Pesquisa de Preços Lite is the public, simplified version of the Brazilian Federal Government's official tool for consulting values from procurements recorded in Compras.gov.br. It can be accessed without registration, login, or password. The benefit is broader access for public managers, suppliers, researchers, and citizens; the caution is not to confuse an automatic calculation with a ready-made conclusion about what the estimated price should be.
The tool works with a 12-month history and displays mean, median, lowest and highest price, standard deviation, and coefficient of variation. It also allows users to apply filters, decide which procurements are included in the calculations, and generate reports or export the data. The safest workflow is sequential: identify the item correctly, refine the sample, evaluate dispersion, choose the calculation method, and only then export the result.
What you can consult without logging in
According to the official Pesquisa de Preços Lite page, anyone can consult reference values from procurements carried out in Compras.gov.br without authentication. The search uses the system's official catalog: materials are linked to CATMAT and services to CATSER.
In the workflow presented in the official June 2026 tutorial, the user creates a new search, enters basic information, adds a material or service, selects the appropriate catalog item, and defines quantity and supply unit. From there, the platform retrieves procurements used to build the sample.
Each record can be expanded to view information such as process number, procurement method, supplier, product brand, and award criterion. The tool also displays the lowest and highest price found and statistics calculated over the quotes that remain selected to compose the search.
Step by step: from search to export
- Find the correct item. Search for a short description and confirm that the result matches the intended material or service and the appropriate catalog code.
- Check quantity and supply unit. Comparing incompatible units can produce a numerically large set that is editorially and technically weak.
- Apply filters before looking at the reference value. Lite offers filters related to procurement location, awarded values, and quantity ranges.
- Review the retrieved procurements. Open records when necessary and check whether the object, scale, and conditions are comparable to the need being researched.
- Observe dispersion. Compare mean and median, read the standard deviation and coefficient of variation, and investigate extremes before choosing the method.
- Define which records compose the calculation. The “Compor” column allows a quote to be enabled or disabled; records removed from the composition stop affecting the indicators.
- Choose the calculation method. The system allows switching among lowest price, mean, and median for the reference price.
- Export before leaving. In the summary tab, users can generate a summarized or detailed PDF report and export the data as CSV.
Why filters come before mean and median
A statistic only summarizes the sample it receives. If the sample mixes objects, units, quantities, or conditions that are not sufficiently comparable, mean and median merely compress that mixture. Therefore, the first task is not to “choose the mean”; it is to build a sample that makes sense.
Filters for location, awarded value, and quantity range help bring records closer to the situation being analyzed. They do not replace reading the object description or conducting a critical review. A price may be real and officially awarded and still have been practiced at a scale, location, brand, or commercial condition that does not represent the new procurement well.
Mean, median, or lowest price: what each measure answers
| Method | What it does | When it requires more attention |
|---|---|---|
| Mean | Adds the considered prices and divides by the number of observations. It uses information from every value. | It is sensitive to extreme values; a few very high or very low quotes can shift the result. |
| Median | Orders the prices and identifies the central point of the sample. | It is more resistant to extremes, but it still depends on the sample being comparable and relevant. |
| Lowest price | Uses the lowest value among the records considered. | It requires checking whether the value is feasible, comparable, and not caused by a specific condition that will not recur. |
The Brazilian Federal Court of Accounts guidance on IN SEGES/ME No. 65/2021 notes that the mean, median, or lowest value may be used to obtain the estimated price, with critical analysis of the data and justified exclusion of unfeasible, inconsistent, or excessively high values. In procedures subject to the instruction, the methodology must be justified in the case file; the button selected on the screen does not replace that justification.
How to read standard deviation and coefficient of variation without turning an indicator into an automatic rule
Standard deviation shows, in absolute terms, how much the values are spread around the mean. The coefficient of variation relates that dispersion to the mean itself and expresses it relatively, which makes it easier to compare the heterogeneity of sets with different price scales.
In practice, the greater the dispersion, the greater the need to investigate the sample: there may be differences in specification, quantity, region, date, supply unit, or commercial conditions. The Lite tutorial presents these indicators as parameters for assessing homogeneity and data reliability, but it does not establish a universal percentage that automatically validates or invalidates a sample. Avoid creating a “magic cutoff” without support in the process, the object, or guidance applicable to the organization.
The “Compor” column is an analysis tool, not a shortcut for deleting inconvenient prices
Lite lets users enable or disable quotes in the “Compor” column. When a record is disabled, it no longer participates in the statistical calculations or the reference value. A record can also be deleted from the list.
This is useful for cleaning the sample, but an exclusion should stem from a verifiable reason: lack of comparability, inconsistency, an unfeasible price, or an excessively high price, for example. In a formal procurement file, especially where IN SEGES/ME No. 65/2021 applies, the criteria for disregarding values need to be substantiated and described in the administrative record.
What the report shows and what needs to be saved
In the “Resumo” tab, the official tutorial describes two PDF outputs: a summarized report and a detailed report. The detailed version includes, among other information, the mean and median calculated from awarded purchases over the previous 12 months and the data for each item. For editable data, there is an option to export CSV.
There is an important operational limitation: because Lite does not require authentication, searches are not saved for later retrieval. The tutorial warns that using “Home” or “Voltar” ends the search and the data are not saved. If the consultation will support analysis or procurement documentation, export the results before leaving.
The main limit: Lite is not synonymous with the “correct price”
Pesquisa de Preços Lite brings together data from public procurements carried out in Compras.gov.br. That makes it a relevant official source, but it does not turn its database into a complete picture of the entire market. Depending on the object and the rules applicable to the procurement, a formal price survey may require other parameters, sources, or justifications.
There is also no universal choice between mean and median. If prices are comparable and relatively homogeneous, the mean may represent the set well. If there is skewness or extremes that deserve attention, the median tends to be less affected by those values. In either case, the measure should come after reviewing the sample—not before.
Checklist to avoid distorting the quotation
- Confirm CATMAT or CATSER, description, quantity, and supply unit.
- Use filters to bring the sample closer to the real procurement context.
- Open records and check details before accepting the list automatically.
- Compare mean and median; a meaningful difference between them is a signal to investigate the distribution.
- Read standard deviation and coefficient of variation as dispersion indicators, not as a ready-made decision.
- Disable or delete quotes only under an explainable criterion and, when required, document it.
- Choose mean, median, or lowest price according to the sample and the methodology applicable to the process.
- Generate the report and export the data before leaving, because Lite does not save the search without login.
Official sources and references
- Pesquisa de Preços Lite — Compras.gov.br.
- Federal Government Procurement Portal manuals, including the Pesquisa de Preços Lite tutorial, version 1.0 from June 2026.
- Compras.gov.br Notice No. 24/26, on implementing the tool without registration or login.
- TCU — defining and applying the method for calculating the estimated procurement value.
