IMG_3441.jpg
IMG_3441.jpg

PI System Health


PI System Health

SCROLL DOWN

PI System Health


PI System Health

PI System Health

Product

The Health Monitoring Web Application leverages real-time data and enables administrators to dynamically monitor the health of their PI systems, allowing them to feel secure about their operational data. 

Challenges

  • Delivering useful and appropriate information to users when they need it the most.

  • Designing appropriate metaphors, interactions, and appearances.

  • Prioritizing and managing requirements for a large-scale project.

  • Working across multiple development teams to ensure the product has a seamless experience for our users.

  • Demonstrating the value of UX design on the first project to have an embedded designer

 

Team and Responsibilities 

I acted as a design lead, primary designer, and researcher. The team comprised of a product manager, development lead, development team (4), and a QA engineer  

My Key Responsibilities were to:

  • Gather insights from User Research and Analysis

  • Translate research into actionable insights for team

  • Create ideal journeys and wireframes based on research results

  • Communicate user needs to build empathy amongst cross-functional team

  • Provide designs, strategy, and user research throughout the agile process

  • Usability Testing with both external and internal users

  • Facilitate collaborative discussions involving designers, developers, and managers

  • Manage design specs and requirements

  • Communicate design rationales with executive and product teams


Research: Understanding User Needs

At the beginning of the project we were given the initiative to improve the manageability of the PI System. The data that is held, visualized, and analyzed within the PI System is critical to operations for our customers, making it critical that their PI systems are reliable and functional. 

From previous research we identified that management of the PI System was a large pain point for our users. The first task we had was understanding what specifically was difficult. 

Card Ranking

Users were asked to use a card ranking exercise to help us prioritize the areas of PI Management that needed the most improvement. The number one priority for administrators was understanding the health of their systems. 

 

User Requirements

After analyzing the research, we outlined user requirements, and organized them by topic and priority. The core requirement was that users needed a quick way to assess and manage the health of their system. 

 

Word Cloud

A quick word cloud from the interviews showed that (1) users wanted a single tool for management and (2) health was a critical component of this tool. 

 

Research Findings 

  • Most of the administrators maintain other systems outside of PI and therefore have limited time to spend on PI.

  • Administrators have limited visibility into the health of their systems and incidents largely go unnoticed until they escalate or cause major issues.

  • Each product has its own independent management tool. There is no central way to manage the products collectively.

  • Customer systems are often broken up into smaller groups with limited cross-communication, making solutions difficult to implement company-wide.

  • Administrators must rely on past knowledge of an issue. If it is a new issue, they rely on technical support.

  • Many users have created their own documentation for troubleshooting.

  • New administrators often feel lost when troubleshooting and do not know where to begin when an issue arises.

  • Previous attempts at health monitoring have failed due to nuisance alerts and general unreliability.


Wireframes: Translating the Research

Taking what I learned from the research, I then moved into ideating on features and solutions to address the pain points we uncovered. 

Simplify the Administrative Experience by

1. Limiting the number of admin touch points (less tools)

2. Provide an easy method of managing assets and components

3. Reliably provide health and guidance of the PI system from a central location 

4. Reduce the amount of setup time, and provide smart default states

Current Pain Points

1. Too many tools 

2. No Ability to navigate between tools 

3. No standard for installation

4. Inconsistent behaviors across applications

5. Difficulty troubleshooting: identify, understand, resolve

6. No visibility into the health of the PI system 

7. No pre-built displays, analysis, or notifications 

 

Sketching the Flow

Beginning with paper and pencil, we outlined general flows and touch points that a user might take in a health monitoring tool. 

Our users needed to monitor the health of their system data flow, assuring that the data from their sources effectively made it into the data historians reliably and in a good state. Our second goal was to provide useful tips and guidance when an incident arose.

 

The screens in the sketch represent a data flow diagram with health indicators, dashboards of each component, and incident reports with helpful guidance.

