Dissenting opinion filed by Circuit Judge NEWMAN.
BRYSON, Circuit Judge.U.S. Magnesium LLC appeals from a judgment of the United States Court of International Trade (“the Trade Court”) in this antidumping duty case. The Trade Court sustained the Department of Commerce’s final determination in an administrative review of the antidumping duty order on pure magnesium from the People’s Republic of China for the period of review May 1, 2009, to April 30, 2010. We affirm.
n
b>
The United States imposes duties on foreign goods sold in the U.S. at less than fair value. 19 U.S.C. § 1673. To determine whether goods are being sold for less than fair value, Commerce compares the export price, i.e., the price of the goods sold in the U.S., to the “normal value” of the goods, which is ordinarily the price at which such goods are sold in the exporting country. Id. § 1677b. When merchandise is exported from a nonmarket economy country, the normal value is constructed from “the value of the factors of production utilized in producing the merchandise and to which shall be added an amount for general expenses and profit plus the cost of containers, coverings, and other expenses.” Id. § 1677b(c)(l)(B). Costs are generally “calculated based on the records of the exporter or producer of the merchandise, if such records are kept in accordance with the generally accepted accounting principles of the exporting country (or the producing country, where appropriate) and reasonably reflect the costs associated with the production and sale of the merchandise.” Id. § 1677b(f)(l)(A).
B
This case concerns the importation of magnesium metal from the People’s Re*1025public of China. The imported magnesium is produced using a manufacturing process known as the Pidgeon process. That process begins by crushing dolomite, a mineral containing magnesium, into granules. The dolomite granules are then calcinated by roasting them to remove carbon. The calcinated dolomite is then mixed with fer-rosilicon and fluorite, and the mixture is pressed into individual briquettes. The briquettes are then loaded into stainless steel reaction vessels known as retorts. The retorts are placed under vacuum and heated, resulting in the separation and vaporization of the magnesium. The magnesium vapor condenses into crowns of solid magnesium metal. The crowns of magnesium metal are then removed from the retorts, melted down, purified, and cast into ingots for sale.
The retorts used in the Pidgeon process must be replaced over time in a commercial operation, as the intense heat and the chemical reactions gradually degrade the interior of the retorts. After approximately 6.0 days of use in multiple cycles of the manufacturing process, the retorts become unsuitable for the production of magnesium. The retorts are then recycled, and the recycled steel is used to produce new retorts.1 This appeal focuses on how to classify the costs of the retorts in constructing the normal value of the exported product.
C
In 1995, the Commerce Department entered an antidumping order on magnesium metal from the People’s Republic of China. Pure Magnesium from the People’s Republic of China, 60 Fed. Reg. 25,691 (Dep’t of Commerce May 12, 1995). On May B, 2010, Commerce provided notice of an opportunity for the parties to seek review of the antidumping order, Tianjin Magnesium International (“TMI”), a foreign exporter of magnesium produced in China, and U.S. Magnesium (“USM”), a domestic producer of magnesium, requested that Commerce review TMI’s sales. From June 30, 2010, to May 30, 2011, Commerce solicited comments and information from the parties, including TMI’s business records, surrogate value and country selection, and freight rates.
On June 8, 2011, Commerce released its preliminary results for the 2009-2010 review. Pure Magnesium from the People’s Republic of China: Preliminary Results of the 2009-2010 Antidumping Duty Administrative Review, 76 Fed. Reg. 33,194 (Dep’t of Commerce June 8, 2011). As part of its nonmarket economy review, Commerce constructed a normal value for magnesium by creating surrogate values for the raw materials used in the manufacturing process. It considered ferrosilicon, fluorite powder, dolomite, flux, and coal to be direct materials, and it included them directly in the calculation of normal value. However, it did not include a surrogate value for steel retorts, because it did not regard the retorts as direct materials. Instead, it treated the retorts as indirect materials and accounted for the cost of the retorts as manufacturing overhead.
In a memorandum accompanying the preliminary results, Commerce explained why it classified retorts as indirect inputs and accounted for them as a component of overhead rather than as direct materials.2 *1026Pure Magnesium from the People’s Republic of China, 76 Fed. Reg. 76,945, 76 ITADOC 76,945, Issues & Decision Memorandum, at Comment 4 (Dep’t of Commerce December 9, 2011). First, Commerce explained that “retorts are not physically incorporated into the final product.” While noting that retorts are necessary to the production process, Commerce stated that “they are more similar to a kiln or furnace,” the costs of which Commerce generally treats as manufacturing overhead. Commerce also found that retorts are reusable and “are not replaced so regularly as to represent a direct factor rather than overhead.” Finally, Commerce found that it was “unclear how retorts are typically treated in the industry.”
