector on their website asking users if they were able to find the information they were looking for. Our team noticed a trend that on doctor search pages users were only succeeding at a rate of 41%. Knowing a key goal of MultiCare was connecting patients to doctors, we decided to dive deeper into this issue.

Team goal: To allow both patients and doctors to find a provider that fits their needs on MultiCare's online search.

Process for this project

Group 187.png

Foundational research

We knew the success rate for the doctor search was a low 41%, but we needed to figure out why. To do so, I worked closely with the UX researcher to learn exactly how it was being used and what was going wrong.

Methods

  • Analyzed search analytics

  • Interviews with doctors

  • On-site surveys

  • Stakeholder workshop

  • Baseline testing

Research Goals

  • How are patients currently using the search? What about the providers themselves?

  • What common pitfalls do users fall into when using the current search?

  • What filters are most useful to users? How do patients zero in on the provider they want to go and see?

  • What key features were missing from the current design?

20180509_130318.jpg
20180514_211636.jpg

Designing a solution

Through our foundational research, we saw a few key opportunities to enhance the experience:

 

Confusing filters

The problem
During our baseline testing we found users often felt the need to fill out every field presented, often resulting in no results being returned. On top of that, many of the existing filters were confusing, such as a filter for “locations” (which actually meant facilities) that came before city (which was the zip or city they were looking in).

  • Users felt they needed to fill out every drop down presented

  • Locations vs cities was led to confusion in users

The problem: The old search function with examples of what’s below Locations.

The problem: The old search function with examples of what’s below Locations.

The solution
To help alleviate this problem, we designed the system such that users perform a broad search on specialty or name, and then are later presented with options to filter down results.

 

Typos and synonyms

The problem
Patients were frequently using search terms such as “GP” for “General Practitioner” resulting in failure.

The solution
We advised MultiCare to create a list of synonyms and common misspellings of words in order to bolster this find as you type search, captured by reviewing search analytics and from our interviews and usability testing. Using this data, we implemented a find as you type search that displays the specialties we assumed they were aiming for with their search.

 

Examples from the prototype I created of how a search-as-you-type greatly improved the experience.

Prototyping

After designing the first iteration of our solution, I built a prototype so we could test it on both mobile and desktop.

Because the find as you type functionality as well as the search logic needed to be somewhat accurate in order to test properly, we knew a simple click through prototype (like one that Figma or Invision could provide) wasn’t going to cut it. Due to this, I ended up using Axure to make a complex and in-depth prototype of the new system.

Try out the final iteration of the prototype

Prototype Features

  • Working search as you type text field

  • Real search logic based off of limited data

  • Adaptive, resizes from desktop to mobile

 

Evolution of the prototype from low to high fidelity.

Testing

RITE Method
To validate our solution we tested the prototype out on patients, as they are the primary audience in MultiCare’s organizational goals. We utilized Microsoft’s RITE method to find and fix issues, then validated with the next batch of users. This gave me experience with performing many quick edits and iterations on the fly, which I was able to execute via use of masters and repeaters throughout the prototype.

A Mid-Test Addition
Halfway through these tests, we met with the client only to learn a function we thought was off the table could now be implemented. This was the option to highlight if a doctor was accepting new patients or not, a key feature that was mentioned during our user interviews. We implemented this feature into our living prototype, resulting in positive reactions from patients.

Delivery
Once we were done testing and we got the go ahead from the client, I worked closely with the graphic designer and developer to ensure the experience came through in the final product

 
Screen Shot 2021-04-11 at 5.38.13 PM.png
Screen Shot 2021-04-11 at 5.04.15 PM.png

Follow up

Once we got a working development version going, we did one final round of tests to make sure the live site showed the same success as our prototype.

  • After testing the live version, the average task success rate went from 50.0% in baseline testing to 86.11%.

  • After a period of months our feedback collector showed an increase from 41% self reporting they found their doctor, to 71%.

Group 105.png
Previous
Previous

Washington State Historical Society Collection Search

Next
Next

YMCA Class Registration