Low fidelity mockups created in Sketch.

 

Sketches in a Higher Fidelity. 

The initial designs covered component health indicators and incident reports, however one critical aspect we had not designed for was the "single-pane of glass" system overview. That is, one screen to represent our users' entire system that could scale from a small system all the way to an enterprise system. 

Users can be spread out geographically. One approach to the single-pane of glass overview was to have a map with pins that represented a system. We initially explored a data flow diagram, however it did not scale for our users with larger systems, and in future iterations our health monitoring tool would include components that are nonessential in data flow. 


Iterative Testing: Are We on the Right Path?

Testing Early

We took our early concepts to users to get their feedback, and gauge if we were on the right path. We visited customers on-site and during our user conference so that our users could be within a natural context for feedback. 

Customer Testing Session in Europe

Customer Testing Session in Europe

 

Cross-Cultural Evaluation

Our sales team noticed low sales with visualization products in European and Middle Eastern markets. The conclusion was drawn that our products were not being designed to the quality that was expected in these markets. To address this pain point, we conducted user testing in these markets and made sure our designs would fit the needs of the users that had previously been unsatisfied. 

Findings from Testing

  • The map is not the best visual for users - We needed to think of a more informative visual for a high level overview.

  • The biggest benefit the tool provides is the visual relationship between components (understanding the issue and it’s impact, as well as the ’where’) and the guidance (how to resolve the issue).

  • Users need to know what happened when the incident occurred. What is the context to the incident?

  • Users want to know if the incident has occurred before, and how it was resolved.

  • We needed to update the navigation so that it is more intuitive and matches the users' expectations.

I have everything I need on one screen, I don’t have to look at multiple places to get to the same conclusion.
— Customer

After the initial round of conceptual testing we uncovered that we needed a different visual to represent a system. We explored:

  1. Treemaps

  2. Network diagrams

  3. Flat list

  4. Sunburst

The sunburst visual provided a single view of a hierarchical model in a concentric space. It allowed a user to see their system at a glance and drill in to see specific segments and their relationships. 

IMG_2948.jpg
 

Usability Test with Clickable Prototype

Clickable prototype created in Principle. Download file here.

Before

  • Incident list grouped together with tree view of system.

  • Flat list and user-defined list of components separated by tabs.

  • Badge for incidents appears near active incidents near the bottom-half of the screen.

After

  • Added toggle to switch between an organized view and a flat list of components. The toggle provides a way to easily translate the information from flat to organized/hierarchical views.

  • Added breadcrumbs to give users context of where they are within the sunburst.

  • Removed incidents from the dashboards and placed them into their own tab to distinguish between a detail page and the notification/incident page.

Below are a few Health screens after a few rounds of user interviews and usability testing. 

PI System Overview, the home page of the PI System Health tool. Users can see at a glance how many incidents there are.

Component Overview Page. Each component within the PI System has its own dedicated page with health indicators, guidance, and related components.

Incident template. Users will be notified of an incident with information on what the issue is, where it is, what is its impact, and how the issue can be resolved it.

Improved out of the box experience

In order to alleviate the pain point of "no default states in the current experience" we designed the system to automatically create displays, analysis, and notifications. Users can edit these default states and subscribe/unsubscribe to notifications so that they can tailor the health tool to fit their needs. 


The Big Picture: Integrating with Enterprise Suite

One of the main pain points for users was accessing configuration and all other administrative applications easily. As we shifted from on-premise installations to web applications, we had the opportunity to provide proper touchpoints between our applications. The health tool was one of the first of the web applications and thus we had to assure we understood how all other applications fit together. 

 

Journey Map

I created a journey map of health integration with other administrative products. We used this visual to convey the need for providing proper entry points within the different administrative tools such that users can flow easily from task to task. 

Global Menu

We designed a universal menu that provides entry-points to other administrative tools from within our main web applications. We also organized the different administrative tools by task to help inform users of the purpose of each tool. 

