Market analysis falls into four broad categories: market research (primary and secondary), marketing analytics, strategic frameworks such as PESTLE and Porter's Five Forces, and advanced analytics including predictive and prescriptive methods. Primary research answers product design and customer validation questions. Strategic frameworks suit market entry and competitive positioning. Predictive analytics handles pricing and demand forecasting. Platforms like Ontherice add a fifth layer: continuous AI signal detection that spots nascent trends before they surface in conventional studies.
Market intelligence is broader than any single study. Leading firms now treat it as an ongoing capability rather than a one-off project, combining internal CRM data with external signals and AI to monitor trends in near-real time. That shift changes which methods you reach for first.
Key takeaways
Matching your business question to the right type of market analysis is the single most important decision before any study begins. Rigour without relevance wastes budget; speed without data quality produces confident mistakes.
| Point | Details |
|---|---|
| Match method to question | Use primary research for validation, strategic frameworks for positioning, and predictive analytics for forecasting. |
| Validate secondary data | Check original collector, date, sample frame, and methodology before citing any secondary source. |
| Size the market in three layers | TAM, SAM, and SOM give progressively realistic estimates; state your assumptions explicitly in every forecast. |
| Follow the six-step process | Industry research, competitor review, gap identification, target definition, barriers assessment, and sales forecast produce a complete, repeatable analysis. |
| Ontherice for continuous monitoring | Ontherice's AI signal detection surfaces early trends between discrete studies, shortening time-to-insight across sectors. |
Table of Contents
- What are the main types of market analysis?
- Primary methods and practical techniques
- How do descriptive, diagnostic, predictive, and prescriptive analytics differ?
- How to apply PESTLE, Porter's Five Forces, and SWOT
- Market sizing, segmentation, and pricing: what to measure
- How to run a market analysis: a practical six-step process
- Common pitfalls and data-quality checks
- How AI-driven signal detection complements traditional analysis
- The trade-off between speed and rigour is real, but manageable
- Ontherice gives you continuous market intelligence, not just a snapshot
- Sources
What are the main types of market analysis?
The three terms that cause the most confusion are market research, marketing analytics, and market intelligence. They overlap, but each answers a different question.
Market research is project-based. You commission it to answer a specific question: is there demand for this product? What price will the market bear? It draws on primary data (collected directly from respondents) or secondary data (existing published sources). According to Investopedia, market research splits cleanly into primary methods such as surveys, interviews, and focus groups, and secondary methods such as industry reports and government statistics, each with characteristic trade-offs in cost, speed, and depth.
Marketing analytics is continuous and operational. It measures campaign performance, customer behaviour, conversion rates, and channel ROI using tools such as Google Analytics 4, HubSpot, or Salesforce. The data is mostly internal and near-real-time.
Market intelligence is the broadest category. The Pragmatic Institute defines it as encompassing four core types: competitive intelligence, product intelligence, market understanding, and customer understanding. Unlike project-based research, market intelligence is continuous and forward-looking.
Cutting across all three categories are four analytic approaches that determine what kind of answer you get:
| Analytic type | Business question it answers | Typical data sources |
|---|---|---|
| Descriptive | What happened? | Sales data, web analytics, CRM reports |
| Diagnostic | Why did it happen? | Cohort analysis, survey responses, session recordings |
| Predictive | What is likely to happen? | Panel data, ML models, historical transactional data |
| Prescriptive | What should we do about it? | Optimisation models, A/B test results, scenario planning |
| Sentiment / social listening | How do people feel about it? | Social feeds, review platforms, NLP-processed text |
Mapping your business question to one of these rows before choosing a method saves considerable time and budget.
Primary methods and practical techniques
Market research methods range from structured surveys to ethnographic fieldwork. Each reveals something different, and the right choice depends on what you need to know, how quickly, and at what cost.
- Surveys deliver broad quantitative data quickly. Online surveys via tools such as Typeform or Google Forms can reach hundreds of respondents within days at low cost. Best for measuring awareness, satisfaction, or stated preferences across a large sample. Weakness: self-reported data can diverge from actual behaviour.
- One-to-one interviews go deep. A 45-minute structured interview with a target customer uncovers motivations, objections, and language that no survey can capture. Time-intensive and expensive per respondent, but irreplaceable for early-stage product discovery.
- Focus groups sit between surveys and interviews. Six to ten participants discuss a topic together, generating reactions and group dynamics that individual interviews miss. Useful for concept testing and messaging. Risk: dominant voices can skew the output.
