Dark Metrics: Illuminating the Negative Impact of Digital Health Design
Summary
Traditional design metrics and KPIs are often geared towards measuring product success. Dark metrics challenge this paradigm by proactively measuring the unintended yet harmful psychological, social, and physical effects of our technologies. The examples within digital health are plentiful. From accelerating burnout among clinicians to widening racial disparities in quality of care, we can only reach the height of our most courageous solutions when we expose our deepest failures.
Key Insights
-
•
Traditional product metrics, like the Google HEART framework, often miss broader impacts on users, focusing narrowly on product success rather than holistic well-being.
-
•
Dark Metrics is a framework designed to measure negative unintended effects in digital health across four dimensions: disempowerment, exclusion, addiction, and distraction.
-
•
Disempowerment occurs when technology removes users’ autonomy, such as opaque black-box AI systems that undermine clinician or patient decision-making.
-
•
Exclusion can be subtle, as algorithms that proxy biased variables like healthcare cost can reproduce racial disparities without explicit intent.
-
•
Racial equity in design can be assessed using heuristics or rubrics co-created by diverse teams, as demonstrated by Raven’s IBM colleagues Dre Barbara, Sherees Cooper, and Morgan Foreman.
-
•
Addiction to technology is an overused concept in consumer health, but distinguishing healthy from excessive use requires linking engagement data to well-being measures.
-
•
Distraction from core tasks is common in clinical environments when new tech disrupts workflows, evidenced by studies with ER staff and clinical trial recruitment tools.
-
•
Ethics frameworks like the Institute for the Future’s Ethical OS help anticipate risks like surveillance, bias, and data control, which inform Dark Metrics design principles.
-
•
Engaging diverse stakeholders and including co-creation early in research helps uncover biases and unintended consequences before launch.
-
•
Addressing negative impacts requires transparency with clients and a strong ethical posture, even when business priorities may conflict with user protection.
Notable Quotes
"The traditional product metrics focus narrowly on the product or near-term impact but fail to capture what success means for the whole person."
"Within IBM Watson, we prefer the term augmented intelligence rather than artificial intelligence to emphasize support, not takeover."
"An AI algorithm that didn’t explicitly consider race still produced racial disparities by using healthcare costs as a proxy."
"I am a Black person, but I do not have every Black experience. Not having experienced something is not proof that it doesn’t exist."
"The difference between technology and slavery is that slaves are fully aware they are not free."
"Doctors want to help people, not be on a machine all day; many health technologies are more distracting than helpful."
"We can assess distraction by observing time spent on screens versus with patients, and self-reported mental effort and stress."
"It’s important to ask, before any new release, what are all the things that could possibly go wrong?"
"Our jobs are to protect users from harm. If clients don’t care about side effects, it may be necessary to draw a line and walk away."
"Storytelling is highly effective in helping stakeholders understand the complete performance of products, including the darker sides."
Or choose a question:
More Videos
"Hypertext notes coexist on the same level and link to each other arbitrarily, creating emergent structures."
Jorge ArangoExploding the Notebook: How to Unlock the Power of Linked Notes (2nd of 3 seminars)
April 19, 2024
"Recruiting diverse partners takes time and effort, and we are actively building programs to include people with diverse abilities."
Tracy McGoldrickIBM User Experience Program—The What, Why and How
October 15, 2021
"IKEA's instruction manual is rigid; you must do exactly what you're told to get a specific result."
Michelle MorrisonPractice What You Preach
January 8, 2024
"If everyone’s looking at different versions of the plan, you get frustration and confusion."
Meaghan Waters Fotina KoutropoulousLack of Product Thinking will Doom Your Legacy Modernization
June 9, 2021
"Collusion is this dynamic process where both parties, through negative interpretation and reaction, perpetuate the conflict."
Sylvie Abookire Susan Abookire Caitlyn NalderA Civic Designer's Guide to Mindful Conflict Navigation
November 17, 2022
"Chatbots aren’t just the language model—they have structure around intent classification, safety checks, and session memory simulation."
Peter Van Dijck Louis RosenfeldCoffee with Lou #4: Taking a Peek Under the Rosenbot's Hood
June 14, 2024
"Every speaker before me had these pristine backgrounds; I’m a slob and feeling self-conscious about it."
Dan WillisTheme 3: Intro
January 8, 2024
"Designers often don’t realize they themselves are providing services and selling a promise of value."
Shreya Dhawan Victor Udoewa Xènia Viladas Florian VollmerMaking service tangible: the fastest path to higher performance
November 19, 2025
"Cultural differences mean improv expectations vary widely—what works in the U.S. might not in Italy or Africa."
Cheryl PlatzMerging Improv with Design
March 7, 2019
Latest Books All books
Dig deeper with the Rosenbot
What role can familiar communication channels like WhatsApp play in deploying AI tools in resource-limited settings?
What new workflows or tools have been effective in bridging the gap between design, engineering, and product teams using AI?
What skills are essential for designers to communicate effectively with AI to achieve meaningful outputs?