The Marketing Research Process in 6 Steps (2026)

The marketing research process is a structured way to define a business question, collect and analyze relevant data, and use what you learn to make better marketing decisions.

It’s a valuable tool in developing an effective marketing strategy. For example, conducting marketing research helps a business discover key information about its market standing and target customers. It also provides essential details about demographics and where to prioritize marketing investments.

This guide shows you how to effectively conduct marketing research.

What is marketing research?

A marketing research strategy is the systematic collection and analysis of information that helps businesses understand customers, identify opportunities and problems, evaluate marketing decisions, and track performance.

The American Marketing Association (AMA) defines it as the function connecting consumers, customers, and the public with marketers through information used to guide those decisions.

That can include studying customer needs, testing a new product concept, comparing competitors, measuring brand awareness, or finding out why sales changed. The research question determines what information you need and how you collect it.

This is not to be confused with market research. Qualtrics defines market research as gathering and interpreting information about your target audience, market, or competitors. But marketing research can extend beyond the market itself to examine the decisions you make around the 4 P’s of marketing: products, pricing, place, and marketing performance.

The 6 steps in the marketing process

These six steps take you from defining what you need to learn to collecting, analyzing, presenting, and applying the findings.

1. Identify the opportunity

Start by separating the business problem from the question your research can answer. In Marketing Research: Applied Insight, Daniel Nunan, David Birks, and Naresh Malhotra distinguish between a management decision problem—what the business needs to do—and a marketing research problem—what information you need to make that decision.

For example, declining sales identifies the symptom. Your research, then, needs to investigate the possible causes. Turn those into specific questions and, where appropriate, a measurable hypothesis before investing in surveys, interviews, or experiments.

That’s what Copper Cow Coffee Founder Debbie Wei Mullin did when she noticed customers were ordering only a couple of times a year, despite saying they loved the brand. She spoke directly with customers to understand the gap and learned they saw Copper Cow as a special-occasion coffee. That insight helped shape the company’s move into ground coffee designed for everyday drinking.

“If you have a product that you think the world needs, go and show the world and see what they say,” Debbie says.

2. Develop a research plan

Once you know what you need to learn, decide how you’re going to learn it. Your marketing research plan should set out the research design, the data you need, how you’ll collect it, who or what you’ll study, and practical constraints such as time and budget.

Key research categories

In Naresh K. Malhotra’s Marketing Research: An Applied Orientation, research designs fall into three core categories: exploratory, descriptive, and causal.

Exploratory research:

This category includes research that helps you investigate a problem you don’t yet understand well and identify possible explanations or opportunities. Think customer interviews, qualitative research, or existing data that reveals something worth investigating. If you sell on Shopify, your Behavior reports can provide useful starting signals. For example, the Searches with no results report shows exactly what shoppers looked for but couldn’t find.

Descriptive research:

This measures characteristics or behavior, including who buys, what they buy, how often, and how those patterns differ between groups. Shopify Analytics includes sales, customer, product, and behavior reports for this kind of analysis. The Customer cohort analysis report, for example, can compare retention, sales, average order value (AOV), and repeat purchases across cohorts.

Causal research:

This tests whether changing one variable causes a measurable change in another. For instance, an A/B test might ask whether a lower product price increases units sold enough to improve profit. For eligible stores, Shopify’s Smart Pricing early-access feature can run price experiments by randomly assigning the tested price to 50% of traffic and measuring outcomes such as selling price, conversion, and units sold.

Primary research vs. secondary research

Your research plan should also specify whether you’ll collect new data (primary research), use existing evidence (secondary research), or combine both. According to the Illinois Institute of Technology, the choice depends on factors such as how closely the data needs to fit your research question, how much control you need over its collection, and the time and resources available. Here’s how the two approaches differ:

Doe Lashes 'Founder Jason Wong used both primary and secondary research before launching the brand. He studied competing products and online beauty communities, then surveyed prospective customers, eventually collecting about 130 responses on the brands they used, what they liked and disliked, and what they wanted changed.

“The first thing I like to do is to see if there’s any history of that product. If someone has tried to create something, and if so, why did they fail? Why did they not get to market? Is it because of product defects, or is it because the market wasn’t ready for it? Or was it because the vendor just wasn’t able to supply what we envisioned?” says Jason.

He also bought about 20 pairs of lashes from different brands and ran informal product-study groups. Jason’s research paid off—Doe Beauty grew from a $500 startup into a multimillion-dollar business in less than five years.

3. Collect the data

At this step, put your research plan into practice. Keep in mind that your collection method also depends on the kind of data you need. According to Splunk, qualitative research is descriptive and typically non-numerical, used to explore the why and how behind behaviors—whereas quantitative research uses numerical data to measure the what, how many, or how much.

So, customer interviews, focus groups, and open-ended survey responses are common qualitative sources; closed-ended surveys, experiments, and analytics produce quantitative data. The two can, and often do, complement each other—qualitative responses can help explain patterns you first spot in quantitative data.

