Skip to content

Problem Statement

beta  ·  Family discovery-docs  ·  Phase define  ·  Sizes lean, full  ·  ~3,300 tokens

The document that frames which problem is worth solving, who has it, and the evidence that it is real, before anyone names a solution. Lean carries the statement, background, who is affected, impact, evidence and the decision it informs; full adds coping today, scope, success signals and open questions.

How to tell whether you need one, how to pick lean or full, and how to grade what comes back before it goes to the decider named in Decision This Informs.

Before you start: is this the right document at all?

Section titled “Before you start: is this the right document at all?”

Write a problem statement when you need to name which problem is worth solving, who has it, and what evidence shows it is real, before anyone names a solution. Nielsen Norman Group’s own definition is the anchor this bundle builds from: “A concise description of the problem that needs to be solved.” This document’s whole job, per this library’s discovery-docs family contract, is to frame one problem and name no solution, option, project, or investment.

Write something else if:

You actually need Because
a business case business-case’s own Problem and Opportunity section asks the same questions inside a funded investment comparison. This document comes earlier, asks for no money, and decides whether a problem is worth that comparison at all. One Six Sigma source draws the same line by name: the problem statement “describes what, where, and how much, without the persuasive framing”. The order, statement first, is this library’s own position
a project brief a project brief asks a named approver for authority to start initiation. This document asks for no authority and names no project. The Association for Project Management’s own published glossary has no entry under this document’s name, and defines Brief and Charter as two different things
a PRD this library’s own prd companion already names a problem statement as one of the things that precede a PRD
a How Might We question it already points toward generating solutions, where this document names none
a list of constraints a limit any answer must respect, such as a budget, a deadline or a regulation, bounds a solution, and this document names none. Engineering-design teaching and a capstone proposal label constraints, and one engineering paper puts them inside the problem statement, a dissent this library does not follow; project-brief’s Constraints and Who Should Be Involved section carries them. That routing is this library’s own position

An opportunity statement is a separate type in this library, and it is not built. Nielsen Norman Group names the two as aliases, “written and used in the same way”, phrased positively instead of as a problem. One business-analysis source dissents for projects specifically, naming the move where “practitioners treat problems and opportunities interchangeably when developing Business Cases and Project Charters” as an error, not a harmless alias. On 2026-10-17 the maintainer ruled the two different documents, and the opportunity statement became catalog entry 207, a candidate type of its own. On 2026-10-22 the maintainer parked it rather than build it (#298), because no template with sections for it was found. Frame an opportunity here when Nielsen Norman Group’s reading fits your case. A design-thinking Point of View statement, by contrast, is listed as an alias of this document.

A net-new initiative, where no product exists yet, can use this document too. Jeff Gothelf publishes a Business Opportunity Statement for that case, and still says: “My advice is to try the Business Problem Statement first.” To fill this document from his formula, put the people the gap leaves unserved in Who Is Affected, named closely enough that someone could go and find one. Put the gap itself in Evidence, with where you saw it. His formula’s line for it reads “What existing products / services fail to address is [this gap].” Leave out his next line, “Our product / service will address this gap by [vision / strategy].” That line names a direction for a solution, which this document never names. Carry it to the document you name in Decision This Informs.

Write nothing at all if no one is positioned to decide, once this document exists, whether the problem is worth pursuing further. GOV.UK frames exactly this decision: “Your discovery is finished when you’ve decided whether or not you want to move on to alpha.” A statement with no real decider named in Decision This Informs has not finished the job this document exists to do.

Lean carries Problem Statement, Background, Who Is Affected, Impact and Why Now, Evidence, and Decision This Informs: enough to frame the problem, show it is real, and name the decision it feeds. Full inserts How People Cope Today, What Is Not Part of the Problem, How We Will Know It Is Solved, and Open Questions and Assumptions, in that order, between Evidence and Decision This Informs. The signal to move from lean to full: will a reader reasonably ask how people work around the problem today, what this statement deliberately excludes, what solved would look like, or what is still unknown, not only who has the problem and what shows it is real.

