Problem-Solving Skills: A Step-by-Step Process for Any Problem

Most people who struggle with a problem are not short of answers. They reach for one too early. Good problem solving skills are the habits that slow that moment down: a repeatable way to work out what is wrong before choosing what to do about it, and to check afterwards whether the fix worked. The same six steps handle a late report at work and a chaotic morning at home:

1. Define the problem
2. Find the causes
3. Generate options
4. Choose against criteria
5. Test on a small scale
6. Review what happened

The first and the last are easy to skip.

Prefer to watch? The video covers the same ground as this article.

VideoProblem‑Solving Skills: A Step‑by‑Step Process for Any ProblemWiserHours

What employers mean by problem solving skills

When employers ask for problem solving, they mean a bundle of thinking skills rather than one talent. Analytical thinking is the core skill employers name most often, with about 7 in 10 of the more than 1,000 companies surveyed for the World Economic Forum’s Future of Jobs Report 2025 calling it essential.¹ Creative thinking came fourth, and what psychologists count as creative is narrower than the everyday label suggests. These are employers’ views of what their workforce needs, not measurements of what makes people succeed.

O*NET, the US Department of Labor’s occupational database, defines complex problem solving as identifying complex problems and reviewing related information to develop and evaluate options and implement solutions.² In the US National Association of Colleges and Employers’ Job Outlook 2025 survey, nearly 9 in 10 employers said they look for evidence of problem-solving ability on students’ résumés.³ Both are US sources, so check the skills framework used where you work or study.

It is a bundle because a problem can go wrong at several points: you can spot it but misread it, or find the cause and never follow up. Read side by side, the definitions share five sub-skills (our grouping), each matching a step here:

• Diagnosis: spotting that something is wrong and stating it precisely.
• Analysis: separating causes from symptoms.
• Creative thinking: producing more than one possible answer.
• Judgment: weighing options against what matters.
• Follow-through: putting a fix in place, checking it and adjusting.

Compare two illustrative CV lines. “Strong problem-solver” proves nothing. “Traced late weekly reports to a sign-off queue, tried a deputy approver for a month and checked the result” shows diagnosis, options, a test and follow-through in one sentence. In an interview, tell one real problem in that order, ending with how you checked that the fix worked. The STAR method for structuring interview answers gives that story a shape interviewers recognize.

Why a process beats a flash of insight

The OECD treats problem solving as a set of processes that can be learned and measured. Its framework for the 2012 PISA test of 15-year-olds names four: exploring and understanding the problem, representing and formulating it, planning and executing a solution, and monitoring and reflecting on progress.

Those four echo the mathematician George Pólya’s 1945 book How to Solve It, which split the work into understanding, planning, carrying out and looking back; Pólya also held that the skill is learned by imitation and practice. The framework makes no assumption that the processes run in order, and real solvers may not move through them in a straight line.⁴

Myth

Good problem solvers simply see the answer.

Fact

The OECD assesses problem solving as processes, such as exploring a problem and monitoring progress, and its framework calls it a learnable skill, gained through imitation and practice.

The OECD’s 2023 Survey of Adult Skills, run in 31 countries and economies, measured adaptive problem solving: reaching a goal in a changing situation where no method of solution is obvious.⁵ On average across the participating OECD countries and economies, 29 percent of working-age adults scored at Level 1 or below, so roughly 3 in 10 are low performers on problems with no ready-made method. The survey’s description of its top level is about how people work through a problem: they define the nature of a messy problem, set aside irrelevant information and adapt when the situation shifts.⁶

A process makes room for the unglamorous moves, describing and checking, that are easy to skip when an answer feels obvious. The six steps work as a checklist: when a small test fails, going back to the causes is the method doing its job. One work example runs through the next three sections: a team’s weekly sales report keeps reaching managers late on Monday.

