For design to work, you need to get multiple stakeholder types on the same page. After all, design is a team sport...but everyone also has an opinion. How will you manage expectations and keep making progress? Learn:

• Identify additional stakeholders who may need to be involved in your solution to drive adoption
• Ramping up your domain knowledge
• The role of empathy in understanding your business stakeholder's perspective and how to ensure you're providing what users actually need (and not just want)
• Understand who to put in your inner ring and outer ring team, and what to watch out for when interacting with different disciplines (e.g. product/biz, data, engineering, design, SMEs, sales, etc.)

Designing a successful solution begins with having the right conversations with the right people. Unfortunately, you may be faced with opaque or vague requests from a business stakeholder to "use AI on this project" or "solve it with machine learning." In this module, we'll learn how to:

• Turn vague requests for "AI" from business stakeholders into actionable business problems that are worth solving
• Help you build credibility and trust with stakeholders
• Define all of the key people (core users, downstream users, stakeholders, obstacles, and partners)
• Learn to conduct qualitative research using the power of Empathy and Why questioning to illuminate the different pains, problems and obstacles faced by these people
• Creating a research plan (questions, target users, and learning objectives) and key factors to consider when the solution is likely to utilize machine-learning / predictive analytics

This is where we turn our research into an actionable design brief that your team and your customer can all see and agree to.  You'll learn to: 

• Develop an initial design brief for your solution summarizing the goals, metrics, and values your team and your stakeholders have agreed upon. From here, your team will know what successful outcomes look like before you start building anything.
• Decide whether to benchmark or not (so you can measure the impact of the solution later on)
• Learn to separate tactics from goals so that outcomes, not assets, drive the project
• If machine learning is a likely technology to be utilized, we will learn to consider additional stakeholder roles to consider involving including raters, review panels, ethicists
• Considerations for high-stakes vs. low-stakes solutions - what considerations need to be given if you're using AI

A human-centered data product or solution is rarely just a dashboard. Chances are, your solution may require touching multiple applications and people before it can become a valuable decision support tool. A journey map helps uncover all of these touch points so you can understand what opportunities exist for innovation, and what non-technical blockers may prevent your solution from getting adopted by the people who will ultimately determine its success. We'll also be looking for ways to cut scope, and deliver value sooner. You'll learn to:

• Understand your customer’s workflow today
• Identify new (and old) places in the customers' experience where data-driven solutions may be able to create impact.
• Uncover places where change management will be required–before it blocks adoption of your solution.
• Learn how to get the support you need when your solution depends on factors outside your immediate control
• Revisit your original plan when new problems/opportunities emerge that could allow you to deliver a smaller, but different solution more quickly. (This is in the spirit of MVPs and quick-wins that let you prove the value of your work sooner to customers and stakeholders.)
• Plan for onboarding of your users and how you'll ensure AI/ML-driven products/solutions are understood by customers who may not trust or be used to AI.

The design jam is a working session with your teammates and partners that focuses on using whiteboard and sketching to clarify ideas, needs, and UX solutions. This is where we turn design thinking into design doing.  Some of the activities we will discuss include:

• Ideation - With your team, learn to generate multiple solution ideas using non-analytical divergent thinking. Don’t worry, it doesn’t hurt! This will:
  • Prevent you and your team from jumping into the solution space too soon
  • Learn to balance speed and showing value soon (MVP mindset) vs. the giant moonshot project
  • Techniques to generate creative ideas when working backwards from existing data
• Map the workflows required in your tool to uncover what design assets will be required to bring your UX to life
• Machine learning data collection - how does the desired UX affect the data that is required/needed to produce a feasible model?
• Machine learning algorithm design:
  • How will you tune the model? For recall or precision?
  • What does the feedback loop look like with users?
  • Is the data rater's UX aligned with the project's needs?
• Gathering artifacts - using our prior journey map, determine the screens, services, notifications/emails, and integrations with other software that may be required based on what is actually needed (not from a business sponsor's requirement document!)
• Determining what modalities may work best for your data solution (phone, tablet, desktop, SMS, etc.)

This is the phase most people think of when they think of "design." Learning to design "transparent" interfaces that focus on simplicity, maximum signal to noise ratio, and help customers to keep moving forward to the next step. Remember, the last-mile of any analytics or data science solution is the part that the people interface with. This can make or break a solution, no matter how powerful the analytics or predictive power of the model is. We will:

• Learn to apply my CED Framework for integrating advanced analytics into decision support software. We will discuss how to expose transparency and model interpretability in your designs to build trust with users.
• Get Brian’s cheat-sheet for designing effective high-fidelity user interfaces
• Designing for different states including:
  • Empty/no data
  • Low-confidence predictions
  • High-confidence predictions
  • Errors
  • Balancing model transparency with information overload
  • Honeymoon considerations
• Learn how mobile-first mentality can accelerate and simplify your data product
• Interpreting likes vs needs, balancing speed with quality
• Critical factors when designing data-rich prototypes used in user validation
• How branding, visual design, and aesthetic choices impact your product, data, and insights
• Get an overview on the principles of Universal Design (accessibility) - designing solutions that consider visual and other impairments users may have.

Learn how to properly evaluate your design or product with real customers and your team.  You'll learn to:

• Run a productive design review or critique with your team
• Uncover ways to test customer or business stakeholder reactions to data model predictions prior to building out any technology solution
  • How much user control will be needed with an AI solution?
  • Can we properly conduct realistic tests of the solution (i.e. high stakes scenarios vs. low stake?)
• Plan and conduct a usability study and use it to get team/stakeholder buy-in when they aren't "believers" in your solution
• Determine how much to test before you ship, who to test, and how to interpret results so you know how the product needs to be adjusted
• Feedback collection: understanding when to design for and capture explicit feedback vs. implicit feedback to improve model quality

You learn all this stuff. You go to work. You apply it. And then what? Does it stick with you? In this module, we will talk about "next steps" so that you can hopefully begin to teach and proselytize what you have learned.