Desk research and the analysis of existing data: what you learn before fieldwork

1. What desk research is and why every project starts with it

Desk research, also called secondary research, is the analysis of data already collected by someone else for another purpose: national statistics, customs records, sector reports, publications, commercial databases, published prices, and the organisation's own internal data. Unlike primary research, surveys, focus groups, interviews, which produces new data for your question, desk research starts from what exists and asks: what is already known, how reliably, and what remains unknown.

It is the rational first step of any project for three reasons. It is fast and inexpensive: two to three weeks of analysis against two months of fieldwork. It avoids paying for existing information: the size of a market, the trend in imports or the demographic structure of a region are not asked in a survey. And it makes primary research better: a questionnaire written after desk research is shorter, more precise and asks only what could not be learned otherwise. The data analysis and desk research service covers exactly this phase.

2. Data sources: what exists and where

3. Assessing a source: five questions

A figure found in a report is not information until its origin is known. Every source goes through five questions, and the answers are noted in the report, next to the figure.

• Who produced the data? A statistical institute, a producers' association, a supplier selling a product, a consultant? The author's interest shapes definitions and selection.
• How were they collected? A census, a survey with a known sample, an estimate, an aggregation of other sources? A report that does not describe its method has described an opinion.
• When? The date of collection, not of publication. Data from 2023 published in 2025 describe 2023.
• With which definitions? Does "the soft drinks market" include water, juices, energy drinks or not? Two reports with different definitions give figures that differ two-fold for the same market.
• For what purpose? A report for investors emphasises growth; one for authorities emphasises problems. Neither necessarily lies, but both choose.

4. Which questions are answered without fieldwork

Desk research alone answers a surprisingly large share of the questions in a typical brief.

• How large the market is for an imported or locally produced category, and how it has evolved over the last five years. The top-down method described in the guide Market sizing and segmentation.
• Who the competitors are, where they import from, at what prices they enter and at what prices they sell, what assortment they carry.
• How many people are in the target population: households with children under 6, rural residents aged 60+, companies with 10-50 employees in a sector.
• How income and spending are distributed, and therefore how much of the budget goes, on average, to your category.
• What previous studies showed on the same topic, including those of international organisations, and which questions remained open.
• Who the company's current customers are and how they differ from the market, from internal data.
• What people search for online about the category, in which words, in which seasons.

What desk research does not answer: perceptions of your brand, reasons for choice, satisfaction, purchase intention at a given price, reaction to a concept. These require people asked directly, and the guide Which type of research solves your problem shows with which method.

5. Analysis: from tables to estimates

Collected data are not yet results. Analysis brings them to a common denominator, correlates them and turns them into estimates with a stated confidence interval. A few standard operations:

• Reconciling definitions: the same category, delimited identically across all sources, with what was included and excluded.
• Apparent consumption: production plus imports minus exports, by year, in quantity and value.
• Per capita and per household consumption, compared with similar countries to estimate growth potential.
• Time series: growth rates, seasonality, turning points and the events that explain them.
• Price analysis: the distribution of shelf prices by segment, the price index against competitors, the depth and frequency of promotions.
• Segmentation on internal data: customers by value, frequency, recency and product, as a hypothesis for market segmentation.
• Modelling: market development scenarios based on explicit assumptions about income, prices and penetration.

For large data sets, transactional sales, prices collected online, traffic, the analysis is done in programming tools, Python or R, which allow the cleaning, structuring and modelling of volumes that spreadsheets can no longer handle. The result remains the same, however: an estimate, with its method, its assumptions and its range of uncertainty, not a single figure.

6. The desk research report

A good desk research report has a fixed structure: the questions it answers; the sources used, each with its assessment; the results, with figures, ranges and the method of each estimate; the findings that cannot be established from existing data, listed explicitly; and the recommendation for the fieldwork phase, if needed: which population, which questions, which method. The last two sections are the most valuable: they turn desk research from a report into a plan.

All sources are cited precisely, so that any figure can be verified by the reader. A report that says "according to estimates, the market is X" without saying whose estimates and how they were made is not desk research; it is recycling.

7. Frequent mistakes

• The single figure. An estimate is taken from a report and built on, without triangulation.
• Recycled data. Three reports cite the same figure, which comes from a single old study; their convergence is false.
• The wrong definition. The customs code also includes products of no interest, and the market appears twice as large.
• Internal data taken for the market. The company's customers are not the market; they are the part of the market the company has already won, with its specific profile.
• Import price taken for market price. Between customs value and shelf price come taxes, logistics and margins, and the ratio between them varies by category.
• Skipping the phase. Fieldwork starts directly, and the survey asks things that were known and fails to ask things that could not be learned otherwise.

8. What to ask the agency

• The list of sources consulted, with the assessment of each on the five questions.
• The method of each estimate and its assumptions, written out, not just the figure.
• Triangulation of the key figures from independent sources.
• The category definitions, explicitly delimited.
• The section "what cannot be established from existing data".
• The recommendation for the fieldwork phase, with the questions that remain open.
• The data files and calculations, delivered together with the report.

Desk research is the part of research most organisations can partly do themselves, and the part most of them skip. Done systematically, with assessed sources and triangulated figures, it answers half the questions before the first interview and makes the other half cheaper and more precise. It does not replace fieldwork; it prepares it, and sometimes makes it unnecessary.

See also: Data analysis and desk research Market research and segmentation Market sizing and segmentation Which type of research solves your problem Decision support Methodology FAQ

The Perspicuus research team · 25 May 2026