Choosing a research approach for a business question: from the decision to the method (Research and Evidence for Better Decisions) | READOS

Businesses research all the time, even when they do not call it research. An owner rings a few customers before launching a service. A manager compares supplier quotes and checks references. An engineer reviews warranty returns to find a pattern. A team runs a pilot before rolling out a new process. Some of this work is excellent; much of it answers the wrong question, uses a method that cannot support the conclusion drawn from it, or collects data that nobody uses. The difference usually lies not in the effort or the budget but in the choices made before any data is gathered.

Good research starts with a decision. What will be done differently depending on the answer? That determines the question, and the question determines the approach: whether the business needs numbers that describe how common something is, explanations of why people behave as they do, evidence that one thing causes another, or an evaluation of whether something worked. Each kind of question calls for different methods, different samples and different ways of analysing results. Choosing the method first and the question second is one of the most common and expensive mistakes in practical research.

This article explains how to move from a business decision to a research question, the main kinds of research questions, quantitative, qualitative and mixed approaches, the menu of practical methods, how to judge whether research will be credible, and how to plan research that is proportionate to the decision it supports. It opens a series on research and evidence for better business decisions.

Start with the decision

Before choosing any method, write down:

the decision the research will inform, such as whether to launch a product, change a supplier or adopt a process;
the options being considered;
what would change the decision: the findings that would favour one option over another; and
the deadline and budget for deciding.

If no plausible finding would change the decision, the research is not worth doing. If the decision must be made next week, a six-month study is irrelevant. The the value of information article explains how to judge how much information is worth before a decision.

Kinds of research questions

Research questions generally fall into a few types:

Exploratory: What is going on? What issues matter to customers? Exploratory research maps unfamiliar territory and generates ideas and hypotheses.
Descriptive: How many, how often, how much? What share of customers use a feature, how long jobs take, how satisfied people are. Descriptive research measures.
Explanatory or causal: Why does this happen? Does this change cause that result? Explanatory research tests relationships.
Evaluative: Did it work? Did the new process reduce defects, did the campaign increase enquiries, did training change behaviour? Evaluative research judges an intervention against its aims.

Many projects combine types: exploring first to understand the issues, then describing their extent, then testing a solution and evaluating it.

Quantitative, qualitative and mixed approaches

Quantitative research collects numerical data from enough cases to measure, compare and generalise: surveys with fixed answers, counts from records, measurements and experiments. It answers how many, how much and how strongly. Its strength is precision and the ability to generalise from a sample; its weakness is that it can miss the reasons behind the numbers.

Qualitative research collects words, observations and stories from fewer cases in depth: interviews, focus groups, observation and documents. It answers why, how and what it means. Its strength is depth and discovery; its weakness is that findings from a few cases cannot be assumed to represent everyone.

Mixed methods combine both, often in sequence: qualitative work to understand the issues and design good questions, then quantitative work to measure them, or quantitative results followed by interviews to explain them. Mixed approaches suit most business decisions.

Primary and secondary research

Secondary research uses information that already exists: industry reports, government statistics, published studies, competitor information and the business’s own records. It is usually faster and cheaper and should come first. Primary research collects new information directly, through surveys, interviews, observation or experiments, when existing information does not answer the question. The market research on a small-business budget article describes low-cost sources of both.

Common practical methods include:

Desk research: reviewing published sources and internal data.
Analysis of existing records: sales, quality, maintenance, timesheets, complaints and web analytics.
Surveys: structured questions to many people.
Interviews: in-depth conversations with individuals.
Focus groups: guided discussions with small groups.
Observation: watching how work or customer behaviour actually happens.
Case studies: detailed examination of a few examples.
Experiments and trials: changing one thing deliberately and measuring the effect, including A/B tests.
Pilots: limited real-world implementation to learn before scaling up.

Each method suits some questions and not others. Surveys measure prevalence but rarely explain causes. Interviews explain but do not measure. Only experiments, or carefully designed comparisons, give strong evidence of cause and effect.

Matching methods to questions

A simple matching guide:

To discover issues and generate ideas, use interviews, observation and desk research.
To measure how common something is, use surveys and records with a suitable sample.
To understand why something happens, use interviews, observation and case studies, then test explanations with data.
To test whether a change causes an improvement, use experiments, controlled trials or before-and-after comparisons with a comparison group.
To decide whether to scale up, use pilots with clear success criteria set in advance.

Credibility: validity and reliability

Research is useful only if its findings can be trusted. Two ideas help judge this:

Validity: does the research measure what it claims to? Internal validity concerns whether conclusions about cause and effect are justified within the study. External validity concerns whether results apply beyond the people, places and times studied.
Reliability: would the research give similar results if repeated? Consistent questions, procedures and coding improve reliability.

Common threats include biased samples, leading questions, small numbers, measurement errors, the researcher’s own expectations and confounding factors that explain a result better than the factor being studied.

Triangulation

Triangulation uses more than one method, source or perspective to examine the same question. If customer interviews, sales records and a survey all point the same way, confidence rises. If they disagree, the disagreement is itself a finding worth investigating. Triangulation is one of the most practical ways for small businesses to strengthen evidence without large budgets.

Writing a research plan

A short plan, often one or two pages, clarifies the work and makes it easier to review. It typically covers:

the decision and the research question;
specific objectives or sub-questions;
the approach and methods;
who or what will be studied, and how they will be chosen;
the data to be collected and how;
how the data will be analysed;
ethical issues, including consent and privacy;
timeline, budget and responsibilities; and
how results will be reported and used.

