Thinking · Lessons from practice

Designing the System, Not Just the Screen

How I turn user problems into focused product opportunities.

Users rarely describe problems in business language. They describe delay, uncertainty, repeated work, or a task they cannot complete. That is valuable evidence, but it is not yet a product opportunity.

User problem → desired outcome → business consequence → system dependency → product bet → evidence

This is how I use systems thinking. I look beyond the visible screen, then narrow the problem into an intervention a team can build, test, and learn from.

01 · Diagnose the outcome and the cost of the problem

“Make this easier” is not enough. I first ask what people are trying to achieve, what blocks them, and what the failure costs.

The user cost might be time, uncertainty, rework, exclusion, or loss of confidence. The business consequence might be low adoption, support demand, operational cost, risk, or a missed strategic capability. These remain hypotheses until evidence supports them.

TakeawayA user problem becomes strategically useful when we can explain both the outcome at risk and why solving it matters.

02 · Find where the system creates or destroys value

Most products are connected systems. A person’s outcome can depend on data, policies, permissions, teams, technology, and transitions between tools. The interface may reveal the problem without causing it.

From a user problem to an evidence-backed product bet

  1. 01Observe the problem

    What is difficult, uncertain, slow, or impossible?

  2. 02Define the outcome

    What is the person actually trying to achieve?

  3. 03State the business consequence

    Why might this matter to adoption, cost, risk, or growth?

  4. 04Map the dependencies

    Which data, policies, teams, tools, and transitions shape it?

  5. 05Choose a focused bet

    Where can a realistic intervention create meaningful value?

  6. 06Decide the evidence

    What would justify continuing, changing, or stopping?

A reusable decision model. It converts observed friction into a testable opportunity without assuming that every problem needs a feature.

Mapping the whole outcome exposes the important seams: where context disappears, responsibility becomes unclear, information loses trust, or a policy prevents progress.

This does not mean redesigning everything. It means understanding enough of the system to avoid polishing one screen while the surrounding experience still fails.

In practice: one access layer across a reporting system

The athenahealth product platform Legacy reports Older data foundation (pre-modernization) New reports Modernizing data platform Templates Athena-curated standard library User-saved reports Users' configured queries and outputs Report Library Access layer Discover · Understand · Retrieve · Manage · Govern a mature library of accumulating reporting content Users access reporting from within the workflow they're already in Reporting is embedded in the product platform, not a separate destination
A worked example from Report Library. Four kinds of reporting content had grown up on different foundations. The brief was not to replace that system but to design one access layer across it, inside the product people already used. Diagram drawn for this portfolio; not an original artifact.

TakeawayProduct boundaries organize teams; value is created across the user’s outcome.

03 · Frame the opportunity before choosing the solution

A pain point does not automatically deserve investment. I compare opportunities through five questions:

  • How important is the user outcome?
  • What business value could improve if we solve it?
  • How often and how widely does the problem occur?
  • Where do we have evidence, ownership, and a feasible path?
  • Could the change create reusable capability or reduce future complexity?

Signals worth investigating

Observed signalSystem questionValue hypothesisEvidence to seek
People abandon or work around a critical taskIs the failure in the interaction, information, policy, permission, or handoff?Higher completion and adoptionTask success, usage, and support patterns
Teams repeatedly perform the same manual workWhich rule, state, or decision could the system support safely?Lower operating cost and faster throughputTime, error, exception, and recovery data
Users see conflicting answersWhere do definitions, sources, or transformations diverge?Greater trust and lower decision riskConsistency checks and verification behavior
Context disappears between tools or teamsWhat must travel with the user for the outcome to continue?Workflow continuity and reduced reworkHandoff success and repeated-entry behavior
These are opportunity signals, not automatic feature requests or proven business cases.

The value column contains hypotheses, not measured outcomes. Making that distinction helps a team decide what evidence is needed before increasing investment.

This framing prevents two common mistakes: jumping from feedback to a feature, and using a large system map to justify a large project.

TakeawayThe strongest opportunity connects user value, business value, evidence, and a realistic point of intervention.

04 · Choose the smallest bet that can compound

I prefer a focused intervention that resolves an important dependency and creates learning. The bet should be small enough to test, but meaningful enough to change the outcome.

Move from opportunity to evidence

  1. 1
    User outcome

    Define what must become possible or meaningfully better.

  2. 2
    System dependency

    Identify the relationship creating the greatest loss of value.

  3. 3
    Focused product bet

    Choose the smallest intervention capable of changing the outcome.

  4. 4
    Interaction and delivery

    Turn the bet into behavior the team can build and support.

  5. 5
    Evidence

    Measure whether the user outcome and value hypothesis improve.

↺ Evidence can strengthen, redirect, or stop the investment.

The model connects strategy to execution while keeping the initial business claim testable.

Compounding value can come from shared interaction patterns, clearer state models, better information, reusable components, preserved context, or a policy that removes recurring friction. These assets can improve more than one screen, but only if teams can own and maintain them.

TakeawayThink beyond feature output. Look for a focused bet that improves today’s task and makes tomorrow’s decisions easier.

05 · Define success before increasing investment

Systems thinking does not replace measurement. It makes the assumptions visible so each can be tested.

  • User evidence asks whether people can complete the intended outcome with confidence.
  • Behavioral evidence shows whether they adopt and reuse the workflow.
  • Operational evidence examines support demand, failure, maintenance, or handoff cost.
  • Business evidence tests the value hypothesis that justified the work.

Research, prototypes, technical reviews, shipped behavior, and business results answer different questions. One should never be presented as proof of another.

TakeawayBe precise about what was learned, what changed, and what remains a hypothesis.

What I carry forward

I do not start with the screen or assume every user complaint deserves a feature. I start with the outcome, identify the business consequence, map the dependencies, and choose a focused bet with explicit evidence.

That is the business value of systems thinking: it helps teams invest in the right problem, at the right level, with a clear reason to believe.

This view has been sharpened by three ideas from the wider field: GOV.UK’s principle of solving a whole problem for users, Donella Meadows’ framing of places to intervene in a system, and GitLab’s emphasis on workflow continuity over feature isolation.