For example, jewelry brand Abbott Lyon uses Sidekick to identify high-value repeat customers in their Shopify data, then has their customer service team invite those customers to focus groups as the brand researches what distinguishes their most loyal buyers.

A survey is another research instrument, a structured set of questions used to collect standardized data from a sample of respondents, so that answers can be compared and analyzed at scale rather than one anecdote at a time.

Here are some best practices for collecting usable, high-quality data:

Write questions that produce usable answers

The questions’ wording can change the data you collect. Pew Research Center recommends:

• Ask one thing at a time. Avoid double-barreled questions such as asking customers to rate your product’s quality and value in a single response.
• Keep wording clear and neutral. Don’t build your preferred answer or an assumption into the question.
• Make response options mutually exclusive. A respondent should fit into only one category when only one answer is permitted.
• Make options exhaustive. Include all reasonable responses, with an “other” option where appropriate.
• Pretest new questions. Pew uses techniques such as focus groups, cognitive interviews, and pretesting to see how people understand questions before fielding them more widely.

Choose a sample size based on the decision

Start with the population you want to understand, the precision you need, and whether you plan to analyze smaller customer segments.

For probability samples, Penn State’s sample-size methodology calculates the required sample using your confidence level, desired margin of error, and expected population proportion. With no prior estimate, using 50% as the proportion gives a conservative calculation: at 95% confidence, roughly 385 completed responses produce a ±5 percentage-point margin of error, while about 601 are needed for ±4 points.

More responses alone don’t guarantee better research. The American Association for Public Opinion Research cautions that a larger sample isn’t necessarily better if the people surveyed don’t adequately represent the population. If you want reliable findings for individual customer segments, you’ll also need enough responses within each subgroup.

If you’re on Shopify, you can collect responses directly within the customer journey using survey apps from the Shopify App Store:

• Grapevine supports post-purchase, NPS, CSAT, attribution, and onsite surveys, with targeting by factors such as product, order value, country, and customer type.
• Fairing can connect post-purchase responses with order and customer lifetime value (CLV) data and analyze survey data within Shopify Analytics.
• POWR offers customizable surveys and polls with multiple-choice, rating, conditional-logic, and data-export options.

Backyard Butchers, for example, serves more than 600,000 customers across 48 states. The company switched to Shopify POS partly to capture cleaner first-party customer data across their pop-up locations and ecommerce stores. Once they had customer, order, and feedback data connected in one system, Backyard Butchers reported a 30% increase in CSAT and roughly 50% faster resolution of support tickets where information was tied to Shopify.

“Shopify makes knowing the customer easy, and if we leverage the data properly, the customer feels known,” says Tyler Medina, head ofmMarketing.

4. Analyze your data

Return to the research question and hypothesis you started with, then look for the patterns, differences, and relationships in your data that can help answer them. Here, analysis turns the responses and metrics you’ve collected into evidence you can interpret, but the technique you use depends on what you’re trying to learn.

According to IBM, there are four main types of business analytics:

• Descriptive analytics: What happened? Summarize historical or current data to establish the pattern you’re trying to understand. In Shopify reports, you can visualize metrics in charts, tables, heatmaps, and histograms, then compare periods or results against a target.
• Diagnostic analytics: Why did it happen? Break the result down by variables that could explain it. Shopify’s Analytics Insights is one tool that can help analyze sales, sessions, and fulfillment data daily and surface the five findings ranked highest by business impact.
• Predictive analytics: What might happen next? Use past and current data to estimate a future outcome. Shopify’s Inventory reports, for example, estimate how many days of stock remain for each product variant based on current inventory and its recent average daily sales rate, while Sidekick can go further into forecasting sales revenue for a specific period.
• Prescriptive analytics: What should you do next? Combine what you’ve learned with business constraints to choose an action. Sidekick can help you interrogate Shopify data in plain language, generate custom reports and visualizations, and work through possible next steps.

Analytics tools (like those listed above) can be incredibly useful for helping businesses make use of their data. Take Jaded London. Their lean ecommerce team was fielding 10 to 15 analytics requests a week, including questions about swimwear performance in Australia, inactive customers in Korea, and customer segments for a New York pop-up.

Now, they use Sidekick to ask questions about product performance, customer segments, and individual markets in plain language. The brand has also used city-level session, revenue, and customer data to evaluate locations for international pop-ups.

“I literally used Sidekick the other day to analyze potential markets … the data helps us make faster, more confident decisions,” says Jamie Evans, head of ecommerce.

Just keep in mind, while data analysis can be powerful, Shopify VP of Data Nell Thomas warns against running with an appealing interpretation before checking the methodology and underlying data: “People tell very powerful stories with data. So you’d better make sure the story is true.”

In other words, analysis can reveal an answer, but you still need to test whether the evidence supports the story you’re telling.

5. Present your results

Bring the research back to the question you started with. Your data presentation should make clear what you investigated, what the evidence shows, any limitations or uncertainty, and what the findings mean for the decision at hand.

