01
Intent
What are people trying to accomplish?
- Interviews
- Search terms
- Jobs to be done
- Support questions
Intent reveals the goal behind an action. A click without intent is only movement.
Evidence / judgment / learning after launch
Good design learns.
Data-driven design gives a product memory. It connects what people say, what they do, where they struggle, and what changes after a design decision reaches the world.
HAAM / EVIDENCE SYSTEM 01
Signals become useful when they change a decision.
Observe / interpret / intervene / learn
00 / POSITION
Data gives design memory.
Every interface begins as a hypothesis about people, context, and value. Research improves the hypothesis before release. Measurement reveals what happened afterward. Design judgment connects the two.
A metric can reveal a pattern, but it cannot decide what deserves to exist. Data-driven design works when evidence sharpens responsibility, expands perception, and helps a team learn without surrendering its judgment.
01 / EVIDENCE
One metric is never the whole experience.
Strong product decisions combine several layers of evidence. Each layer answers a different question, and each corrects the blind spots of the others.
02
Behavior
What do people actually do?
- Journey data
- Interaction events
- Navigation paths
- Feature adoption
Behavior shows where the designed path and the lived path begin to diverge.
03
Friction
Where does effort, doubt, or failure appear?
- Drop-off
- Error states
- Repeated actions
- Usability observation
Friction is often the most useful signal because it reveals where the system asks too much.
04
Outcome
Did the experience create meaningful value?
- Task completion
- Retention
- Comprehension
- Support reduction
Outcome keeps the team focused on what changed for the person and the organization.
05
Inclusion
Who succeeds, who struggles, and who disappears from the average?
- Accessibility
- Device and network
- Language
- Segment differences
Averages can hide exclusion. Good measurement makes uneven experiences visible.
06
Time
What happens after novelty wears off?
- Cohorts
- Repeat use
- Recovery
- Long-term trust
Time separates a persuasive launch from a product that remains useful.
02 / LEARNING LOOP
A product should become more intelligent through use.
The loop is the real capability. Research, analytics, experiments, and prototypes are instruments inside it.
01
Frame the decision
Name the product decision, the user outcome, the business outcome, and the risk of getting it wrong.
02
Build the evidence trail
Combine qualitative research, behavioral data, operational signals, accessibility, and technical performance.
03
Find the uncertainty
Identify what the team still does not understand. The best experiment reduces a consequential uncertainty.
04
Design the intervention
Create the smallest change capable of producing a meaningful difference, then define the guardrails around it.
05
Observe the whole effect
Measure the intended outcome together with side effects such as confusion, exclusion, pressure, or slower performance.
06
Keep the learning
Document what changed, what was learned, and what should happen next so the product develops a memory.
03 / BETTER QUESTIONS
The question determines the quality of the data.
Narrow optimization questions produce narrow answers. Better questions connect performance to intent, trust, inclusion, and the wider journey.
Onboarding
Weak question
How do we increase completion?
Stronger question
Which step creates uncertainty, for whom, and can we remove it without reducing informed consent or future success?
AI product
Weak question
How do we increase AI usage?
Stronger question
When does the AI improve the task, when does it create doubt, and what correction or fallback does the user need?
Ecommerce
Weak question
Which button converts better?
Stronger question
What information helps a customer make a confident decision, and does that confidence survive delivery, use, and return?
Public service
Weak question
How many people reached the final screen?
Stronger question
Who completed the service independently, who needed help, and which requirements created avoidable exclusion?
04 / FAILURE MODES
Data can make weak thinking look scientific.
Measurement becomes dangerous when precision replaces understanding. These failure modes are common because each can produce convincing dashboards while degrading the actual experience.
01
Vanity metrics
Numbers that look active but do not explain whether people completed a meaningful task or received lasting value.
02
Proxy capture
A measurable proxy becomes the goal, then the team optimizes the proxy while the real outcome quietly deteriorates.
03
False precision
A dashboard presents exact numbers from incomplete tracking, biased samples, unclear definitions, or unstable instrumentation.
04
Local optimization
One screen improves its conversion while the wider journey becomes more confusing, extractive, or difficult to recover from.
05
Average-user blindness
Aggregate performance improves while people on older devices, slower networks, different languages, or assistive technology fall behind.
06
Research theatre
Teams collect interviews, heatmaps, and dashboards but the evidence never changes a roadmap, interface, or product decision.
05 / PRINCIPLES
A measurement ethic for design.
- 01
Measure outcomes, not interface activity alone.
- 02
Pair every important metric with a human explanation.
- 03
Treat accessibility, trust, and comprehension as product performance.
- 04
Use experiments to reduce uncertainty, not to decorate a predetermined answer.
- 05
Keep the cost of measurement proportional to the value of the decision.
- 06
Preserve what the team learns so the same uncertainty is not purchased twice.
HAAM / DATA-DRIVEN DESIGN
Build a product that can explain what it learned.
HAAM connects user research, product analytics, experimentation, accessibility, and interaction design into one evidence system for better product decisions.