How a small business can run marketing research on its own: frame the question, choose a method for the task, stay within a small budget and turn the answers into decisions on ads, the website and the product.
KelyanMedia editorial team · Updated November 13, 2026

Research is not for a report but for one decision: which service to promote, who to show ads to, what to put on the first screen of your website, what price to set.
If the decision is not named in advance, data piles up but nothing gets done. That is why the work starts with a question, not with choosing a questionnaire.
In short: what you need to know
- TargetOne study answers one management question that can be written in a single sentence.
- MethodInterviews explain the reasons, a survey shows how common they are, and search data shows demand.
- ResultEvery conclusion becomes an action with an owner and a deadline, otherwise the research will not pay off the time spent on it.
Which question to start your research with
A good research goal sounds like a decision you will make based on its results.
Compare two wordings. "Study the coffee shop market" commits you to nothing. "Find out whether to open a second location in Chilanzar or launch delivery" already tells you who to ask and what data to collect.
- Write down the decision you need to make and the deadline for it.
- List 3 or 4 assumptions your current plan rests on: "customers choose based on price", "people find us through Instagram".
- Mark the assumption that would cost you the most if it turned out wrong. Start with that one.
- Decide what will count as confirmation: for example, most respondents name the same reason.
If you have more than five hypotheses, split the research into stages. A small test of one hypothesis is more useful than a big questionnaire about everything at once.
Which method to choose for the task
The method depends on the question: "why" is answered by talking to people, "how many" is answered with numbers.
| Method | Question it answers | Limitation |
|---|---|---|
| In-depth interview | Why the customer chose you or left, and how they describe the problem | Few people, so you cannot apply the percentages to the whole audience |
| Poll | How often a reason found in interviews actually occurs | Answers depend on how the questions are worded and who answers |
| Competitor analysis | What offers, prices and arguments already exist on the market | You see what competitors do, but not what actually works for them |
| Search data | Whether people search for your service, in what words, in which months | Does not show demand from people who buy through referrals or social media |
| Ad test | Whether people are ready to click and leave a request for a specific offer | Requires a budget, and the result depends on the quality of the ad and the page |
← Swipe to scroll the table →
A common combination for a small business: first 5 or 6 interviews, then a short survey on the reasons you found, then an ad test of the strongest offer. How to prepare and run a conversation with a customer is covered in the article CustDev and in-depth interviews.
How to run research on a small budget
Most of the data is already in your hands: in client chats, in your CRM, in reviews and in your website analytics.
| Source | What you get from it | Time needed |
|---|---|---|
| Chats and calls | Frequent questions, doubts and reasons for refusal | An evening to review recent inquiries |
| Reviews of you and your competitors | What people praise and complain about, and in what words | A few hours |
| Yandex Wordstat, Google Trends | Search query wording, seasonality, comparing directions | An hour or two |
| Customer survey in Telegram | How they heard about you, why they chose you, what was missing | A week to collect responses |
| Calling competitors as a customer | Response speed, prices, sales arguments | A day for 5 or 6 companies |
← Swipe to scroll the table →
Keep the survey short: 5 or 7 questions, mostly closed, with one open question at the end. Do not ask "would you buy", ask about past experience: "when did you last order a service like this and where did you look for it".
How to turn data into conclusions and actions
A conclusion is useful only if it leads to a specific change in the product, price, copy or channel.
- Put all the answers into one table: source, quote or figure, and which hypothesis it relates to.
- For each hypothesis, mark whether it was confirmed, disproved, or there is not enough data.
- For confirmed hypotheses, write down an action: change the offer on the website, turn off a channel, add an argument to the sales script.
- Set a metric to check in a month whether the change worked: number of requests, website conversion, share of refusals after a call.
A hypothetical example: in interviews, dental clinic patients talk about fear of pain more often than about price. A survey confirms this on a larger sample. Action: put information about anesthesia on the first screen of the website and check whether bookings go up.
Separate observations from interpretations. "Three out of six mentioned delivery" is a fact. "Delivery matters most" is a hypothesis that still needs to be tested with a survey.
Learn about full-service digital marketing →Important
A small study helps you make decisions with more confidence, but it does not give an exact sales forecast. Results depend on the sample, the wording and the market. If you collect customers' personal data, check the storage requirements with a lawyer.
FAQ
How many people do you need to survey?
Can you survey friends and employees?
How often should you do research?
Next steps
Need to test demand for an offer: see search advertising as a quick way to test.
Need to build marketing based on the research: learn about integrated online marketing.
Still have questions: call +998 33 339 33 34.
Discuss your marketing strategy →