Following the preliminary results, USM continued to argue that retorts should be classified as direct materials rather than as overhead. In the final results, however, Commerce stood by its classification, explaining that the retorts are best classified as overhead “because they are not physically incorporated into the final product and are replaced too infrequently to be a direct material.” Commerce concluded that “retorts are not an input added into the production process; rather, they are manufacturing equipment, like an oven or crucible, all of which are necessary components of the production line to produce pure magnesium.” Consequently, Commerce declined to treat retorts as a direct material.
D
After the closing of the administrative record in the review, USM sought to submit new evidence contradicting one of the answers TMI provided to Commerce. USM contended that the new evidence was indicative of fraud on TMFs part. Commerce rejected USM’s submission as untimely. The Trade Court, however, remanded the case to Commerce for consideration of the new evidence.
On remand, Commerce found that USM’s newly submitted evidence did not constitute prima facie evidence of fraud. Commerce also found that the new evidence did not call into question its finding that retorts are properly treated as factory overhead.
The Trade Court affirmed Commerce’s remand results. U.S. Magnesium LLC v. United States, 72 F.Supp.3d 1341 (Ct. Int’l Trade 2015). After reviewing the financial records of TMI’s supplier, the court sustained Commerce’s conclusion that the supplier did not treat retorts as a direct material. The court also upheld Commerce’s finding that the evidence was inconclusive as to whether the industry as a whole treated retorts as a direct material input. And the court held that Commerce was “right in its claim that treating the retorts as an indirect material is consistent with its past practice of characterizing materials as overhead when ‘they are not physically incorporated into the final product and are replaced too infrequently to be a direct material’ ”
USM appealed to this court from the Trade Court’s judgment.
II
Commerce’s determinations in an anti-dumping duty case must be upheld unless they are “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(1)(B)(i). Substantial evidence means “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938).
*1027In conducting substantial evidence review of Commerce’s determinations, we apply the same standard of review that the Trade Court used in reviewing the administrative record. Downhole Pipe & Equip., L.P. v. United States, 776 F.3d 1369, 1373 (Fed. Cir. 2015). As we have. explained, however, we “will not ignore the informed opinion of the Court of International Trade.” Diamond Saioblades Mjrs. Coalition v. United States, 612 F.3d 1348, 1356 (Fed. Cir, 2010) (quoting Suramerica de Aleaciones Laminadas, C.A. v. United States, 44 F.3d 978, 983 (Fed. Cir. 1994)).
A
USM argues that substantial evidence does not support Commerce’s decision to classify the retorts as overhead rather than as direct materials. USM contends, first, that TMI’s supplier characterized retorts as a direct cost, as shown by its books and records, and that Commerce should have based its determination on that data from TMI’s supplier. In support of its argument, USM points to three documents of TMI’s supplier, all from December 2009: the supplier cost sheet, the material out specification sheet, and the cost of production subledger.
■ The supplier cost sheet is a one-page document detailing the expenditures of TMI’s supplier. It includes both total and unit costs of a variety of different materials essential to the production of magnesium. In the category of “raw material,” the chart shows cost breakdowns for FeSi (fer-rosilicon), dolomite, flux, fluorite powder, sulfur powder, and sulfuric acid. Retorts are not included within the category of “raw material,” but are listed separately.
Like the supplier cost sheet, the material out specification sheet details the supplier’s expenditures. The material out specification sheet separately lists the costs for retorts along with the costs for the common ingredients, materials, and equipment used in processing magnesium.
The cost of production subledger, which contains financial information, includes entries for a number of items that it describes as “materials consumption,” a category that includes raw materials, retorts, certain equipment used in the manufacturing process, and certain other itemized expenses.
USM argues that these documents show that the internal accounting records of TMI’s magnesium supplier characterized retorts as a direct input. It argues that the retorts were not characterized as “accessory expenses,” i.e., indirect materials, and were not' characterized as “manufacturing expenses,” i.e., equipment. Therefore, USM argues, TMI’s supplier necessarily treated retorts as direct material inputs, not as overhead.