Both sizes stay short. NHS Wales puts a typical instance at “about 18 words long”, and Nielsen Norman Group’s own brevity rule holds at both sizes: “If you can condense your problem statement down to a few sentences, others will quickly understand what you focus on and why, and what’s out of scope.” The full variant adds four named fields around the statement; it does not add length to the statement itself, and no source read for this bundle abandons that brevity convention for a longer form.

Score each row 0, 1 or 2. Below 14 out of 20 on a full statement, a reader cannot tell whether this problem is worth taking to a project brief or a business case without going back to ask the author questions this document should have answered on its own. Below 9 out of 12 on a lean statement, the same gap opens with four fewer sections standing between a weak answer and the decider.

# Criterion 0 1 2
1 No solution or cause The statement names a feature, a tool, or a fix to build, or states a cause as settled fact No solution or cause is named, but you cannot tell from the sentence who has the problem or what is happening to them You can point to the sentence, name the affected person and what is happening to them, and no solution or cause appears anywhere in it
2 Affected group is findable “Users” or “customers,” with nothing more A group is named, but too broadly for anyone to go find a member of it Someone could take the sentence and go find a person who matches it, without guessing
3 Origin is traceable No indication of how the problem came to attention, or which team or product owns it An origin is named, but you cannot tell what specifically surfaced it now rather than earlier You can name who or what flagged the problem, and point to the specific event that surfaced it
4 Evidence outranks opinion Every row in the table is an assertion with no named source At least one row names a source, but the strongest row is still an unsourced assertion The strongest row names where it came from, specifically enough that someone could go check it themselves
5 Cost tied to timing No cost or consequence is named, or a cost is named with no reason to act now rather than later A cost and a timing reason both appear, but you cannot tell whether the cost actually grows if nothing changes You can point to a trend or a date in the document, and tell from it why waiting costs more than acting now
6 Decision has an owner No name, role, or date appears, and the next document is left generic A decider, a date, and a next document are all named, but nothing shows the decider agreed to decide You can point to when and how the named decider agreed to decide, and the specific next document the decision leads to
7 Workaround stays evidence (full only) No workaround or prior attempt is named A workaround is named in enough implementation detail that it reads as a proposed fix The workaround is named at the level needed to judge how painful it is, and a prior attempt plus why it fell short is also named
8 Exclusion draws a boundary (full only) The field is blank, or says everything else is out of scope A neighbouring problem is named, but you cannot say who would have assumed it was included A specific neighbouring problem is named, and you can say who would reasonably have assumed it was in scope before this line drew the boundary
9 Measure carries no target (full only) No measure is named, or the row names a solution’s output instead of a measure of the affected person’s experience A measure and a current level both appear, but a target number or a solution’s output also appears in the row The row names a measure, its current level, and a direction, with no target number anywhere in it
10 Root cause stays hypothesis (full only) A cause is stated as settled fact anywhere in the document A cause is labeled a hypothesis in this section, but an open question or assumption elsewhere still reads as settled Every suspected cause in the document sits only here, labeled a hypothesis, and at least one open question is named that would change the statement if answered differently

Which rows apply to what. Full ships all ten rows because a reader of the full variant can ask about coping, exclusion, success, and open questions as well as who, what, and why. Lean ships six, every row the lean variant has a section to support.

Document Rows Maximum Score against
lean 1-6 12 9
full all 10 20 14

A solution in disguise. Digital NSW and GOV.UK both publish the identical corrective move, rewriting a handed-down solution as a problem, and Build for Good states the rule directly: “Do not include solutions in your problem statement.”

Scope set too broad or too narrow. A UX design guide states the tension directly: the scope should be “broad enough to allow for creative thinking and innovative solutions, but narrow enough that it can provide a direction for that solution.”

Opinion standing in for evidence. NHS Wales’s own warning is blunt: “without evidence of your problem, you are left with opinion.” Digital NSW’s review checklist asks the same question as a gate: “Is this based on evidence rather than assumptions?”

A predetermined cause dressed up as a problem statement. A good statement “does not offer commentary on a proposed solution. It does not attempt to diagnose the problem, nor assign blame. It is objective and factual.” NHS Wales draws the same line from the other side: “problem statements do not describe the solution or root causes.”