1. Define the problem: what is happening, where, when it occurs, and why it matters
2. Find the causes: check each hunch against evidence that could prove it wrong
3. Generate options: several before judging any of them
4. Choose: against criteria written down before the options
5. Test small: on one unit or team first, with a written prediction
6. Review: what worked, what did not, what to change next time

Steps 1 and 2: define the problem and find its causes first

Defining and diagnosing are easy to rush. In surveys of 106 senior executives from companies in 17 countries, reported in Harvard Business Review in 2017, the author Thomas Wedell-Wedellsborg found that 85 percent agreed their organizations were bad at diagnosing problems.⁷ That is opinion rather than a measured result, yet it matches the pattern he describes: managers jump to solutions before checking that they understand the problem.

1. Define the problem

Write one or two sentences that describe what is happening, not what should be done about it. ASQ, a professional body for quality management, suggests answering plain questions: what happened, where it was noticed, when it occurs, how much of it there is and why it matters, while keeping facts apart from opinions.⁸

Take the running example. “We need a better reporting tool” is a solution in disguise. “The weekly sales report has reached managers after its Monday deadline most weeks this quarter, so pricing decisions wait a day” is a problem. The second version says what good looks like and what the delay costs. If your definition names a tool, a solution or a person to blame, rewrite it.

2. Find the causes

A cause is something that, removed, would stop the problem coming back; a symptom is what you see. The trap at this step is checking your favorite explanation only in ways that could confirm it. In a classic 1960 experiment, the psychologist Peter Wason asked 29 young adults to work out a simple rule about numbers, and only about 1 in 5 reached the right rule without first announcing a wrong one.⁹ In the task, people guessed the rule behind a trio of numbers by proposing new trios, and participants tended to offer trios that fitted their own guess rather than ones that could disprove it.¹⁰

Later researchers argued that testing cases that fit your hunch is not always a mistake.¹⁰ It misleads in situations like Wason’s task, where many wrong explanations fit the same evidence.⁹

In the example, the team suspects the data export is slow. A check that could prove this wrong is to compare when the export finishes with when the report leaves. If the export is done early and the report still goes out late, the cause lies elsewhere, perhaps in the queue for finance sign-off. Tools such as the 5 whys and fishbone diagrams give this step structure; how root cause analysis works, and why it often falls short sets them out in detail.

Steps 3 and 4: widen the options, then choose against criteria set first

Picking from a list of one is linked to failed decisions. Studying 358 decisions in medium-sized and large US and Canadian organizations, the management researcher Paul Nutt found that about half failed. He traced the failures to managers who imposed a solution, limited the search for alternatives and used power to push plans through; managers who set objectives and searched widely for solutions were more likely to succeed.¹¹ Because these were observational studies of real decisions, they reveal a pattern rather than proof of cause.

3. Generate options

ASQ’s advice is to hold off choosing until several alternatives are on the table, because groups too often start planning their first idea.⁸ A useful minimum is three: one cheap option, one that removes the cause entirely and one that sounds slightly odd. For ways to widen the list, see twelve techniques for coming up with ideas; for problems about how people use a product or service, design thinking’s observe-and-prototype approach adds watching real users.

In the example, if the cause is the sign-off queue, the options include moving sign-off to Friday, letting a deputy approve routine weeks, dropping sign-off when figures are within the usual range, or sending the report on time marked “provisional”.

4. Choose against criteria

Write the criteria before you score the options, so your favorite cannot set the rules. ASQ’s page lists the questions to ask of each candidate: how well it will work, whether it can be carried out, whether it creates new problems and whether the people involved will accept it.⁸ For a choice with many factors, a weighted decision matrix turns those criteria into scores.

Measured against “on time”, “no extra errors” and “finance agrees”, deputy approval wins. The “provisional” report fails the error test, because managers would act on figures that might change.

