Downloading a table from the Local Data Bank takes five minutes, but drawing the correct conclusion from it can take several days – and this is precisely the stage at which errors most often occur. GUS (Statistics Poland), Eurostat, and NBP data in market research provide the foundation for desk research, but using them carelessly leads to distorted market size estimates, incorrect segmentation, and conclusions that cannot be defended before a management board.<\/p>\n
Public statistics are useful wherever a benchmark is needed: the demographic structure of the population, the number of business entities in a given PKD classification, the growth rate of sold industrial production, wage levels, exchange rates, the balance of payments, core inflation indicators, or foreign trade data. GUS provides a picture of Poland at the national, voivodeship, county, and municipality levels. Eurostat makes it possible to compare member states using harmonized methodologies. NBP is responsible for monetary, exchange rate, banking sector, and balance-of-payments data.<\/p>\n
GUS, Eurostat, and NBP data in market research serve three functions. First, they define the population frame – in other words, they answer the question “how many are there?” before the researcher asks “what are they like?” Second, they are used to weight and calibrate samples in quantitative research so that the respondent structure reflects the distribution in the population. Third, they provide macro-level context for interpreting primary research findings – without it, it is difficult to distinguish a seasonal effect from a lasting change in behavior.<\/p>\n
The problem begins when an analyst treats public statistics as a ready-made answer rather than as input material. GUS supports market research with structural data, but it does not explain motivations, attitudes, or purchasing decisions. Eurostat shows comparable aggregates, but it conceals differences in national definitions beneath a common denominator. NBP reports financial flows, but it does not explain why companies have changed their lending policies.<\/p>\n
Working with secondary data requires methodological discipline similar to that applied when designing quantitative research. Each source has its own metadata: what exactly it measures, how the unit has been defined, what the reference period is, what level of aggregation is used, and when the data were last updated.<\/p>\n
In practice, desk research using secondary data from GUS, Eurostat, and NBP proceeds through several steps that should be completed in the appropriate order:<\/p>\n
As Hume’s Institute experts point out, GUS and Eurostat data are a starting point, not an answer – only combining them with primary research provides a complete market picture, because public statistics indicate “what is happening,” while qualitative or quantitative research explains “why” and “what this means for consumer decision-making.”<\/p>\n
A specific example is an analysis of the market for care services for older people. Eurostat statistical data will show the demographic structure and population aging projections in EU countries. GUS will provide the number of single-person households in age groups above 65, as well as data on social benefits. NBP can supplement the analysis with information on household deposits and savings and, in selected cases, also on remittances from abroad. However, only primary research – interviews with caregivers, a survey among families, and ethnography in care facilities – will answer the question of who actually makes the purchasing decision, which price barriers are critical, and which communication channels reach decision-makers.<\/p>\n
Similarly, in B2B projects, the REGON register provides the number of entities registered under a selected PKD classification, but it does not indicate how many of them actually conduct operational activity in a given product category. This information can only be obtained through telephone screening or a CATI survey conducted on a random sample drawn from the sampling frame.<\/p>\n
Public statistics are reliable, but their users are not always. In Hume’s Institute projects, a recurring set of pitfalls can be observed that lead to incorrect conclusions even when the source data are accurate.<\/p>\n
Confusing active entities with registered entities.<\/strong> The REGON register includes all registered units, not only those that are economically active. Estimating the potential B2B market based on the raw number of registrations inflates the population by as much as several dozen percent in some industries.<\/p>\n Unwittingly comparing time series based on different classifications.<\/strong> The change from PKD 2004 to PKD 2007, NACE revisions, and COICOP updates – each of these means that data from before and after the change are not directly comparable without correspondence tables.<\/p>\n Ignoring the informal economy and self-employment.<\/strong> Some consumer service categories, such as beauty services, tutoring, minor repairs, and care services, are underestimated in public statistics. A conclusion about a “small market” drawn solely from GUS data may therefore be misleading.<\/p>\n Confusing correlation with mechanism.<\/strong> An increase in indicator X and a decrease in Y over the same period do not indicate a causal relationship. Public statistics do not replace hypothesis testing or econometric models.<\/p>\n Overlooking methodological differences between countries in Eurostat data.<\/strong> Although Eurostat statistical data are harmonized, in some areas, such as labor force surveys and education statistics, national definitions affect comparability. This requires reading methodological notes, not only tables.<\/p>\n Using nominal rather than real data.<\/strong> Comparing year-on-year retail sales values without adjusting for inflation leads to overstated assessments of growth.<\/p>\n Old time series with a new interpretation.<\/strong> Data from before the pandemic, before Russia’s full-scale invasion of Ukraine, or before a regulatory change do not describe the current market. Yet some industry reports continue to cite them as current.<\/p>\n Lack of triangulation.<\/strong> A single source, even a reputable one, provides only one slice of the picture. Triangulating GUS, Eurostat, and NBP data with industry data and primary research is a standard part of sound research practice, not excessive caution.<\/p>\n Before including a time series in a report or model, it is worth going through a short checklist that eliminates most input-stage errors:<\/p>\n This checklist does not replace subject-matter expertise, but it protects against the most costly errors – those that become apparent only when a report is discussed with the management board or regulator.<\/p>\n GUS publishes data from censuses and representative surveys covering demographics, the labor market, wages, prices, production, trade, construction, transport, education, healthcare, culture, and household living conditions. Data are available at the national level and in territorial breakdowns, including NUTS, county, and municipality levels, through the Local Data Bank. GUS also maintains administrative and official registers, including REGON and TERYT, and publishes census results.<\/p>\n Public statistics do not contain information on attitudes, motivations, brand preferences, loyalty, customer journeys, or detailed behavioral segmentation. They also lack data on niche product categories, the informal economy, and certain digital services. These gaps are filled by primary research – quantitative, qualitative, and mixed-methods research.<\/p>\n The key is consistency in definitions, periods, and levels of aggregation, as well as the informed use of correspondence tables between classifications. It is worth building a single working table in which each time series has its own metadata: source, definition, unit, period, and download date. Triangulation – comparing GUS, Eurostat, NBP, and industry data – makes it possible to identify discrepancies that point to methodological differences or errors in raw data.<\/p>\nWhat should be checked before using data in an analysis?<\/h2>\n
\n
Frequently asked questions<\/h2>\n
What data does GUS provide?<\/h3>\n
What is missing from public data?<\/h3>\n
How should data from different statistical sources be combined?<\/h3>\n