Commerce rejected that interpretation of the documents. Instead, Commerce noted that TMI’s supplier grouped other expenses together with retorts, even though those other items are not considered material inputs. Accordingly, Commerce concluded that “the fact that TMI’s supplier lists retorts [in the cost of production sub-ledger] does not indicate that they are a direct material, but rather part of the cost of production.”
The Trade Court agreed with Commerce’s interpretation of the supplier’s documents. The court noted that “at first blush, the description ‘materials consumption’ [in the cost of production subledger] might indicate that the listed items were direct inputs, [but] an examination of the nature of the entries shows that they are not.” Thus, although the subledger lists retorts among the items classified as “materials consumption,” it also includes other items under that category that are clearly not direct materials, such as “crucible,” *1028“packing,” and “accessory expense.”3 Likewise, the cost sheet does not include retorts under the category of “raw material.” In light of that evidence, the Trade Court concluded that Commerce permissibly found that TMI’s supplier did not treat retorts as a direct material.
Commerce’s findings as to the supplier’s treatment of retorts are supported by substantial evidence. The documents are not definitive, and Commerce’s reading of them is plausible.
The documents that USM highlights contain line items that range from items that are plainly direct materials to others that are plainly not. Given that retorts are not listed as raw materials, and that retorts are grouped together with other expenses that are plainly not direct materials, it was reasonable for Commerce to conclude that the records do not show that TMI’s supplier treated retorts as direct inputs. We will not second-guess the agency’s choice between plausible interpretations of record evidence, particularly in light of the Trade Court’s conclusion that Commerce’s analysis of the record materials was reasonable. In re Jolley, 308 F.3d 1317, 1326 (Fed. Cir. 2002).
B
USM next argues that in classifying the retorts as indirect materials, Commerce unjustifiably departed from its prior practice in similar cases. In USM’s view, Commerce abandoned its prior reliance on a four-part test for distinguishing between indirect materials, which are accounted for as overhead, and direct materials: “1) whether the input is physically incorporated into the final product; 2) the input’s contribution to the production process and finished product; 3) the relative cost of the input; and, 4) the way the cost of the input is typically treated in the industry.” Certain Steel Nails from the People’s Republic of China, 78 Fed. Reg. 16651, 78 ITADOC 16651, Issues & Decision Memorandum, at Comment 4 (Dep’t of Commerce Mar. 5, 2013). The government responds that Commerce has never followed such a strict four-factor test, but instead has characterized materials as direct or indirect depending on a variety of relevant factors.
We agree with the government. In distinguishing between direct materials and overhead, Commerce has not confined the inquiry to particular defined factors, but instead has employed a “totality of the circumstances” test. For example, in the Certain Steel Nails ease, after summarizing the factors considered in previous determinations, Commerce cautioned that:
As demonstrated by the variety of considerations, there is no conclusive test for reaching the appropriate classification of inputs that are not easily distinguished on-their face as direct materials or [overhead]. Further, contrary to Petitioner’s assertion that meeting any one of these factors demonstrates that an input is a direct material, the Department instead finds that it is the totality of the evidence that must guide its decision in each case.
Certain Steel Nails, Issues & Decision Memorandum, at Comment 4.
*1029In some investigations, such as Citric Acid and Certain Citrate Salts, Commerce has made findings, with regard to all of the four factors that USM cites. 76 Fed. Keg. 77, 772, 78 ITADOC 77,772, Issues & Decision Memorandum, at Comment 18 (Dep’t of Commerce Dee. 7, 2011). In others, Commerce has focused on individual factors that appeared significant in the particular investigation. For example, in Certain Steel Nails, Commerce based its classification of dies as indirect materials on the fact that “dies are not consumed on a directly proportional basis,” but are reused until they have worn out and can no longer be used. Issues & Decision Memorandum, at Comment 4. In another case, Diamond Sawblades and Parts Thereof from the People’s Republic of China, Commerce relied on the lifespan of molds in determining whether they were properly classified as indirect materials, as well as the fact that graphite molds were partially incorporated into the final product. 71 Fed. Keg. 29,308, 71 ITADOC 29,303, Issues & Decision Memorandum, at Comment 2 (Dep’t of Commerce May 15, 2006),
In this case, Commerce relied primarily on two factors: the fact that the retorts were not physically incorporated into the final product, and the fact that the retorts were not replaced frequently. Commerce’s focus on those factors was neither inconsistent with its past practices in analogous cases nor unreasonable as a way of distinguishing between direct and indirect materials. As long as its analysis was reasonable, as it was here, Commerce was not required to examine and rely on every factor that it has used in the past. See Bridgestone Ams., Inc. v. United States, 710 F.Supp.2d 1359, 1364 (Ct. Int’l Trade 2010) (Commerce is not bound by the four-factor test and has discretion “to rely on various criteria to value factors of production.”).
