Secondary Research and Market Intelligence

Introduction

Entrepreneurial decisions are rarely made in conditions of complete information. Before investing in product development, entering a market, selecting a customer segment, or responding to competitors, an entrepreneur must assemble evidence about demand, industry structure, regulation, technology, customers, and rivals. Secondary research is the systematic use of information that already exists, while market intelligence is the organized process of interpreting that information to support decisions.

Secondary research is not simply “searching online.” It involves defining an information need, locating appropriate sources, evaluating their credibility and relevance, identifying limitations and bias, triangulating conflicting evidence, and converting findings into defensible managerial judgments. At graduate level, the central challenge is not finding data but determining what the data can legitimately support.

This chapter presents a rigorous framework for locating, evaluating, and synthesizing government, academic, industry, commercial, digital, and competitor sources. It also considers data quality, evidence bias, source decay, ethical and legal issues, artificial intelligence, and the translation of market intelligence into entrepreneurial action.

Learning Objectives

By the end of this chapter, you should be able to:

1. Distinguish secondary research from primary research and market intelligence.
2. Classify major secondary-source types and select sources appropriate to a research question.
3. Evaluate authority, methodological quality, relevance, timeliness, transparency, and possible bias.
4. Synthesize apparently conflicting evidence without treating all sources as equally reliable.
5. Assess competitor, digital, commercial, academic, and public-sector information critically.
6. Document an evidence trail and use secondary findings in entrepreneurial decisions.
7. Recognize ethical, legal, privacy, and governance risks in market intelligence.

1. The Nature and Scope of Secondary Research

Primary research generates new data for a specific inquiry, such as interviews, surveys, observations, or experiments. Secondary research analyzes information collected previously by another person or organization, usually for a different purpose. Examples include a national census, a peer-reviewed article, a trade association report, a competitor’s annual report, app-store reviews, patent records, and web-traffic estimates.

Secondary data may be quantitative, qualitative, structured, or unstructured. A statistical table is structured quantitative data; customer reviews are usually unstructured qualitative data; a regulatory consultation may contain both. Secondary research therefore includes more than numerical market-size estimates.

Market intelligence is broader than secondary research. It includes the continuous collection, analysis, interpretation, and dissemination of information about an organization’s external environment. Secondary sources are a major input, but market intelligence may also include primary research, internal sales records, expert judgment, and monitoring of competitor behavior.

A useful distinction is:

Secondary evidence can support market definition, segmentation, demand estimation, trend analysis, competitor mapping, regulatory scanning, pricing analysis, and risk assessment. It cannot automatically prove that a particular product will succeed. Historical market growth, for example, does not establish future demand for an untested offering.

2. A Research Architecture for Entrepreneurs

High-quality secondary research begins with a decision, not a search engine. The researcher should state the decision to be informed, the population or market of interest, the time period, the geographic scope, and the level of certainty required.

A practical research architecture has six stages:

1. Frame the decision. Define the managerial choice and the consequences of being wrong.
2. Specify information requirements. Translate the decision into answerable questions.
3. Develop a source strategy. Identify likely source classes, search terms, and inclusion criteria.
4. Retrieve and record evidence. Save documents, metadata, quotations, measures, and access dates.
5. Appraise and synthesize. Evaluate quality, compare findings, and explain disagreements.
6. Convert evidence into action. State implications, uncertainties, assumptions, and next research steps.

For example, “Is the market attractive?” is too vague. A better set of questions might be: What customer problem is being addressed? How many potential users meet the proposed eligibility criteria? What alternatives do they currently use? What regulatory barriers affect entry? Which competitors serve the same use case? What evidence indicates willingness to pay?

A research protocol improves discipline. It may specify databases, search strings, publication dates, geographical boundaries, acceptable source types, quality thresholds, and procedures for resolving contradictory results. Graduate researchers should maintain a research log containing the search date, query, source, author, publisher, data period, method, relevant finding, limitation, and intended use.

3. Major Source Families

Government and intergovernmental sources

Government sources include censuses, labor-force surveys, business registers, customs statistics, regulatory filings, legislative documents, procurement databases, and economic indicators. Intergovernmental organizations may provide cross-country datasets and methodological standards.

Strengths include broad coverage, formal definitions, transparent publication procedures, and continuity over time. Limitations include publication lags, political pressures, revisions, missing informal activity, and definitions that may not match a start-up’s market. A government category such as “information services” may be much broader than a proposed software niche.

