We find that the commission did not abuse its discretion in excluding a portion of the development costs of CG&E’s customer service system from its rate base. We find, however, that the commission did act unlawfully in imputing revenue levels for ratemaking purposes that differ significantly from the actual revenue collected from customers during the test period pursuant to contracts previously approved by the commission.
*58Rate Base
CG&E argues that the commission abused its discretion by ignoring evidence and the cross-examination of witnesses, thereby producing an order that was contrary to the manifest weight of the evidence. It has been long held in commission matters that “this court will not reverse or modify a [commission] decision as to questions of fact where the record contains sufficient probative evidence to show that the [commission’s] determination is not manifestly against the weight of the evidence and is not so clearly unsupported by the record as to show misapprehension, mistake, or willful disregard of duty.” Ohio Edison Co. v. Pub. Util. Comm. (1992), 63 Ohio St.3d 555, 556, 589 N.E.2d 1292, 1294.
Pursuant to R.C. 4909.15(A), the commission in a rate case is to determine the rate base as being “[t]he valuation as of the date certain of the property of the public utility used and useful in rendering the public utility service for which rates are to be fixed and determined.” (Emphasis added.)
There is no dispute in this case that the CSS itself is used and useful. The dispute resides in the development costs of the project, which were capitalized and treated as property in CG&E’s rate base. Still, those expenses were in the first instance operating expenses, and, as such, they must meet the prudency test in R.C. 4909.154 before they can gain the status of “used and useful.” R.C. 4909.154 provides:
“In fixing the just, reasonable, and compensatory rates, joint rates, tolls, classifications, charges, or rentals to be observed and charged for service by any public utility, the public utilities commission shall consider the management policies, practices, and organization of the public utility. * * *
“* * * [T]he public utilities commission shall not allow such operating and maintenance expenses of a public utility as are incurred by the utility through management policies or administrative practices that the commission considers imprudent.”
Thus, it follows that operating expenses that the commission determines under R.C. 4909.154 were imprudently incurred will be disallowed, when capitalized, as not being used and useful property under R.C. 4909.15(A)(1) and therefore deducted from capitalized expense included in rate base.
This court set forth in Cincinnati v. Pub. Util. Comm. (1993), 67 Ohio St.3d 523, 530, 620 N.E.2d 826, 830, what constitutes a prudent decision:
“ ‘[0]ne which reflects what a reasonable person would have done in light of conditions and circumstances which were known or reasonably should have been known at the time the decision was made.’ In the Matter of the Investigation into the Perry Nuclear Power Station (Jan. 12, 1988), PUCO No. 85-521-EL-*59COI, at 10-11. The standard contemplatés a retrospective, factual inquiry, without the use of hindsight judgment, into the decisionmaking process of the utility’s management.”
The commission considered CG&E’s mismanagement of the CSS project to constitute imprudence, resulting in the disallowance of a substantial portion of CG&E’s proposed CSS rate-base property. We find that the commission did not abuse its discretion in making that determination.
Institutional hubris is at the heart CG&E’s problems with the development of the CSS. The stubborn belief that the project could be completed in-house, despite continued rising costs and missed deadlines, was the source of much of the project’s expense. By the time the company hired an independent consultant to complete the CSS on deadline for a fixed fee, the damage had been done. It is that period before the hiring of the consultant, when the company sought to develop the project itself, that the commission points to in finding that CG&E acted imprudently. We find that the record supports that conclusion.
We need not slog through the daily history of the development of this project to determine that it was mismanaged. The big picture tells the tale. When CG&E commenced development of the system in 1987, it projected that the project would.be completed by July 1990. In 1990, CG&E revised its projected completion date to June 1991, and projected the capitalized cost to be $24 million. Prior to 1990, CG&E had already spent $17 million in ordinary operating expenses on the CSS project, but those expenditures are not at issue in this case. As of April 1991, the CSS project team was reporting in its status reports that the project was running only two months behind schedule. By fall 1991, the truth was known: since 1990, CG&E had spent $47 million on the CSS, and it was only half complete. All of these figures reveal that CG&E was an entity in over its head.
The evidence showed that at least ten other utilities undertaking computer projects of similar complexity sought, without’exception, independent professional supervision of project development. From virtually the time project development began in 1987, CG&E knew, or should have known, that it was involved in a massive undertaking that far exceeded the expertise of its in-house project team. CG&E had already sunk $47 million in capitalized expenditures into the project when it finally relinquished project development supervision to CSC in November 1991. Based on its 1990 projection of $24 million to complete the project, CG&E went one hundred sixty percent over its budget.
Evidence obtained from independent consultant Andersen Consulting established that the project was approximately fifty percent completed when CSC took over supervision in 1991. The CSC contract called for project completion for a *60relatively frugal fixed fee of $14,875 million and established July 1993 as the deadline. CSC completed the job on time.
The CSC contract amount of $14,875 million and the fifty-percent completion figure when CSC took over the job are the two most important elements in the commission’s disallowance of a portion of the CSS costs from the rate base. First, those figures help establish that the project was mismanaged by CG&E. CG&E had nearly doubled its budget and had completed only half of the project when CSC stepped in. For less than $15 million, CSC was then able to complete in two years as much of the project as CG&E had completed at a cost of $47 million in four years.