- Ethnographic and observational research watches people in context rather than asking them to describe behaviour. A researcher shadowing a warehouse manager for a day will spot workflow friction that no questionnaire would surface.
- Field trials and test marketing expose a real product to a limited market before full launch. Supermarkets in the UK regularly pilot new lines in selected regions before national rollout. Expensive but produces genuine purchase data rather than stated intent.
- Social listening and sentiment analysis monitor public conversations on platforms such as Reddit, X (formerly Twitter), and review sites. Tools such as Brandwatch or Sprout Social aggregate and classify mentions in real time. Particularly useful for brand monitoring and early trend detection.
- Secondary data and desk research draw on existing sources: ONS datasets, Mintel reports, Companies House filings, academic journals, and trade association publications. Fast and cheap, but the data may be dated or not precisely matched to your market definition.
Pro Tip: The single biggest source of bias in primary research is a leading question. Before fielding a survey, run every question through a neutral colleague who has no stake in the outcome. If they can guess the "right" answer from the phrasing, rewrite it.
How do descriptive, diagnostic, predictive, and prescriptive analytics differ?
Each analytic approach requires a different data quality threshold and produces a different type of output.
Descriptive analytics is the baseline. It summarises what has already happened: revenue by region last quarter, website sessions by channel, or product return rates. Every business already does this, often without calling it analytics. The data requirement is low: clean, consistent historical records are enough.
Diagnostic analytics goes one step further by asking why a pattern occurred. A sudden drop in conversion rate might be explained by a price change, a competitor promotion, or a broken checkout flow. Cohort analysis, funnel breakdowns, and customer surveys are the typical tools. The data requirement rises: you need enough granularity to isolate variables.
Predictive analytics uses historical patterns to estimate future outcomes. Demand forecasting, churn prediction, and pricing elasticity models all fall here. Machine learning models trained on transactional or panel data are common. The data requirement is high: volume, variety, and cleanliness all matter, and a poorly trained model can produce confident-sounding nonsense.
Prescriptive analytics goes furthest, recommending a specific action. Yield management systems in airlines and dynamic pricing engines in e-commerce are prescriptive. They combine predictive outputs with optimisation logic to suggest the action that maximises a defined objective.
Sentiment analysis occupies a distinct lane. It applies natural language processing (NLP) to unstructured text, classifying opinions as positive, negative, or neutral. Where quantitative analytics tells you how many customers churned, sentiment analysis tells you what they said on the way out. The two together are more powerful than either alone.
The AMA's market analysis guide makes a related point: market analysis is strongest when quantitative sizing and qualitative motives are combined. Conjoint analysis and discrete-choice methods are particularly useful for fine-grained product trade-offs once you have the broader picture.
How to apply PESTLE, Porter's Five Forces, and SWOT
Strategic frameworks are not checklists. The BDA Global guidance on market intelligence is explicit on this: PESTLE and Porter's Five Forces should be applied dynamically to interpret how macro-forces interact with an organisation's specific competitive position. Ticking boxes produces a document; interpreting implications produces a decision.
PESTLE scans the macro-environment across six dimensions: Political, Economic, Social, Technological, Legal, and Environmental. Its value is in forcing you to look outside your industry. A UK food manufacturer running a PESTLE in 2025 would flag post-Brexit import regulations (Political/Legal), rising energy costs (Economic), growing consumer demand for low-carbon packaging (Environmental/Social), and automation in logistics (Technological). Each signal then feeds a question: does this change our cost base, our supplier strategy, or our product positioning?
Porter's Five Forces zooms in to industry structure. The five forces are: competitive rivalry, threat of new entrants, threat of substitutes, bargaining power of buyers, and bargaining power of suppliers. For a UK SaaS company entering the HR software market, the analysis might reveal high competitive rivalry (many incumbents), low barriers to entry (cloud infrastructure is cheap), and high buyer power (large enterprises negotiate hard on price). That combination argues for a niche positioning strategy rather than a direct assault on the market leaders.
SWOT bridges external analysis and internal capability. It is most useful after PESTLE and Porter's Five Forces have mapped the environment, because Strengths and Weaknesses only mean something relative to the Opportunities and Threats the environment presents. Running SWOT first, without that context, tends to produce generic lists. For a practical comparison of when to use SWOT versus PESTLE, this MOGHQ guide is worth reading.