C
USM next argues that Commerce erred in finding that retorts are replaced too infrequently to be treated as a direct input. Citing Commerce’s decision in Citric Acid, USM suggests that in the past Commerce has required an item to have at least one year of useful life in order to be classified as overhead. In that investigation, however, Commerce noted that the resins at issue were used for more than one year; it concluded that they were therefore properly treated as indirect materials. Commerce did not create a rule of thumb that items replaced more often than once per year must be treated as direct materials. In fact, Commerce has found factory items having a useful life much shorter than the retorts in this case to be properly categorized as indirect materials whose costs should be treated as overhead. In Certain New Pneumatic Tires Ojf-the-Road Tires from the People’s Republic of China, Commerce characterized curing bladders used in tire manufacturing as overhead when the record evidence showed that they were replaced as frequently as once every two to eight days. 77 Fed. Reg. 14,495, 77 ITADOC 14,495, Issues & Decision Memorandum, at Comment 3 (Dep’t of Commerce Mar. ,5, 2012).
USM also criticizes Commerce’s comparison of the retorts with the steel molds at issue in Diamond Sawblades, because the public record does not reflect the lifespans of the graphite and steel molds that were at issue in that case. However, given Commerce’s obligation to maintain the confidentiality of investigation respondents’ business proprietary information, it is inevitable that the public records of investigations will not always disclose such facts. It was not unreasonable for Commerce to conclude that the 60-day lifespan of the retorts was long enough to justify *1030its conclusion that the retorts should not be treated as direct materials in the manufacturing process.
As part of its argument about the replacement rate for the retorts, USM quarrels with Commerce’s conclusion that the record evidence did not show that the retorts were traceable to specific magnesium products. USM argues that Commerce improperly conflated “traceability” with “physical incorporation,” and that because the retorts were consumed in the course of the magnesium production process, they were traceable to specific magnesium products. Defined in that manner, however, the term “traceable” would apply even to items such as furnace components that have to be replaced as infrequently as once a year.
By “traceable,” Commerce appears to refer to items that are continuously consumed during the production process and must continually be replaced. See, e.g., Sil-icomanganese From the People’s Republic of China, 65 Fed. Reg. 31,514, 65 ITADOC 31,514, Issues & Decision Memorandum, at Part IV, Comment 1 (Dep’t of Commerce May 18, 2000) (electrode paste is a direct material because it is “burned-off during production and must continually be replaced”); Wooden Bedroom Furniture From the People’s Republic of China, 69 Fed. Reg. 67,313, 69 ITADOC 67,313, Issues & Decision Memorandum, at Comment 6 (Dep’t of Commerce Nov. 17, 2004) (abrasives used in production of furniture are direct materials because they are “consumed in' large quantities and their consumption is tied directly to the amount of subject merchandise each respondent produced”); Silicon Metal From the Russian Fed’n, 68 Fed. Reg. 6,885, 68 ITADOC 6,885, Issues & Decision Memorandum, at Comment 25 (Dep’t of Commerce Feb. 11, 2003) (electrodes were direct materials because they “were burned away each day and are continually replaced”). It was reasonable for Commerce to conclude that retorts, which last for many production cycles, are not traceable to specific magnesium products and are not required to be characterized as direct materials on that basis.
D
USM argues that Commerce failed to take into account practices among other companies in the magnesium production industry with respect to the accounting treatment of retorts. USM contends that the industry practice is to treat retorts as a direct input.
In support of its argument, USM relies on the records of three other producers. The first, a Malaysian producer, classified retorts as a direct input, which is undisputed by all parties in this case. As to the second, an Indian company that had ceased production 10 years before the review period, Commerce found that the company had classified retorts as a direct expense in one year, 1994-95, but not in any subsequent years. As to the third, a Chinese company, Commerce found that it was unclear whether that company classified retorts as a direct input.4
Based on the record before it, Commerce found that USM’s evidence with respect to industry practices in the ac*1031counting treatment of retorts was inconclusive. The Trade Court determined that “it is difficult to quarrel with the Department’s conclusion that USM has presented little evidence that retorts are treated as a direct input by the magnesium industry.”