Researchers should inspect the statistical unit, sampling frame, coverage, nonresponse treatment, revision policy, seasonal adjustment, and whether values are nominal or inflation-adjusted. A figure reported in current currency should not be compared directly with a constant-price series without checking the price basis.

Academic sources

Academic sources include journal articles, scholarly books, conference papers, dissertations, and research datasets. Peer review can improve methodological quality, and academic work often makes assumptions and limitations explicit. It is particularly useful for theories of adoption, entrepreneurship, competitive strategy, consumer behavior, and research methods.

Peer review is not a guarantee of truth. Researchers should examine the design, sample, measures, statistical analysis, context, and whether the findings have been replicated. A study of university students in one country may have limited generalizability to business customers in another. Citation counts are indicators of attention, not necessarily of validity.

Industry and trade sources

Trade associations, professional bodies, specialist journals, standards organizations, conference materials, and consultancy publications can provide current sector terminology, operating benchmarks, regulatory interpretations, and practitioner insight. They may be more timely than official statistics.

However, trade sources can favor incumbents, promote a policy agenda, or define a market in a way that benefits sponsors. A trade association representing manufacturers may report industry optimism differently from an independent economic survey. Sponsorship, sample selection, and the commercial purpose of a report should be recorded.

Commercial sources

Commercial databases and market reports may offer forecasts, syndicated surveys, pricing information, panels, firmographic data, and estimates of market size. They can save time and provide specialized segmentation.

Commercial reports require special scrutiny. “Market size” may mean supplier revenue, retail sales, transaction value, or an estimate of total potential demand. Forecasts depend on assumptions that may be hidden behind a paywall. A polished chart does not substitute for a disclosed methodology. Researchers should ask who paid for the research, how the population was defined, how estimates were modeled, and whether the report separates observed data from analyst judgment.

Company and competitor sources

Annual reports, investor presentations, earnings calls, regulatory filings, product documentation, pricing pages, job advertisements, patents, app-store listings, customer-support materials, and press releases reveal strategic priorities and operational claims. They are valuable for identifying positioning, target segments, capabilities, partnerships, geographic expansion, and technology investments.

Company information is self-presentational. Revenue disclosures may be aggregated; product claims may use favorable comparisons; press releases describe intended rather than realized performance. Competitor evidence should therefore be treated as a statement of position or claim unless independently corroborated.

Digital and user-generated sources

Search trends, social-media discussions, online communities, reviews, creator content, web pages, app-store data, and digital advertising libraries can reveal language, complaints, emerging use cases, and competitor visibility. Digital data are often timely and granular.

They also create serious inference risks. Online participants are not necessarily representative of the market. Algorithms shape what is visible, duplicate accounts may exist, reviews may be incentivized, and platform metrics can change without notice. Search interest indicates attention, not necessarily purchase intent. Digital traces should generally be used for hypothesis generation and triangulated with stronger evidence.

4. Evaluating Source Quality

A source should be evaluated in relation to a claim, not labeled simply “good” or “bad.” A useful appraisal framework considers six dimensions.

A further criterion is provenance: can the researcher trace a number back to its original dataset or publication? Secondary reports often repeat one another. Ten websites citing the same consultancy estimate do not constitute ten independent sources.

Data quality dimensions

Data quality is multidimensional. Accuracy concerns closeness to the phenomenon being measured. Completeness concerns missing values, omitted groups, and coverage. Consistency concerns whether definitions and measurements remain comparable. Timeliness concerns the delay between observation and use. Validity concerns whether the measure actually represents the intended construct. Precision concerns the level of detail or statistical uncertainty.

A large sample can still have poor validity. For example, millions of page views may measure browsing behavior accurately while providing weak evidence about willingness to pay. Conversely, a small but carefully designed expert study may provide strong contextual validity while offering limited population estimates.

Bias and error

Important forms of bias include:

• Selection bias: observed cases differ systematically from the target population.
• Survivorship bias: analysis includes successful firms but excludes failures.
• Measurement bias: the instrument or definition systematically distorts results.
• Confirmation bias: researchers favor evidence supporting an existing belief.
• Publication bias: positive or novel findings are more likely to be disseminated.
• Commercial bias: funding or incentives influence framing, measures, or interpretation.
• Recency bias: recent evidence is overvalued even when it is less rigorous.
• Automation and platform bias: algorithms determine which content is collected or displayed.