The practical discipline is to turn framework outputs into named strategic options. A PESTLE signal about rising energy costs is not an insight until it becomes: "We should lock in a three-year energy contract before Q3" or "We should accelerate our shift to lower-energy production methods." Frameworks that stop at observation are half-finished.
Market sizing, segmentation, and pricing: what to measure
Market sizing gives a number to the opportunity. The standard framework uses three nested estimates:
| Measure | Definition | Quick calculation |
|---|---|---|
| TAM (Total Addressable Market) | The total global or national demand for a product category | Annual units sold × average price, or revenue per customer × total potential customers |
| SAM (Serviceable Addressable Market) | The portion of TAM your business model and geography can realistically reach | TAM × proportion matching your target segment and delivery capability |
| SOM (Serviceable Obtainable Market) | The share of SAM you can realistically capture given competition and resources | SAM × realistic market share percentage based on competitive analysis |
Wikipedia's market analysis entry draws a useful distinction between market volume (realised sales) and market potential (the upper limit of demand). TAM approximates market potential; your current revenue sits somewhere inside SOM. The gap between the two is where growth strategy lives.
Segmentation breaks the market into groups that behave differently. The three most useful segmentation bases for most businesses are:
- Demographic: age, income, geography, company size (for B2B). Fast to measure, but a weak predictor of purchase behaviour on its own.
- Behavioural: purchase frequency, channel preference, feature usage. Stronger predictor, but requires transaction or product data to calculate.
- Value-based: lifetime value, willingness to pay, strategic importance. The most commercially useful segmentation, but requires modelling rather than simple counting.
Pricing analysis feeds directly from segmentation. Understanding what different segments will pay, and why, is more useful than a single average price point. Conjoint analysis, van Westendorp price sensitivity surveys, and competitive price benchmarking are the standard tools. For technology sector benchmarking, StockMarketCap's technology sector data provides useful market cap context when sizing relative competitive positions.
For forecasting, the simplest defensible approach is a bottom-up model: estimate the number of reachable customers in your SAM, apply a realistic conversion rate based on comparable products or your own pilot data, and multiply by average revenue per customer. State your assumptions explicitly. A forecast with named assumptions is far more useful than a precise-looking number with hidden ones.
How to run a market analysis: a practical six-step process
Coursera's market analysis framework outlines six steps that cover the full scope of a practical analysis. Here they are with the deliverables each step should produce:
-
Research the industry. Gather data on market size, growth rate, key players, and macro trends. Use ONS statistics, trade association reports, and industry databases. Deliverable: industry overview document with TAM estimate and growth rate.
-
Investigate competitors. Map direct and indirect competitors across pricing, positioning, product features, and customer reviews. A structured competitor matrix (company, price point, key differentiator, customer segment, perceived weakness) is the standard output. Deliverable: competitor matrix spreadsheet.
-
Identify gaps. Cross-reference customer needs (from primary research) with what competitors currently offer. Gaps are where unmet demand exists. Deliverable: gap analysis table linking customer pain points to unserved or underserved needs.
-
Define the target market. Combine segmentation analysis with gap findings to specify the customer group most likely to buy and most valuable to serve. Build one or two customer personas with demographic, behavioural, and motivational attributes. Deliverable: target market definition and persona documents.
-
Assess barriers to entry. Consider regulatory requirements, capital requirements, switching costs, and incumbent advantages. For UK businesses, this includes sector-specific FCA or CMA considerations where relevant. Deliverable: barriers assessment with mitigation notes.
-
Build a sales forecast. Use the bottom-up approach described in the previous section. Produce three scenarios: conservative, base, and optimistic, each with named assumptions. Deliverable: three-scenario forecast model with assumption log.
For M&A contexts, Compass Business Acquisitions' pre-acquisition analysis guide extends this process with due diligence steps specific to buying an existing business, including customer concentration risk and supplier dependency checks.
Common pitfalls and data-quality checks
Most market analysis errors are not statistical. They are structural: the wrong question was asked, the wrong sample was used, or the wrong source was trusted.
Unrepresentative samples are the most common failure in primary research. A survey of 200 LinkedIn followers of a B2B SaaS company will over-represent early adopters and under-represent the mainstream buyers the company actually needs to reach. Always define the target population before designing the sample, not after.
Survivorship bias distorts competitive analysis. Studying only the companies that succeeded in a market tells you what winners look like, not what the market actually selects for. Include failed entrants in your competitive review.