A laundry list of unrelated problems. Nielsen Norman Group’s own rule holds that a discovery effort “should have one problem statement, and the problem statement should be focused on one problem.”

One neutral statement flattening genuinely different stakeholder views. A Canada School of Public Service job aid writes a point-of-view statement once per stakeholder rather than merging them into a single voice, illustrated from one inspector’s own line about a device that “is not secure, which makes me feel vulnerable and inefficient.” Where stakeholders genuinely disagree on what the problem is, write more than one statement rather than flattening the disagreement into a false consensus.

A costed estimate where a directional figure belongs. A Six Sigma source draws the line this document stops short of: “the problem statement describes what, where, and how much, without the persuasive framing”, while a specific dollar estimate “gives a sponsor something concrete to weigh” in a business case, not here. Writing a full costed estimate with its own methodology has stepped into business-case’s job, not this document’s.

A target number smuggled into the success-measure row. (full only) Queensland’s own “Defining success” play draws a line this section is built to respect: “The outcome is not an output or a metric of your project.” Name the measure and its current level; leave the number that says how far to move it for the document that proposes how to move it.

When the statement names no solution anywhere in it, when the strongest row in the Evidence table names a source a reader could go check, and when a named person has actually agreed to decide whether the problem is worth pursuing further, by a real date.

Then delete every HTML comment. A statement nobody has agreed to decide on has not finished the job it exists to do, whatever else it contains, so the last thing worth checking is not a section of the statement at all: has the person named in Decision This Informs actually agreed to decide, not merely been copied on the draft.