Steps 5 and 6: test small, then review what happened

The last two steps turn a decision into evidence. ASQ recommends trying a chosen solution first on a typical subset, such as one region or one unit, and building in ways to monitor it once it runs.⁸

5. Test on a small scale

A small test limits the damage if you are wrong and exposes problems a plan hides. The same logic sits under the plan-do-study-actplan-do-study-act: A four-step loop for testing a change: plan it with a written prediction, try it on a small scale, study what happened against the prediction, then act by adopting, adapting or dropping it. It comes from the quality methods of Walter Shewhart and W. Edwards Deming.Full entry in the glossary cycle used in health-care improvement: test a change against a prediction, start small and track data over time. A 2014 systematic reviewsystematic review: A review that fixes its question and its rules for including studies in advance, then searches out every study that fits and weighs them together. Some systematic reviews pool the results into a meta-analysis; others describe what the studies found without combining the numbers.Full entry in the glossary of peer-reviewed health-care studies that used the cycle found that many did not follow these core features.¹² A test without a written prediction is easy to reinterpret afterwards, so write yours down before it starts.

In the example, the team tries deputy approval for a month on one region’s report, predicting it will arrive before the deadline every week.

6. Review what happened

A review, also called a debrief or after-action review, is a structured look back that builds in reflection, discussion and goals for next time.¹³

The study

Moderate evidence

What 46 samples say about reviewing your work

Across 46 samples of published and unpublished debrief research, 2,136 people in all, Scott Tannenbaum and Christopher Cerasoli found that on average people and teams who debriefed were about 25 percent more effective than controls. The average effect held about equally for teams and for individuals, for simulations and real work, and for medical and non-medical groups.¹³

A short, structured review is one of the cheapest ways to get better at the next problem, though the pooled studies used different designs and blended simulations with real work. Nathanael Keiser and Winfred Arthur’s later meta-analysismeta-analysis: A study that combines the results of earlier studies on the same question into one overall estimate. Pooling makes the estimate more precise, but it cannot repair the studies it pools: a meta-analysis of surveys is still survey evidence.Full entry in the glossary (2021), which measured training outcomes, also found a sizable benefit, and reviews built on objective records of performance tended to work better.¹⁴

In the example, the month-end review shows the report arrived on time in every week but one, when the deputy was on leave, so the change is to name a second deputy. Keep a review like this to about ten minutes, base it on records rather than memory, and end with one named change.

Before you call a problem solved

• The problem is written down without a solution in it
• Each suspected cause has been checked against evidence that could prove it wrong
• At least three options were listed before choosing
• The criteria were written before the options were scored
• The fix was tried on a small scale with a written prediction
• A review compared the prediction with what happened and named one change

Download PDF ↓

Get this checklist as a one-page PDF. We'll email you the download link.

Your emailAlso send me the weekly newsletter (launching soon)Email me the PDF

We use your address to send this link and, only if you tick the box, the newsletter. Privacy policy.

A worked everyday example: the morning that always runs late

At household size, each step takes minutes rather than a meeting. Take an illustrative family whose problem, defined without a solution in it, is “we leave after the bus time on most school days”. The first hunch is that the children wake late, but a check shows they are up on time; the delay is a hunt for shoes, keys and signed forms in the last ten minutes. The options: an earlier alarm, a fixed spot by the door for things that leave the house, bags packed the night before, a fridge checklist.

The criteria: it costs no sleep, takes a few minutes a day and the children can do it themselves. That rules out the alarm and favors the spot by the door plus bags packed at night. The family tries it for one school week, predicting on-time departures every day. At Friday’s review, the two late days both involved a form needing a signature, so the change is to put forms on the table on Thursday night.

How well tested is each of the six steps?

The six steps come mostly from frameworks and professional practice, not from trials of each step. One reason is that a single step is hard to test on its own. The review has the firmest direct evidence, from meta-analyses of debriefs; the others rest on surveys, observational studies, one classic experiment and expert guidance.

Treat the early steps as sound practice that is hard to test directly, and the review as the habit with a measured payoff.

The bottom line

Of the six steps, the review has the strongest evidence. After your next attempt at a problem, spend ten minutes comparing what happened with what you predicted, and write down one change for next time.