Confirmation bias is the subtler problem. Teams that commission research to validate a decision they have already made tend to design studies that find what they are looking for. An independent reviewer of the research design, before fieldwork begins, is a cheap insurance policy.
Overfitting predictive models produces models that describe historical data perfectly and forecast future data poorly. If a model has more variables than it has data points to train on, treat its outputs with scepticism.
Poor secondary data is widespread. Many industry reports recycle figures from earlier reports without updating the underlying data. Before trusting a secondary source, check: Who collected the original data? When? What was the sample frame? Is the methodology described?
Pro Tip: Apply a four-point check to every secondary source before citing it: (1) identify the original data collector, not just the report publisher; (2) confirm the publication or data collection date is within the last three years for fast-moving markets; (3) check whether the sample frame matches your target market; (4) look for a methodology section. A source with no methodology section is a red flag.
How AI-driven signal detection complements traditional analysis
Traditional market analysis is discrete: you commission a study, collect data over weeks or months, and produce a report. By the time the report lands, the market may have moved. AI-driven market intelligence addresses this by combining internal and external data sources for continuous, near-real-time monitoring of trends and competitor activity.
The value of AI signal detection is highest in three situations. First, when you need to monitor a large number of signals simultaneously across sectors, geographies, or product categories. Second, when speed of detection is a competitive advantage, as it often is in technology, consumer goods, and financial markets.
Early signal detection does not replace primary research. It identifies where to look. A rising signal in AI-assisted diagnostics, for example, tells a medical device company that the category is gaining momentum. It does not tell them which specific product features customers want or what price they will pay. That still requires interviews and conjoint analysis. The two approaches are complementary, not substitutes.
Ontherice operates precisely at this intersection. Its AI engines scan global public data across domains including finance, technology, products, and brands, producing ranked signals and sector leaderboards. The platform's early trend discovery capability is designed to shorten the time between a trend emerging and a business acting on it. A practical workflow: use Ontherice signals to identify categories worth investigating, then deploy targeted primary research to validate the commercial opportunity. For a broader view of how market intelligence trends are shifting in 2026, the Ontherice blog covers the integration of AI monitoring with traditional methods in detail.
Pro Tip: Treat AI-generated signals as hypotheses, not conclusions. When a signal ranks highly, ask: what primary research would confirm or refute this? A brief set of customer interviews or a targeted survey is usually enough to separate genuine trends from noise.
The trade-off between speed and rigour is real, but manageable
The most common mistake practitioners make is treating method selection as a binary: either you do rigorous research that takes three months, or you move fast and trust your gut. Neither extreme serves a business well.
The six-step process described above is not a three-month project by default. Steps one and two (industry research and competitor review) can be completed in a week using secondary data and desk research. Steps three and four (gap identification and target definition) take longer if they require primary research, but a focused set of ten customer interviews can be completed in two weeks and will answer most of the critical questions. The full process, done with discipline, typically takes four to six weeks for a new market entry decision.
What matters most is repeatability. A documented process with named assumptions and clear data sources can be updated as the market changes. A one-off study that lives in a single slide deck cannot. The teams that make the best market decisions are not the ones with the biggest research budgets; they are the ones with the most consistent analytical habits.
Combining methods is where the real gains are. Quantitative sizing tells you how big the opportunity is. Qualitative research tells you why customers behave as they do. AI signal detection tells you where the market is moving before the data fully confirms it. None of these replaces the others, and the step-by-step market intelligence guide on the Ontherice blog is a useful companion for translating these frameworks into practitioner tasks.
Ontherice gives you continuous market intelligence, not just a snapshot
Most market analysis tools hand you a report and leave you to monitor the market manually. Ontherice works differently: its AI engines run continuously, scanning global data to surface ranked signals and early-trend alerts across finance, technology, products, brands, and more.
For professionals and students who have just mapped their analytical approach using the methods above, Ontherice fills the gap that discrete research cannot: the space between studies, where markets actually move. The platform's AI-detected opportunities feed surfaces signals before they appear in industry reports. The Rankings Generator Engine produces sector leaderboards so you can prioritise which trends merit deeper investigation. Core content is free; Access Points unlock premium signal cards and advanced feeds when you need greater depth.
Visit Ontherice to explore the signals feed and see which trends are gaining momentum in your sector right now.
Sources
- What Is Market Intelligence? Definition, Frameworks & Examples
- How to conduct a market analysis: a complete guide for businesses and marketers | AMA
- Market Analysis: What It Is and How to Conduct One | Coursera