problem-statement_template-lean.md · ~3,300 tokens

---
title: "{{problem_name}} Problem Statement"
problem_name: "{{problem_name}}"
author: "{{author}}"
decider: "{{decider}}"
status: "{{status}}"
last_updated: "{{date}}"
doc_type: problem-statement
size: lean
source_template: problem-statement
source_template_version: 0.1.2
---
<!--
LEAN PROBLEM STATEMENT. The shortest form that still names which problem is worth solving, who has
it, and what evidence shows it is real, before anyone names a solution. Use it to decide whether a
problem is worth taking to a project brief or a business case at all, not to make that case itself.
To grow it into a form that also carries success measures, a boundary on what the problem excludes,
and open questions, see problem-statement_template-full.md. ADD sections between Evidence and
Decision This Informs; never rename or reorder the ones below, because the full variant is a strict
ordered superset of this one.
SHORT AT BOTH SIZES, NOT ONLY THIS ONE. Nielsen Norman Group's own brevity rule reads, "If you can
condense your problem statement down to a few sentences, others will quickly understand what you
focus on and why, and what's out of scope," and NHS Wales puts a typical instance at "about 18 words
long." The full variant adds four named fields around the statement; it does not add length to the
statement itself. See problem-statement_companion.md section 4.
WHAT A PROBLEM STATEMENT IS, AND IS NOT
It names which problem is worth solving, who has it, and what evidence shows it is real, before
anyone names a solution, an option, a project, or an investment. It is NOT a business case (this
library's own `business-case` bundle asks the same questions inside a funded investment comparison;
this document comes earlier, asks for no money, and decides whether a problem is worth that
comparison at all; one Six Sigma source draws the same line by name, and the order, statement first,
is this library's own POSITION). It is NOT a project brief (which asks a named approver
for authority to start initiation; this document asks for no authority and names no project; an
Association for Project Management consultation draft read for this library's `project-brief`
research names the three documents as interchangeable for one early rationale, but APM's own
published glossary carries no "problem statement" entry and defines "Brief" and "Charter" as two
different things). It is NOT a PRD (this library's own `prd` companion already names a problem
statement as one of the things that precede a PRD). It is NOT a hypothesis statement, a How Might We
question, or a job story (each of those already points at a solution or tests one). It is NOT, in
this library, an opportunity statement: that is a separate catalog type, not yet built, though
Nielsen Norman Group names the two "written and used in the same way." A design-thinking Point of
View statement is listed as an alias of this document.
See problem-statement_companion.md section 6 and section 8.
HOW TO FILL THIS IN
1. Read the comment under each heading: WHAT it wants, WHY it matters (with a pointer into
problem-statement_companion.md), guiding questions to ASK, a GOOD and a WEAK example, and the
TRAP to avoid. For the table, PRIORITY explains the ordering rule and ROW HINT says what a good
row contains.
2. Replace each {{placeholder}} with your content. Name no solution, option, project, or investment
anywhere in this document; that is a later document's job.
3. If a section does not apply, write "N/A" and one line of why, rather than deleting it.
4. Before you send this for a decision: self-grade against problem-statement_guide.md, then DELETE
every HTML comment. They are guidance, not content.
-->
# {{problem_name}} Problem Statement
## Problem Statement
<!-- WHAT The statement itself: a few sentences, from the affected person's perspective, naming no
solution.
WHY Nielsen Norman Group's own definition anchors this section, "A concise description of the
problem that needs to be solved," and NHS Wales adds the brevity bar, "about 18 words
long." Atlassian's own Problem Framing play names the same four elements this section
asks for, "who is affected, what is affecting them, why it needs to be solved, and where
the problem is happening." Deep dive: problem-statement_companion.md section 3
(Anatomy > Problem Statement).
ASK Who has this problem? What is happening to them? Why does it need to be solved? Where and
when does it occur?
GOOD "Patrons who borrow ebooks through the library's mobile app cannot renew a loan from
within the app, and must call the branch during open hours to ask staff to renew it for
them."
WEAK "We need to add a renew button to the ebook app." (a solution stated as if it were the
problem; nothing is left to investigate)
TRAP Naming a solution instead of a problem. Nielsen Norman Group's own quality rule reads,
"Leave solutions out of your problem statement. At the beginning of discovery, there are
too many unknowns, so the best solution is not obvious." -->
{{problem_statement}}
## Background
<!-- WHAT Where the problem shows up, how it came to attention, and what is affecting the people
who have it.
WHY Nielsen Norman Group names the question directly, "Which organization or department has
the problem and what is the problem? Why has the problem arisen?" and Atlassian's own
prompts ask, "What is the nature of the problem? What signs point to there being a
problem?" Deep dive: problem-statement_companion.md section 3 (Anatomy > Background).
ASK Which team, product, or service has this problem? How did it come to attention? Why has
it arisen now rather than earlier?
GOOD "Flagged by three branch managers in the October staffing review, after phone-renewal
requests started crowding out other reference calls during the two weekday hours the
branches are fully staffed."
WEAK "The app has some issues." (no organization named, no origin, no reason it surfaced now)
TRAP Restating the Evidence section's data as narrative instead of explaining where the
problem shows up and how it was noticed. -->
{{background}}
## Who Is Affected
<!-- WHAT The people who have the problem, named specifically enough to find them.
WHY Nielsen Norman Group's own wording leaves room for more than one group, "There could be
multiple user groups affected by a specific problem in different ways," and Atlassian
