Defensible price research is not just about putting together three commercial proposals and calculating the average. What needs to be demonstrated in the process is that the estimated value is related to the market for the object that will be contracted, considering quantity, commercial conditions, place of execution and other characteristics capable of changing the price.
Law nº 14,133/2021 allows different parameters to be used to arrive at this estimate, combined or not. Direct quotation with three suppliers is just one of these parameters. In the federal Public Administration, Normative Instruction SEGES/ME nº 65/2021 still determines that official government systems and similar contracts carried out by the Administration are prioritized.
“Three budgets” is not the general rule of research
Confusion often arises because art. 23, § 1, IV, of Law No. 14,133/2021 mentions direct research with at least three suppliers. But this device specifically addresses one of the possible sources. The same article also admits prices for public systems, similar contracts, specialized research and the national base of electronic invoices.
At the federal level, there is another important distinction. IN SEGES/ME nº 65/2021 establishes, as a rule for the calculation, a set of three or more prices, which can come from one or more parameters. This does not mean three supplier proposals. Exceptionally, the standard even allows the determination of the estimated price with less than three prices, as long as there is justification in the records and approval by the competent authority.
Therefore, three commercial proposals are not a safe conduct. A survey with five bad references may be less reliable than a smaller sample, but one that adheres to the objective and is properly justified.
The TCU works in the same direction by recommending a “price basket”, with a preference for public prices and the use of research exclusively with suppliers as a residual solution when there are no public references or other suitable sources.
Before collecting prices, make references comparable
A source is only useful when the price searched represents something close enough to the intended contract.
Imagine purchasing equipment for R$8,000 in a previous public contract. This value cannot be automatically transplanted to the new process. The quantity may have been much larger, shipping may have been included, the warranty may have been different, or delivery may have taken place in another region.
IN nº 65/2021 determines that, whenever possible, the research considers conditions such as delivery time and place, installation, quantity, form and payment term, shipping, guarantee, brand or model, economy of scale and peculiarities of the place of execution. The TCU Bidding and Contracts Manual also warns that differences in these conditions may distort the estimate.
This changes the logic of the job. Before asking “what is the price?”, it is necessary to ask “price of what, under what conditions?”.
How to choose pricing sources
The best source is not necessarily the easiest to obtain. It is one that has sufficient adherence to the object and whose origin can be demonstrated in the process.
| Source | When it tends to be most useful | Main care |
|---|---|---|
| Official government systems | Goods and services with comparable public procurement history | Filter specification, quantity, period and contracting conditions |
| Similar Administration Hiring | When there are effectively comparable objects already contracted | Confirm similarity and update values when necessary |
| Specialized media, tables and websites | Markets with current and verifiable public references | Record date and time, check commercial conditions and avoid merely promotional or incomparable prices |
| Direct quotation with suppliers | Specific objects or situations where other sources are insufficient | Formalize the consultation, justify the choice of those consulted and send everyone equivalent conditions |
| National database of electronic invoices | When used in accordance with applicable regulations and there is comparable data | Check period, item description and transaction adherence to planned contracting |
In federal regulation, the first two parameters must be prioritized. Websites, suppliers and other references are not prohibited; its use needs to make sense in the specific case. The Law allows for combining parameters, and the TCU recommends, whenever possible, a diversity of sources so that the Administration can see a range of the market, instead of depending on a single cut.
If the team opts for direct research with suppliers under the terms of IN nº 65/2021, the request must be formal. The choice of suppliers and the companies consulted that did not respond must also be documented. The relevant characteristics of the contract must be informed to those consulted to improve the comparability of proposals.
Discrepant prices are a signal to investigate
Suppose a search finds values of R$90, R$95, R$102 and R$240 for the same item. Simply withdrawing R$240 because “it’s too high” is not enough. Likewise, automatically keeping it on average can artificially raise the budget.
The first task is to find out why the difference exists.