Agreeing in advance what result would support each option protects against the temptation to interpret results to fit a preferred answer.

Proportionate research

Research should be proportionate to the decision. A low-cost, easily reversed decision needs little research; a large, irreversible investment justifies much more. Quick, inexpensive methods, such as a handful of interviews, a review of existing records or a small pilot, often reduce most of the uncertainty that matters. The goal is a better decision, not a perfect study.

Research when time is short

Many business decisions cannot wait for a full study. Rapid methods trade some rigour for speed: five or six focused interviews in a week, a quick analysis of the last year’s records, a one-day observation of a process, a short online poll of existing customers or a small test with a limited group. Rapid research works best when the question is narrow, the people closest to the issue are involved and the limitations are stated plainly. A quick study that clearly reports what it can and cannot show is more useful than a slow one delivered after the decision has been made.

Involving the people who will act on the results

Research is more likely to change a decision when the people who will act on it help shape it. Asking the sales manager, operations lead or engineer what they need to know, what would convince them and what they already believe brings out hidden assumptions and makes the findings harder to dismiss later. Involving them in reviewing early results also builds understanding of the evidence and its limits.

Researching inside the business

Not all business research faces outward. Questions about processes, staff experience, quality problems, delays and costs are researched inside the organisation, using records, observation, interviews with staff and small trials. Internal research raises particular issues: staff may worry about how findings will be used, managers may be invested in current practices, and the researcher may be a colleague with their own views. Explaining the purpose, protecting individuals’ confidentiality and focusing on systems rather than blame help produce honest information.

From findings to action

Research ends with a decision, not a report. Findings should be summarised against the original question and decision criteria, with the strength of the evidence made clear, the options compared and a recommendation stated. Recording what was decided, and later what happened, builds a body of evidence the business can learn from. Where findings are uncertain, a staged decision, such as a pilot, a limited launch or a decision review date, often makes better use of the evidence than an all-or-nothing choice.

Common traps

Solution-first research: seeking evidence to support a decision already made.
Vague questions that cannot be answered or acted on.
Methods chosen by habit, such as a survey for every question.
Asking people to predict their own behaviour and treating the answers as fact.
Collecting too much data and analysing too little of it.
Producing a report instead of a decision, so findings are filed and forgotten.
Leaving internal staff unsure how research findings about their work will be used.
Ignoring existing data that would answer the question for free.

A worked example

This is an illustrative example. The business and figures are hypothetical.

A small machinery maker is considering launching a paid preventive maintenance service for customers’ equipment. The decision is whether to launch in the next financial year, and the owner agrees that the service would be worthwhile if at least 15% of current customers were likely to sign up at a price that covers costs.

The research plan uses mixed methods. Desk research reviews the company’s service records, which show how often customers call for breakdown repairs. Ten interviews with customers of different sizes explore how they manage maintenance now, what frustrates them and what they would value. The interview findings shape a short survey sent to all 300 current customers, asking about current practices, interest in specific service options and acceptable price ranges. Finally, a three-month pilot offers the service to eight customers to test costs and delivery.

The interviews reveal that customers value guaranteed response times more than scheduled visits, so the service design changes before the survey. The survey and pilot together show interest from about a fifth of respondents at the tested price, and the pilot confirms costs. The owner launches the service, with the response-time guarantee as its centrepiece. The reading a pilot honestly article explains how to judge results like these without overreading them.

How this applies to a small Australian business

Write down the decision before choosing any method.
Classify the question as exploratory, descriptive, explanatory or evaluative.
Check existing data first.
Combine methods to triangulate findings.
Set success criteria in advance.
Record decisions and outcomes so the business learns from its research over time.
Involve the people who will act on the findings from the start.
Report limitations plainly, especially for rapid research.
Keep research proportionate to the size and reversibility of the decision.

Frequently asked questions

Is qualitative research less rigorous than quantitative research? No. It answers different questions. Rigour in qualitative research comes from careful sampling, systematic analysis and transparency about how conclusions were reached.

How much research is enough? Enough to reduce the uncertainties that would change the decision, at a cost justified by what is at stake.

Can a small business run experiments? Often, yes: testing two versions of an offer, changing one process step on some shifts but not others, or piloting a change in one area before others.

Should research be done in-house or by an agency? In-house research suits questions close to the business’s own operations and customers, where context matters and budgets are small. Agencies add value for large samples, specialist methods, independence from internal views or sensitive topics where customers may speak more freely to an outsider.

Questions to ask

What decision will this research inform, and what result would change it?
Is our question exploratory, descriptive, explanatory or evaluative?
What existing data could answer it?
Which combination of methods would give credible evidence?
Who will act on the findings, and have they helped shape the questions?
Have we agreed in advance how we will interpret the results?

Bringing it together

Good business research begins with a decision, turns it into a clear question, and chooses methods that can answer that question credibly: qualitative methods to explore and explain, quantitative methods to measure and generalise, experiments to test causes and pilots to learn before scaling up. Secondary research comes first, triangulation strengthens conclusions, and a short plan with success criteria set in advance keeps the work honest and proportionate.

Source: KEVOS editorial notes. Examples and figures in this article are illustrations.

Choosing a research approach for a business question: from the decision to the method