asks the verification question directly, "Who is experiencing this problem? How do we
know that they are experiencing it?" NHS Scotland's Problem Framer tool carries a closely
related field, "Who The Problem Most Affects." Deep dive: problem-statement_companion.md
section 3 (Anatomy > Who Is Affected).
ASK Who specifically has this problem? Is there more than one affected group, and does each
experience it differently? How do you know they are experiencing it, rather than assuming
it? Does an existing persona already describe this group?
GOOD "Patrons who borrow ebooks through the app, especially those who use it in the evening or
on weekends when no branch phone line is staffed."
WEAK "Our users." (no group named specifically enough to find them)
TRAP Folding more than one differently affected group into one generic line. Nielsen Norman
Group treats more than one group as the normal case, not the exception. -->
- Who is affected: {{who_is_affected}}
- Existing persona, if one already describes this group: {{persona_reference}}
## Impact and Why Now
<!-- WHAT What the problem costs the people who have it and the organization, and why it matters
now rather than later.
WHY Nielsen Norman Group's own prompt runs, "If the problem is not fixed, what will be the
effect on the organization?" and Atlassian asks it from the affected person's side, "What
happens if we don't solve it?" GOV.UK adds the instruction to size the impact where
possible, "You should also consider quantifying the value of solving the problem."
Keep any figure directional: a Six Sigma source draws the line this section stops short
of, "the problem statement describes what, where, and how much, without the persuasive
framing," while a specific dollar estimate belongs to a business case. Deep dive:
problem-statement_companion.md section 3 (Anatomy > Impact and Why Now).
ASK What does this problem cost the affected people, and the organization? Why does it matter
now rather than later? What changes if nothing is done? What is a rough, directional
figure for the cost of not solving it, not a costed estimate?
GOOD "Branch phone lines run near capacity during the two staffed hours, and the October
survey named the call-in renewal step the top app complaint. Ebook loans are a growing
share of total borrowing, so the volume of calls this causes is rising each quarter."
WEAK "This is frustrating for everyone." (no organizational effect, no timing reason, no
figure at all, directional or otherwise)
TRAP Writing a full costed estimate with its own methodology. That is a business case's job,
not this one's; a rough, directional figure is enough here. -->
- Impact: {{impact}}
- Why now: {{why_now}}
## Evidence
<!-- WHAT What shows the problem is real, where each piece came from, and how strong it is.
WHY Digital NSW states the obligation plainly, "Use the evidence you gather to document your
clearly defined problem," and NHS Wales states the cost of skipping it, "without evidence
of your problem, you are left with opinion." Name the extent of the problem, not only
its existence: a Six Sigma source adds the magnitude question, "it is essential to
explain the extent or magnitude of the issue." Deep dive: problem-statement_companion.md
section 3 (Anatomy > Evidence).
ASK What evidence shows this problem is real? Where did each piece come from? How strong is
it, a direct measurement, a survey, or an anecdote? How large or widespread is the
problem, not only that it exists?
PRIORITY List the strongest, most direct evidence first. A statement resting only on the
weakest row is a statement resting on opinion.
ROW HINT A good row names where the evidence came from and what it shows, specifically enough
that a reader could go check it. A weak row asserts a fact with no named source.
GOOD | Branch call logs, Sept-Oct | Renewal-assistance calls are the largest single call
category during staffed hours | Direct measurement |
WEAK | Staff say it's a problem | | Anecdote |
TRAP Opinion standing in for evidence. NHS Wales's own warning is blunt: without evidence, a
problem statement is left with opinion, not a documented fact. -->
| Evidence | What it shows | Strength |
|---|---|---|
| {{evidence_source}} | {{evidence_summary}} | {{evidence_strength}} |
## Decision This Informs
<!-- WHAT Who decides whether this problem is worth pursuing further, by when, and which document
takes over if the answer is yes.
WHY GOV.UK frames the decision itself, "Your discovery is finished when you've decided
whether or not you want to move on to alpha," adding, "It's not a failure to stop at
the end of the discovery phase," and Digital NSW's own checklist ends on the same
question, "Is the problem clear enough to investigate further?" Which document takes over
is this library's own POSITION, from the `discovery-docs` family contract's successor
obligation: name the specific document, a project brief, a business case, or both in
sequence, rather than leaving the pointer generic. Deep dive:
problem-statement_companion.md section 3 (Anatomy > Decision This Informs).
ASK Who decides whether this problem is worth pursuing further? By when do they need to
decide? Which specific document takes over if the answer is yes, a project brief, a
business case, or both in sequence?
GOOD "Decider: the digital services manager, by the end of the month. If the problem is
confirmed worth pursuing, a project brief is written next to ask for authority to
investigate options."
WEAK "We'll figure out next steps later." (no decider, no date, no named next document)
TRAP Leaving the next document generic, or naming an option or a project instead of a
decision. This section names who decides and what happens to the answer, nothing
more. -->
- Decider: {{decider}}
- Decide by: {{decision_deadline}}
- Next document if pursued: {{next_document}}

The reasoning, the history and every source, in the repository:

  • Companion - the long-form argument: why these sections, where the sources disagree, and what the bundle refuses to claim
  • History - what changed in this bundle, and when
  • Research log - every source consulted, with what each one actually supports
  • Catalog metadata - the machine-readable record this page is generated from

Catalog record: 10 sections across 1 format(s), methodology UX discovery / service design, typically owned by PM / UXR.