Smarter Touchpoints

I provided entry points to other applications when it makes sense, in context. An example of how we provided links to other applications was by leveraging the details pane in the health tool. We used descriptive text of what task you could do in the other tool.


Meaningful Guidance

Another pain point our users faced was "lack of clear, and meaningful guidance". Often times users felt that they had to rely on previous knowledge of an incident, or escalate to contacting technical support in order to resolve an incident that affected their system. 

We need to provide our users the right information at the right time. 

Cross-Departmental Collaboration

To alleviate this pain point we worked with our customer support engineers and product experts to create a library of incidents and product resources that we could use in the health tool. Each type of incident received an overview page with guidance on what the issue is, where it is, and how to resolve it. We also included a section that allowed users to document comments and resolutions to incidents; these comments will be historized for future reference. 

Users are likely not sitting at a desk monitoring their systems. From research, 100% of users rely on notifications in some way to know if theres an incident. The health tool automatically creates notifications and analysis for a system that a user can then subscribe to and edit according to their needs. 


Onboarding: Getting Value Right Away 

First impressions are long-lasting. 

Our goal is to minimize set-up time and provide our users with a walkthrough of general features and How-Tos so that they can get value out of the tool right away.

The health tool automatically creates dashboards and incident notifications so that the user only has to open the tool and gain familiarity with new and key-features.

Limit the friction, increase the value. 

Simple overlay walkthrough highlighting key-features. Users can interact with visuals as they explore the new tool.


Visual Design  

A majority of our users are in front of control systems and dated software. Initially our designs were dated and gray-scaled because they matched the software our users used and we did not have a company standard for visual design. With this project, we made it a goal to set the standard across our company and establish a clean, modern visual design that would fit the requirements for a dashboard.

A dashboard needs to be kept simple, only using stop-light colors for pertinent information. 

Design for Accessibility  

We are working towards having all of our designs fit within accessibility requirements. Our first focus is designing the health tool (and all administrative tools) to be colorblind compliant. We use symbols to denote incidents in addition to the traffic light indicators so that users do not have to rely solely on colors. We are currently working towards using colors that are compliant as well. 

Using a sketch plugin to test designs for colorblind compliance.

Communication with Development Team 

We used Zeplin to capture comments and edits of the health screens. We also used Zeplin as a library for styles as no such library existed at the start of the project. 


Continuous User Feedback

At the beginning of the project, our users were thrilled to be a part of the user feedback sessions. This project was the first effort that included user feedback throughout every stage. We continue to talk to users and keep in contact with the ones that we have worked with from the beginning in order to show them our progress and continuously get feedback.  We also included new users throughout every stage to have variety and assure we are talking to users that represent all personas.

By getting continuous feedback through research (usability, conceptual, etc) we are assuring that we are listening to our users. We can fail early, often, with less risk, and assure we are delivering a valuable product. 


Summary

The PI System Health Monitoring tool was designed to help our users use and trust their PI System. The biggest pain points we designed for in V1 were:

  1. Lack of guidance. Help a user answer "where is the issue, when did it occur, how can I resolve it?".

  2. No central management. Reduce the time for configuration by automating processes during installation. Embed touch points for different management tools in context. Create universal access to all management tools from a central location.

Our next step is to continue working on a central administration portal that helps users keep track of all administrative tasks. Such tasks include configuring security for their systems and configuration of administrative components. 

Highlights:

  • First project at OSIsoft that used embedded UX Design

  • Designed and delivered a Health Monitoring tool that allows administrators to monitor the health of their PI System in real time

  • Delivered useful guidance and tips to users within the tool and in context

  • Worked on a cross-functional team that used agile methods

Home Page in System Health. System Overview

Example of a component detail page. Health Indicators and statuses provided for every individual component and user can see where component falls in the data flow stream.

Example of Incident overview. What is the issue, where is the issue, how do I resolve the issue. Every incident is saved along with notes and comments for reference.


Related Work