Bias does not make every source unusable. It changes how the source should be interpreted. A competitor’s press release may be useful evidence of intended positioning, even if it is weak evidence of actual performance.

5. Search, Retrieval, and Evidence Management

Effective searching combines broad discovery with targeted verification. Start with conceptual terms, synonyms, industry classifications, customer language, product categories, and relevant geographic terms. Then search specific databases, official sites, academic indexes, regulatory repositories, and company documents.

Search operators can improve precision. Quotation marks identify exact phrases; a minus sign may exclude an unwanted term; site restrictions can focus a search on official domains; file-type restrictions can locate reports. Search strategies should be documented because search results are personalized, temporary, and influenced by ranking algorithms.

Evidence management should distinguish at least four items:

1. Claim: the proposition being considered.
2. Evidence: the exact source passage, table, or observation.
3. Interpretation: what the researcher believes the evidence means.
4. Decision implication: what action may follow, subject to uncertainty.

This separation prevents an analyst’s interpretation from being mistaken for a reported fact. A source matrix is useful:

6. Synthesis and Triangulation

Synthesis is more than summarizing sources one by one. It involves comparing evidence, identifying patterns, explaining contradictions, and constructing a conclusion proportional to the evidence.

Triangulation can occur across methods, sources, researchers, or theories. For instance, a demand hypothesis may be examined through official business counts, academic adoption research, competitor behavior, and customer complaints. Agreement across independent source types increases confidence, although it does not prove causality.

When evidence conflicts, the analyst should not calculate a simple average automatically. First ask whether the sources measure the same construct, population, period, geography, and unit. Differences may be definitional rather than substantive. A “market” measured by producer revenue will differ from one measured by consumer spending or addressable users.

A disciplined synthesis can use the following sequence:

1. Normalize definitions, currencies, price bases, time periods, and geographic scope.
2. Separate direct observations from forecasts and opinions.
3. Assess each source’s quality and likely direction of bias.
4. Identify convergence and disagreement.
5. Develop alternative explanations for disagreement.
6. State a conclusion with confidence level and unresolved questions.

Worked example: interpreting conflicting market estimates

Suppose Report A estimates a national market at $120 million in 2024, while Report B estimates $75 million. Report A defines the market as total transaction value, including marketplace transactions. Report B defines it as revenue retained by specialist suppliers. The difference is therefore partly a unit-of-analysis problem, not necessarily an error. If the entrepreneur’s revenue model captures only specialist supplier revenue, Report B may be more relevant for a revenue forecast, while Report A may be useful for assessing total customer spending.

The correct conclusion is not “the market is $97.5 million.” It is: “Two estimates use different definitions; the relevant baseline depends on the venture’s position in the value chain. Additional evidence is needed to estimate the portion realistically accessible to the proposed business.”

7. Market Sizing and the Limits of Secondary Evidence

Secondary data often support three related but distinct concepts:

• Total addressable market (TAM): the total theoretical demand if all relevant customers could be served.
• Serviceable available market (SAM): the portion matching the venture’s geography, capabilities, and offering.
• Serviceable obtainable market (SOM): the realistically capturable portion under competitive and operational constraints.

These are strategic constructs, not automatically observed facts. A top-down estimate may begin with an official industry total and apply defensible filters. A bottom-up estimate may multiply the number of target accounts by plausible annual spending. Neither approach is credible if the assumptions are hidden.

A basic bottom-up relationship is:

where N is the number of relevant customers, p is the average price per transaction, f is the expected number of transactions per customer per year, and s is the expected share captured by the venture. The units must be consistent: customers × currency/transaction × transactions/customer/year × share = currency/year. The formula is an estimating model, not a law; it assumes the inputs are defined for the same market and period.

For example, if 2,000 qualified business customers each make 3 purchases per year, the average price is $400, and the venture expects to capture 5% of transactions, estimated annual revenue is:

2,000 × 3 × $400 × 0.05 = $1,200,000.

This result is sensitive to customer count, purchase frequency, price, and capture rate. Sensitivity analysis should vary these assumptions rather than presenting a single number as a prediction.