The visualization should also match the point you’re trying to communicate. Shopify recommends different formats depending on the relationship you want readers to see:

• For trends over time. Line and area charts can show changes, seasonality, and period-over-period movement.
• For comparisons. Bar charts make differences across products, markets, channels, or other categories easier to scan.
• For precise values. Tables, lists, and number displays work when readers need exact figures or rankings.
• For patterns and customer journeys. Heatmaps, histograms, funnels, and cohort visualizations can reveal concentrations, distributions, drop-off, and retention over time.

Shopify expanded these options in Spring ’26, adding four new visualization types plus multimetric support for existing charts. Each shows a different kind of relationship in your data:

• Scatter plots compare two metrics to reveal relationships and outliers. You could plot order value against time of day to see when higher-value purchases cluster.
• Bubble charts add a third metric through bubble size. A retailer could compare gross and net sales while using bubble size to show discount value, making heavily discounted but poorly performing products easier to spot.
• Radar charts compare several metrics across the same products, customer segments, or markets. This format makes relative strengths, weaknesses, and imbalances easier to spot at a glance.
• Sunburst charts show hierarchical relationships. You might break gross sales down from product category into individual products, to see which items are carrying each category.
• Multimetric charts put several measures on the same timeline. Shopify line charts can display up to four metrics, making it easier to examine relationships among the data, such as whether a traffic surge is followed by higher sales or returns.

For example, a bar-and-line chart can put two sides of the same story on one timeline:

Once you know what you need to communicate, build the report around that finding. Shopify’s custom data explorations, available on all plans, let you choose the metrics, dimensions, comparisons, filters, and visualization that appear in a report. You can start from scratch or adapt an existing Shopify report, then save the finished version for future use. Sidekick can also generate or refine the underlying ShopifyQL query from a plain-language request.

6. Incorporate your findings

The final step is to decide what the evidence changes.

Ideally, you made that decision before seeing the results. In research, this is sometimes called an action standard, which is a predefined result or threshold that determines what you’ll do next. For example, you might decide in advance that if research shows one customer segment has substantially higher repeat-purchase intent, then you’ll develop a campaign specifically for that group.

From here, turn the finding into the appropriate business action. If your research changes who you want to reach or what you want to tell them, Shopify’s marketing tools can carry the insights into execution. For instance, customer segments let you define the audience, while Shopify Messaging can send targeted email and SMS campaigns. Shopify Campaigns can then organize activity across multiple channels and track campaign results against metrics such as sales, sessions, conversion rate, and average order value.

For eligible stores, Campaign Autopilot can take that process further. The tool analyzes your products, customers, sales history, and current marketing activity to recommend tactics across channels. You set the guardrails—including budget, target ROAS, regions, and languages—and decide whether campaigns need approval before they run.

Shopify Flow can automate the trigger itself. The Get Analytics data action lets you run scheduled ShopifyQL queries and use the result in workflow conditions. Sidekick can also build Flow workflows from a plain-language instruction, which you can review and test before activating.

For outdoor furniture brand POLYWOOD, research and analysis feed directly into experimentation. They use Sidekick for customer segment analysis, assortment strategy, and decisions about above-the-fold merchandising—while AI supports demand forecasting elsewhere in the business.

More than 150,000 product variations have contributed to the company’s rapid-testing culture rather than treating analyses as the end of the process.

A real-world marketing research process example: SURI

Before launching sustainable electric toothbrush brand SURI, Founders Gyve Safavi and Mark Rushmore wanted to understand whether there was room to challenge an electric toothbrush market dominated by two major brands.

This is how they followed the six steps in the marketing research process:

1. They identified the opportunity. SURI’s founders found that many consumers couldn’t name the toothbrush brand they already used and showed little loyalty to incumbent brands. This insight suggested market dominance didn’t necessarily translate into strong customer attachment.
2. They developed a research plan. The team combined surveys, in-person interviews, expert feedback, product testing, and pre-launch landing-page experiments.
3. They collected the data. SURI used SurveyMonkey and Attest for surveys, interviewed consumers in person, and sought feedback from dentists and sustainability experts. The founders went through roughly 20 product iterations before launch.
4. They analyzed the data. Their research surfaced customers’ recurring frustrations with existing electric toothbrushes, including short battery life, bulky travel cases, difficult-to-clean designs, and hard-to-recycle brush heads. SURI also tested different landing pages and advertising messages to see which drove email sign-ups.
5. They presented the results. The findings gave the founders a clear picture of the gaps they could address: Consumers wanted a more convenient, distinctive, and sustainable product, while pre-launch testing showed which messages resonated with prospective buyers.
6. They incorporated the findings. SURI designed a toothbrush with a 40-day battery, USB-C charging, a streamlined travel case, easier-to-clean construction, and recyclable brush heads. Plus, they refined their marketing before committing heavily to inventory.

The process was, to say the least, successful. SURI reached more than $30 million in sales within two years and secured retail partnerships with Boots in the UK and Erewhon in the US.

“I think a combination of thinking about your infrastructure to help you address customers [and] being proactive in your communication [were key],” says Cofounder Mark Rushmore. “We didn’t have a single refund as a result of that.”

Listen to the full story of how the founders built SURI into a multimillion-dollar brand through marketing research on an episode of Shopify Masters.