Marketing research example: the starting situation
Picture a children's clothing shop in Tashkent. It sells through Instagram and a Telegram channel, and customers pay with Click, Payme or cash to the courier. The owner notices something odd: people message the shop often and the comments under posts are lively, yet orders are noticeably fewer than conversations.
The first instinct is to launch targeted ads and bring in more people. But if interest already exists, new traffic will drop into the same funnel and leak out at the same spot. Before spending any budget, the owner needs to learn where exactly people drop off.
Steps 1–2: the question, the decision and the hypotheses
"Few orders" is a complaint, not a research question. It gets rewritten: at which point in the conversation do buyers leave, and what should change so they reach payment? The owner also lists the decisions that depend on the answer: revisit prices, product descriptions, delivery terms, reply speed or advertising.
Then come the possible causes, written down as hypotheses:
Each one has to be confirmed or dropped. Hypotheses keep the search focused; without them research turns into collecting everything in sight.
- The price is higher than similar items on Uzum and OLX
- Choosing a size is confusing and people fear getting it wrong
- Delivery time and cost to the regions are unclear
- Prepaying an unfamiliar account feels risky
- Messages are answered too late
Step 3: start with the data you already have
The cheapest source is your own records. The owner goes through recent Instagram and Telegram conversations and marks, for each, how it ended and which question it stalled on. At the same time she reads comments, compares the listings of several competitors on Instagram, Uzum and OLX, and checks when messages arrive and when they get an answer.
A pattern shows up quickly. Most conversations stall after a question about size and after "when will you deliver?". People ask about price but rarely leave because of it. Evening messages wait until morning, and by then the person has often bought somewhere else.
Step 4: talk to real customers
Conversations show what happens, not why. So the owner contacts a few people who showed interest but did not buy, and a few regular customers. The talk is short and open: what were you looking for, what stopped you, what convinced those who ordered. A two- or three-question poll in the Telegram channel can add more voices.
The answers are specific. The size chart is buried in an old post, and a parent is afraid of choosing wrong. It is unclear whether an order will reach Samarkand or Fergana and how long it takes. Sending money to a card before receiving the item feels uncomfortable. Regular customers value getting a reply from a real person rather than a template.
Steps 5–6: findings and decisions
The hypotheses are reviewed one by one. Size confusion, delivery uncertainty and distrust of prepayment were confirmed. Reply speed was partly confirmed. Price and marketplace competition were not. From these findings the owner builds an action list with a built-in check:
- A size chart and photos on a model in pinned Stories and in every listing
- One clear delivery block: within Tashkent, to the regions, timing and cost
- Plain return terms and pay-on-delivery where it is realistic
- Reply templates for frequent questions and someone on duty in the evening
- A second review of conversations in a month to see whether more reach an order
What the owner should not have done
Launching ads right away would have sent more people to the same unanswered questions. A discount would have cut the margin even though price was not the cause. The research took a few working days and showed which actions would have been pointless, which is as valuable as the fixes it found.
It is fair to name the limits too. The sample is small, and people who agree to talk are not the whole audience. So the conclusions are not treated as law; they are tested by making changes and measuring again.
How to reuse this in your own business
The order never changes: decision, question, hypotheses, internal data, customer conversations, findings, action list, re-check. Without the last two steps, research ends up as a report in a folder. The main cost is time for reviewing data and interviews, sometimes survey or analytics tools; large panel studies are usually unnecessary for a small business.
If you are short on time or want an outside view, markdev works on marketing strategy and consulting in Uzbekistan: we help frame the question, gather the data and turn findings into a plan. Describe your situation and we can discuss where to start.
FAQ
How many people should a small business interview?
There is no single number. To understand reasons, a few dozen short conversations are usually enough, or even fewer at first, until answers start repeating. Estimating shares or comparing segments needs a larger sample.
How is marketing research different from regular analytics?
Analytics shows what happens in numbers: traffic, leads, sales. Research answers why, drawing on conversations, observation and competitor comparison as well.
Can I run a study myself?
Yes, if you have access to conversations, reviews and customers. Write down the question, the hypotheses and the decisions that depend on the answer beforehand, and avoid bending conclusions toward what you hoped to see.
Where do I begin if I have almost no data?
Start by talking to current customers and to people who walked away, reading reviews and looking at competitor listings on Uzum, OLX and Instagram. It is cheap, quick and gives you first hypotheses.
Let’s look at your task for free
Leave your contact — in a 15-minute call we’ll show where leads get lost and what to fix first.