8. Competitor and Industry Intelligence

Competitor analysis should begin by identifying alternatives, not merely firms with similar product descriptions. Direct competitors offer similar solutions to similar customers. Indirect competitors solve the same problem differently. Substitutes may include internal work, manual processes, postponement, or doing nothing.

A competitor profile can include customer segment, job-to-be-done, value proposition, product scope, price architecture, distribution, partnerships, capabilities, funding, geographic presence, evidence of traction, and vulnerabilities. The analysis should distinguish observed behavior from inferred intention.

Useful frameworks include industry-structure analysis, strategic-group mapping, value-chain analysis, business-model comparison, and technology or regulatory scanning. Frameworks are lenses, not substitutes for evidence. A five-forces assessment, for example, requires evidence about entry barriers, buyer concentration, supplier dependence, substitution, and rivalry; it should not be completed through unsupported adjectives such as “high” or “low.”

Competitor intelligence must remain lawful and ethical. Public filings, public websites, published interviews, and legitimately obtained reports are generally acceptable. Misrepresentation, unauthorized access, theft of confidential information, inducement of employees to disclose trade secrets, and misuse of personal data are not legitimate research methods.

9. A Detailed Illustrative Case: Reusable Packaging for Urban Restaurants

Background

This fictional case concerns LoopBox, a proposed venture supplying reusable takeaway containers to independent restaurants in a large metropolitan area. The founders initially believe that growing concern about plastic waste creates a substantial market opportunity.

Problem

They must decide whether to pilot the service and, if so, which restaurant segment to target. They need evidence about regulation, restaurant numbers, customer behavior, competitor offerings, logistics, costs, and willingness to participate.

Analysis

The team begins with official business-register data and municipal waste-policy documents. The business register reports 8,400 food-service establishments, but the category includes cafés, caterers, mobile vendors, and fine-dining businesses. The team therefore cannot treat 8,400 as its target market.

Academic studies suggest that convenience, hygiene assurance, and return friction influence reuse behavior. However, several studies concern household consumers rather than restaurant operations. The evidence is relevant for hypothesis development but not directly transferable to business adoption.

Competitor websites reveal three existing providers. Their public prices differ because one charges per container, one charges a monthly service fee, and one bundles washing and logistics. A trade report claims rapid sector growth, but the report is sponsored by a packaging manufacturer and does not disclose its sample frame. The team records it as directional evidence rather than a neutral forecast.

Digital reviews of reusable-container schemes identify recurring concerns about deposit refunds, container availability, and hygiene communication. These reviews are self-selected, but the repeated themes help the team design questions for subsequent interviews.

Approach and solution

The founders construct a source matrix and define a narrower SAM: independent restaurants offering takeaway meals within the service radius and with sufficient daily order volume. They triangulate official establishment counts with map listings, local business directories, and interviews. They create three scenarios rather than one forecast: conservative, base, and expansion.

They also identify a critical operational variable absent from most market reports: return density. A container system may be attractive at the city level but uneconomic if containers must be collected from widely dispersed locations. The proposed pilot therefore targets a dense restaurant district and measures return rates, washing cost, loss, customer complaints, and restaurant staff time.

Outcome

The research does not prove citywide demand. It does show that the dense district has a plausible pilot segment, that hygiene communication is essential, and that operational density is more decisive than general environmental interest. The venture delays broad expansion and tests a limited network with explicit performance thresholds.

Learning points

First, a large official category is not the same as an addressable market. Second, source relevance depends on the decision and unit of analysis. Third, competitor claims reveal positioning but require corroboration. Fourth, secondary research can identify key variables for primary research and experimentation. Finally, market intelligence is valuable when it changes the decision, not merely when it produces a larger report.

10. Ethics, Governance, and Regulation

Market intelligence may involve personal information, behavioral traces, employee information, or commercially sensitive material. Researchers should follow applicable privacy and data-protection law, terms of service, copyright rules, intellectual-property obligations, and institutional research policies.

Ethical practice includes data minimization, lawful collection, respect for reasonable expectations of privacy, secure storage, accurate attribution, and avoidance of deceptive identity. Public availability does not automatically make data ethically unrestricted. A public social-media post may still contain personal information whose aggregation creates a new privacy risk.

Researchers should preserve an audit trail: source version, retrieval date, transformations, coding decisions, exclusions, and analytical assumptions. If a report is revised, the original version should be retained where legally permitted. Decision makers should be told when evidence is provisional, modeled, sponsored, or affected by substantial uncertainty.