Price may correspond to different specification, different quantity, exceptional shipping, additional service, shorter term, extended warranty or incompatible commercial condition. It can also be an effectively excessive price. On the other side of the sample, a very low value may represent a one-off promotion, registration error, different object or condition that will not be available to the Administration.
IN No. 65/2021 expressly determines that unfeasible, inconsistent and excessively high prices are disregarded when forming the estimate and requires well-founded and described criteria in the process. The standard also requires critical analysis to be carried out, especially when there is a large variation between the values collected.
The TCU reinforces that the exclusion of references must result from a previously justifiable method, not from an intuitive choice made to produce a certain result.
A defensible treatment might follow this sequence:
- check that all references correspond to the same object and comparable commercial conditions;
- identify the probable cause of relevant differences;
- define and record the criteria used to maintain or delete each reference;
- form the sanitized sample;
- only then apply the estimated price calculation method.
The advantage of this order is simple: the statistical method no longer serves to hide problems in the sample.
Average, median or lowest value?
IN nº 65/2021 admits average, median or lower value, in addition to other methods duly justified and approved by the competent authority. The standard does not determine that a single method be used for all research.
The average works best when the sample, after being sanitized, is relatively homogeneous. As extreme values directly influence the result, using it without first evaluating the dispersion can produce an artificially high or low reference.
The median reduces the influence of extremes and may be more appropriate when prices present greater heterogeneity. This does not eliminate the need to investigate discrepancies: the median does not transform an inadequate reference into a valid reference.
The lowest value should not be chosen just to produce a lower budget. The TCU Manual records jurisprudence in favor of the use of the minimum quote in certain restricted market situations, while averages and medians are associated with competitive market conditions. The choice continues to require connection with the real characteristics of the researched market.
In any case, the question that needs to be answered in the process is: why does this method better represent the price range of the contract analyzed?
The estimated price needs to be explainable by someone else
A good calculation memory should allow someone who did not participate in the research to reconstruct the reasoning.
At the federal level, IN nº 65/2021 requires that the research document identifies the object, those responsible, the sources consulted, the price series, the method used, the justifications for possible exclusions, the calculation memory and the supporting documents. When there is direct consultation, the justification for choosing suppliers must also be included.
It is not enough to write “the median was adopted because it is more appropriate”. Motivation must connect data to decision.
A more useful justification would have this logic: the research gathered references from similar public contracts and from a certain complementary source; specifications, quantities and delivery conditions were compared; two references were removed for documented reasons; the remaining sample presented a certain characteristic; Therefore, the method that best represented the analyzed set was chosen.
The goal is not to create a long justification. It is to allow traceability.
The most recent source may also be the wrong source
Curency matters, but it doesn't solve the search alone.
In IN nº 65/2021, similar public contracts must, as a rule, be in progress or have been completed within a period of one year prior to the research. For specialized websites and research, there are specific requirements for currentness and access registration; Direct quotations with suppliers also have a time limit.
Even within these periods, the researcher needs to evaluate events capable of changing the price: relevant market changes, input variations, product discontinuation, tax changes or logistical differences, for example.
The date is an expiration filter. Grip remains the quality filter.
The error that the “three-budget rule” tends to hide
When the team transforms three proposals into the objective of the process, the research runs the risk of starting with the document and not with the market.
The most robust procedure is the opposite: define exactly what will be contracted, identify which sources represent this market, collect comparable references, explain discrepancies, clean up the sample and only then calculate the estimated value.
This approach also avoids two opposing problems: an inflated budget, which can tolerate overpricing, and an artificially low estimate, which can reduce competitiveness or lead to unfeasible proposals.
Law No. 14,133/2021 requires that the previously estimated value is compatible with market values, considering public databases, quantities, economies of scale and peculiarities of the execution location. It is this result that the procedural instruction needs to demonstrate.
Before ending the preparatory phase, therefore, it is worth doing a simple test: could someone else understand where the prices came from, why some were accepted or rejected and why the chosen method represents the market? If the answer depends on an oral explanation from the person who carried out the research, the documentation can still be strengthened.
