I’ve seen a lot of dashboards fail
Despite the fact that books and courses on information dashboard design have been available for years, many dashboards still fail to meet users' and organizations' expectations. Users have trouble finding answers to basic data-related questions and fail to notice urgent problems because they’re hidden behind clicks, hard to notice, or possibly not even on the dashboard. Because of these and other problems, many dashboards still end up under-used or even abandoned.
Based on Nick Desbarats’ experiences designing dashboards for over 50 large organizations and teaching dashboard design to thousands of professionals, Beyond Dashboards uncovers the real reasons why so many dashboards fail to satisfy users and organizations; reasons that go far deeper than the visual design on which most dashboard books and courses focus. Readers will learn a practical, actionable framework for creating a system of purpose-specific displays (problem-scanning displays, metric diagnostic displays, performance reporting displays, etc.) that enables users to get answers to their data-related questions quickly and easily, including basic ones such as, “Is everything O.K. at the moment?”, often for the first time.
Who should read this book
Beyond Dashboards provides specific strategies, frameworks and best practices for those who are directly responsible for designing or developing information dashboards for employees, partners, stakeholders, and/or customers of their organization. Target readers include business intelligence professionals, data analysts, reporting managers, software developers, user interface designers, and similar roles. The strategies and frameworks in the book are sector-agnostic and are applicable to organizations in the finance, manufacturing, technology, health care, banking, insurance, government, military, non-profit, education, and most other sectors. Senior managers and other decision-makers who consume dashboards will also find the book to be of value since it enables them to ask for dashboard designs that help them to be more effective at their jobs. The book does not assume any specific prior technical knowledge.
Major topics include:
Untangling the term “dashboard”
The 12 fundamentally different types of information displays that are, unfortunately, all called “dashboards”
The three high-level groupings in which the 12 types of “dashboards” fall:
Enterprise information displays for enabling an organization’s employees, partners, customers and other stakeholders to interact with the organization’s data (seven display types)
General information displays for engaging, persuading, and educating audiences (four display types)
Glance displays for real-time monitoring (one display type)
Note that only enterprise information displays are discussed in detail in the book.
The seven types of enterprise information displays:
Scanning displays for enabling users to quickly spot metrics that are underperforming, overperforming, or behaving abnormally among potentially thousands of metric values
Barriers to rapid scanning: 24 causes of hidden gotchas, false alarms, haystacks and decoys
Determining which metrics to include and exclude on scanning displays
Segmenting scanning displays by user role and review frequency (daily, weekly, monthly, etc.)
Why conventional ways of determining which metrics to flag on dashboards are surprisingly problematic:
“% change since previous period”
“% deviation from target”
The “four-threshold” method for flagging metrics that require attention
Using simple statistics to detect and flag problems among thousands or even millions of metrics and metric values
Scanning on mobile
Examples of well-designed scanning displays
Diagnostic displays for quickly understanding why a given metric is underperforming, overperforming, or behaving abnormally
The five types of information on diagnostic displays that enable a metric’s behavior to be diagnosed 90% of the time:
Child metrics, peer metrics, metric history/forecast, influencer metrics, related metrics
The diagnostic cascade: Enabling users to quickly diagnose 90% of problematic metrics on their own, saving analyst resources for the 10% of problems that require advanced analytical skills to diagnose
Examples of well-designed diagnostic displays
Performance reporting displays for assessing and improving organizational performance in light of strategic goals
The important differences between scanning displays and performance reporting displays
Examples of well-designed performance reporting displays
Common pitfalls when selecting performance measures (KPIs) and analyzing performance (brief overview only)
Note that general performance improvement processes such as strategic planning, goal setting, KPI selection, etc. are not discussed in depth
Slice-and-dice displays for browsing and filtering large datasets of customers, transactions, employees, etc.
Best practices and common design mistakes
Examples of well-designed slice-and-dice displays
Item browsing displays for spotting items of interest among a limited number of items (usually fewer than 50) such as projects, portfolio investments, hospitals, etc., based on multiple criteria (budget deviation, wait times, beta, etc.), usually in a tabular layout
Best practices and common design mistakes
Examples of well-designed item browsing displays
Item detail displays for displaying detailed information about a single item (a student, a transaction, a website, etc.)
Best practices and common design mistakes
Examples of well-designed item detail displays
Canned analysis displays for enabling non-analysts to use complex analyses that have been packaged behind a simple user interface (what-if simulators, scenario evaluators, etc.)
Determining which types of analytical questions can be safely answered by non-analysts using preconfigured tools, and which types of questions require a trained analyst to answer
Examples of canned analysis displays are shown but design principles and best practices for this type of display aren’t discussed in depth since these vary greatly from one display to another
Navigation and discoverability
Designing a home screen that enables users to quickly find answers to different types of data-related questions
Linking different types of information displays into a cohesive system that enables users to quickly get answers as new data-related questions arise in their minds
Educating and gathering requirements from users
Review of a typical, complete set of information displays based on the Beyond Dashboards framework
How and where to start
Maintenance and evolution
Topics NOT covered:
How to use specific software products to create dashboards. The frameworks and best practices that are recommended in Beyond Dashboards are software-independent.
How to create visually impressive dashboards. Only enterprise information displays are discussed in detail, and these are recommended to have a plain, minimalist visual design. General information displays, which can have visually impressive designs, aren’t discussed in detail.
Performance measurement and improvement best practices (organizational and personal goal setting, strategic planning, KPI selection, etc.), though books on these topics are recommended. Examples of well-designed performance displays are featured, and the book discusses how performance reporting displays fit into an ecosystem of information displays.
“Glance displays” for real-time monitoring such as car dashboards, status displays on industrial equipment, etc. A few examples are shown but these types of displays aren’t discussed in detail.
Taking a page from one of my favorite authors (pun intended), Dan Pink, I’m going to attempt to crowd-source the editing of this book, and that’s where you come in. In the coming months, I’m going to be posting short excerpts from the new book as blog posts and asking for your feedback. If you hang around long enough, you’ll eventually see all of the major ideas in the book before it’s published, but I know that you’ll still buy it anyway so I’m not worried ;-) If you think that any of my ideas are off-base, even the high-level ones that I’ve discussed here, don’t sugarcoat. Tell me what’s what. That’s exactly the kind of feedback that I’m hoping for.
(To leave a comment about anything on this page, please do so on this identical blog post.)