11. Technology and Artificial Intelligence

Digital tools can accelerate discovery, text classification, competitor monitoring, translation, deduplication, and extraction from large document collections. Generative AI can help produce search terms, summarize documents, compare definitions, and identify possible contradictions.

AI does not remove the need for source evaluation. It can hallucinate citations, merge incompatible definitions, reproduce web bias, overlook minority evidence, and present an uncertain inference confidently. A responsible workflow uses AI for assistance rather than unverified authority: retrieve the original source, inspect the relevant passage, verify dates and definitions, and document the tool’s role.

Automated monitoring also requires governance. Researchers should define alert thresholds, avoid excessive collection of personal data, review false positives, and ensure that automated classifications do not unfairly exclude firms or communities. The more consequential the decision, the greater the need for human review and explainability.

12. Best Practices for Graduate Researchers and Managers

A robust secondary-research process should:

• Begin with a clearly defined decision and information gap.
• Prefer original sources over unattributed summaries.
• Record definitions, dates, units, geography, and methodology.
• Separate facts, estimates, forecasts, interpretations, and assumptions.
• Triangulate across genuinely independent source types.
• Investigate disagreement rather than concealing it.
• Assess likely bias and the direction of its effect.
• Use sensitivity analysis for uncertain market-size assumptions.
• Treat digital signals as context unless representativeness is demonstrated.
• Preserve citations, versions, and an audit trail.
• Protect personal, confidential, and legally restricted information.
• State confidence levels and identify what primary research is still needed.

Chapter Summary

Secondary research uses existing information to answer entrepreneurial questions, while market intelligence organizes evidence into ongoing decision support. Effective research begins with a defined decision and proceeds through question specification, source selection, retrieval, appraisal, synthesis, and action.

Government, academic, industry, commercial, company, competitor, and digital sources each offer distinctive benefits and risks. Authority alone does not establish quality. Researchers must examine methodology, relevance, timeliness, transparency, provenance, coverage, validity, and incentives. Bias should be analyzed rather than treated as an all-or-nothing defect.

Triangulation requires comparison of definitions, populations, periods, units, and methods. Conflicting estimates should not be averaged mechanically. Market-size models such as TAM, SAM, and SOM are useful only when their boundaries and assumptions are explicit. Competitor intelligence should analyze alternatives and distinguish public claims from independently supported performance.

Ethical market intelligence respects privacy, intellectual property, confidentiality, legal restrictions, and research integrity. AI can improve retrieval and organization but cannot replace verification or judgment. The professional standard is an auditable evidence trail that connects source, claim, interpretation, uncertainty, and decision implication.

Key Terms

Secondary research: Analysis of information previously collected by another person or organization.

Market intelligence: Systematic collection, analysis, interpretation, and dissemination of information about an organization’s external environment.

Primary research: Collection of new data for a specific research purpose.

Triangulation: Comparison of evidence from multiple sources, methods, researchers, or theories.

Provenance: The traceable origin and history of a piece of data or claim.

Validity: The extent to which a measure represents the intended concept.

Reliability: The consistency of a measurement or procedure.

Selection bias: Systematic distortion caused by how cases enter the observed dataset.

Survivorship bias: Error caused by examining only cases that remain or succeed.

Measurement bias: Systematic distortion introduced by definitions, instruments, or procedures.

TAM: Total addressable market under an unconstrained theoretical definition.

SAM: Serviceable available market matching a venture’s scope and capabilities.

SOM: Serviceable obtainable market that can realistically be captured.

Competitor intelligence: Analysis of rival and substitute offerings, behaviors, capabilities, and strategies.

Source matrix: A structured record linking sources to claims, methods, limitations, and uses.

Data quality: The suitability of data in dimensions such as accuracy, completeness, consistency, timeliness, and validity.

Commercial bias: Distortion arising from financial incentives or sponsorship.

Audit trail: Documentation of sources, transformations, decisions, and analytical assumptions.

Digital trace: Data generated through online behavior, platforms, and digital interactions.

Evidence synthesis: Structured integration and interpretation of findings across sources.