We agree with the Trade Court that substantial evidence supports Commerce’s conclusion that USM’s evidence on this point was inconclusive. USM quarrels with the inferences Commerce drew from the record evidence, but “the possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence.” Consolo v. Fed. Maritime Comm’n, 383 U.S. 607, 620, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966). Interpreting accounting documents and drawing conclusions from them are tasks within Commerce’s expertise. Fujitsu General Ltd. v. United States, 88 F.3d 1034, 1039 (Fed. Cir. 1996). Commerce considered all of the submitted documents relating to the three foreign producers. Given that there was an evidentiary basis for Commerce to conclude that only one of the three treated retorts as a direct input, it was reasonable for Commerce to find that no industry-wide practice had been shown.
E
Next, USM argues that. Commerce ignored the cost of the retorts relative to the final product. We disagree. Commerce did not ignore testimony about the cost of the retorts; instead, it explicitly discussed that issue in the memorandum accompanying the final results, but found other factors more persuasive. Commerce wrote:
Although the Department may consider the relative cost to determine whether certain items should be attributed to overhead, this consideration is not determinative or considered alone. Further, in determinations where cost has played a large role, the cost has not been related to factory equipment. For example, in Urea/Russia AD Final (02/21/2003), the items were catalysts. In Silicomanga-nese /PRC AD Final (05/18/2000), the electrode paste was a “consumable” used up during production. In this review, retorts are not an input added into the production process; rather, they are manufacturing equipment, like an oven or crucible, all of which are necessary components of the production line to produce pure magnesium.
As Commerce pointed out, the relative cost of particular items has not been regarded as an important factor where the cost in question is related to factory equipment or items that are not used up during production. The relative cost of the retorts therefore provides no reason to reject Commerce’s findings as unsupported by substantial evidence.
F
Finally, USM complained at oral argument that Commerce ignored a declaration by John Haack, a representative of an American company involved in magnesium production. USM infers that Commerce ignored Dr. Haack’s declaration because it did not refer to the declaration in its decision memorandums.
First, to the extent USM argues that Commerce’s decision lacked substantial evidence support because of the failure to cite the declaration, it is wrong. We presume that a factrfinder reviews all of the evidence presented unless it states otherwise, even if its opinion does not “recite every piece of evidence.” Plant Genetic Sys., N.V. v. DeKalb Genetics Corp., 315 F.3d 1335, 1343 (Fed. Cir. 2003).
Second, the Haack declaration added very little by way of evidentiary support for USM’s position. The two-page declara*1032tion consists of undisputed facts about the Pidgeon process and a conclusory statement that “[t]o accurately account for the costs of producing magnesium using the Pidgeon process retorts should be treated as a direct material input, because they are central to the production process, have a short useful life, are consumed by the process, and have a very high unit cost.” In light of the skimpy nature of the declaration, it is unsurprising that Commerce found that the declaration contributed little to its investigation and did not need to be separately addressed.
Ill
We agree with the Trade Court that Commerce’s decision was supported by substantial evidence and not otherwise contrary to law. We therefore uphold the Trade Court’s judgment sustaining Commerce’s Final Results of Redetermination in this case.
AFFIRMED
. Evidence in the administrative record showed that the Pidgeon process takes approximately 12 hours to complete.'Based on the time required for each production cycle and the lifespan of the retort, a single retort can be used for many cycles before being replaced.
. In addition to indirect materials, Commerce treats overhead as including expenses such as building or equipment rental, depreciation, *1026supervisory labor, plant property taxes, factory administration, and other like costs.
. USM argues that Commerce misread the cost-of-production subledger because the "ldln accrued expenses” and "wage accrued" expenses are not listed in the "materials consumption" section of the document. But Commerce’s point was that the "materials consumption” expenses include items that are not regarded as direct materials, so the inclusion of retort costs in the list of "materials consumption” expenses does not show that the supplier treated retorts as direct materials.
. USM argues that, in addition to the Malaysian, Indian, and Chinese companies, the evidence, showed that two other producers treated retorts as direct materials: TMI's supplier and Magpro, LLC, an American producer. We have already addressed TMI’s supplier in part II-A, above. As for Magpro, USM’s evidence ta. the form of a declaration from the managing member of Magpro showed that Magpro used a different process, in which retorts typically last three to five years. The declarant did not state that Magpro treated retorts as direct inputs.