Review Questions

Short Answer Questions

1. What is the difference between secondary research and market intelligence?
2. Identify three dimensions that should be considered when evaluating a source.
3. Why can a large sample still have poor validity?
4. What is the difference between TAM, SAM, and SOM?
5. Why should competitor press releases not automatically be treated as evidence of performance?

Long Answer Questions

1. Explain how a graduate researcher should design and document a secondary-research protocol for a new venture.
2. Compare government, academic, commercial, and digital sources in terms of strengths, limitations, and appropriate uses.
3. Explain how bias and data-quality problems can affect market-size estimates.
4. Discuss the role of triangulation in resolving conflicting secondary evidence.
5. Evaluate the ethical and governance issues involved in using digital data and artificial intelligence for market intelligence.

Analytical / Application Questions

1. Two reports estimate different market sizes because one measures consumer spending and the other measures supplier revenue. Design an analysis that determines which estimate is more useful for a proposed platform venture.
2. Develop a competitor-intelligence framework for a start-up entering an industry dominated by three established firms. Specify the sources, variables, quality checks, and ethical safeguards you would use.
3. A founder wants to rely on rapid growth in online search interest as proof of product demand. Critically assess this reasoning and propose a triangulated research design.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes secondary research from market intelligence?

A. Secondary research uses only government statistics, while market intelligence uses only interviews.

B. Secondary research analyzes existing information, while market intelligence organizes information to support ongoing decisions.

C. Secondary research is always quantitative, while market intelligence is always qualitative.

D. Secondary research studies competitors, while market intelligence studies customers only.

Correct Answer: B. Secondary research analyzes existing information, while market intelligence organizes information to support ongoing decisions.

Question 2

Which source characteristic concerns whether a measure actually represents the intended concept?

A. Timeliness

B. Authority

C. Validity

D. Completeness

Correct Answer: C. Validity

Question 3

Why should two market estimates not automatically be averaged?

A. All market estimates are necessarily inaccurate.

B. Different estimates may use different definitions, populations, periods, or units.

C. Averages can only be used with government data.

D. Commercial reports are always more accurate than official reports.

Correct Answer: B. Different estimates may use different definitions, populations, periods, or units.

Question 4

Which is an example of survivorship bias?

A. Using a survey with too few responses from rural customers.

B. Using only successful start-ups to identify factors associated with entrepreneurial success.

C. Using an outdated price list in a current analysis.

D. Comparing values stated in different currencies without conversion.

Correct Answer: B. Using only successful start-ups to identify factors associated with entrepreneurial success.

Question 5

In the formula Estimated annual revenue = N × p × f × s, what does f represent?

A. The number of relevant customers

B. The average price per transaction

C. The expected number of transactions per customer per year

D. The expected share captured by the venture

Correct Answer: C. The expected number of transactions per customer per year

Question 6

Which statement about digital search interest is most defensible?

A. It proves that users will purchase the product.

B. It is irrelevant because digital data cannot be used in research.

C. It can generate hypotheses but should usually be triangulated with other evidence.

D. It is always representative of the total market.

Correct Answer: C. It can generate hypotheses but should usually be triangulated with other evidence.

Question 7

What is the primary purpose of a source matrix?

A. To replace all qualitative analysis with numerical scoring.

B. To link sources to claims, methods, limitations, and intended uses.

C. To rank competitors by their advertising expenditure only.

D. To guarantee that every source is unbiased.

Correct Answer: B. To link sources to claims, methods, limitations, and intended uses.

Question 8

Which practice is most consistent with ethical competitor intelligence?

A. Obtaining confidential information through deception.

B. Accessing restricted systems to retrieve pricing data.

C. Analyzing public filings, published materials, and legitimately obtained reports.

D. Impersonating a customer to obtain nonpublic documents.

Correct Answer: C. Analyzing public filings, published materials, and legitimately obtained reports.

Question 9

If 2,000 customers make 3 purchases per year at $400 per purchase and a venture captures 5%, what annual revenue is estimated?

A. $120,000

B. $400,000

C. $1,200,000

D. $2,400,000

Correct Answer: C. $1,200,000

Question 10

What is a major risk when using generative AI in secondary research?

A. It makes all official statistics inaccessible.

B. It can produce unsupported or fabricated citations and confident but incorrect summaries.

C. It prevents researchers from recording search dates.

D. It guarantees that source bias is removed.

Correct Answer: B. It can produce unsupported or fabricated citations and confident but incorrect summaries.