Second, the commission used the CSC contract to compute its rate-base deduction. The commission’s staff used the amount paid to CSC to determine the reasonable cost of the entire project. Using simple, but not simplistic, logic, the commission’s staff reasoned that the one-half of the CSS development project completed before CSC took over would reasonably cost the same as the amount paid to CSC for the one-half of the project that CSC completed. The commission thus doubled the fixed contract rate to arrive at its $29.75 million estimate of a reasonable cost for the project. We find that while the staffs methodology is simple, it is justifiable as a means for estimating a reasonable cost for the CSS development project.
The figures relied upon by the staff and the commission were not contrary to the manifest weight of the evidence. The staff presented adequate evidence supporting the fifty-percent allocation. The staff report stated that “CG&E hired Andersen Consulting and CSC to perform a full assessment of the entire project * * *. Both consultants agreed the CSS project was 50% completed at the point of their evaluations.” The fifty-percent level of completion is clearly corroborated by the analytical report of Andersen Consulting entitled “Customer Service System Review.”
In his prepared written testimony, staff witness Francis C. Rack adopted the fifty-percent level of completion and referred to the Andersen Consulting conclusion that “the project was 50% complete at the time of their evaluation.” Rack testified on cross-examination that a CG&E manager he interviewed agreed for the most part with the Andersen Consulting review, but that the manager felt the work completion was closer to sixty-percent complete.
Rack further stated on cross-examination that he felt comfortable relying primarily on Andersen Consultant’s estimation of fifty-percent completion of the CSS development project in part because of a CG&E manager’s assessment that Andersen Consulting did an in-depth evaluation of the project that constituted “the best job.”
*61CG&E cross-examined staff witnesses and elicited from an expert witness his opinion that the CSS was seventy- to seventy-five-percent complete when CSC took over the project. CG&E provided other testimony and expert witnesses that disagreed with or contradicted testimony and opinions of other staff witnesses.
CG&E implies that the commission ignored evidence and cross-examination of witnesses. Rather, it is probable that the commission merely chose to rely on evidence and testimony other than that proffered by CG&E.
The commission’s finding that the CSS development project was fifty-percent complete was based on sufficient record evidence. While CG&E presented testimony from its witnesses of a greater level of completion, the commission’s finding of a fifty-percent level of completion is certainly not manifestly against the weight of the evidence.
The commission and its staff also broke down the big picture, and identified six separate ways by which CG&E was deemed to have mismanaged the CSS development project. The staff concluded and the commission agreed that CG&E had unreasonably missed project deadlines and exceeded project costs, had failed to sign CSS contractors to contracts with identified deliverables with certain deadlines, had chosen to develop a customized DB2-based CSS system without adequate identification of software capabilities and resource availability, had invested in a system in which the costs greatly exceeded benefits, and had failed to maintain an adequate audit trail of CSS system development and managerial decision-making with regard to costs and deadlines.
Each of those elements is supported by record evidence. CG&E presented contradictory evidence as to several of the elements, and as to several of them CG&E challenged whether the evidence presented by the commission’s staff evinced mismanagement. However, the evidence of elements of mismanagement was sufficient to support a conclusion that CG&E had mismanaged the CSS development project.
Therefore, the commission’s deduction from the rate base for CG&E’s imprudent expenditures under R.C. 4909.154 should stand. The $29.75 million figure is nearly twenty-five percent higher than CG&E’s original capitalized cost estimate of $24 million to complete the project and also significantly exceeds the tangible benefit of $20 million that the staff found associated with project completion. We affirm the commission on this issue.
Delta Revenues
A utility’s test-period revenues are a key component in determining whether the commission will grant a rate increase. The test period provides a snapshot of the utility’s revenues. Those revenues form the basis upon which the utility will *62be judged. If the revenues received by the utility during the test year are less than the gross annual revenues to which the utility is entitled, the commission is required to fix new rates that will raise the necessary revenue. R.C. 4905.15(D); Columbus S. Power, 67 Ohio St.3d at 537-538, 620 N.E.2d at 838-839.
R.C. 4909.15(C) addresses test-period revenues. R.C. 4909.15(C) provides:
“The test period, unless otherwise ordered by the public utilities commission, shall be the twelve-month period beginning six months prior to the date the application is filed and ending six months subsequent to that date. In no event shall the test period end more than nine months subsequent to the date the application is filed. The revenues and expenses of the utility shall be determined during the test period. The date certain shall be not later than the date of filing.”
R.C. 4909.15(C) by its terms does not require that the commission apply actual test-period revenues in 'the ratemaking formula. Neither party disputes that estimates and forecasts are permissible. They are necessary for a representative sampling.
But the revenues attributed to CG&E in this case were something other than forecasts. Their lack of connection to reality is what CG&E disputes. A forecast would take into account the contract price for the service provided and multiply that by the volume anticipated to be used by the buyer to determine revenue. The commission substituted the price it thought CG&E should charge for the service. Instead of forecasting based on known factors, the commission ignored reality. Instead of a snapshot of CG&E’s revenues, the commission substituted its own surreal vision. The law requires Ansel Adams; the commission gives us Salvador Dali.
This court has determined that the commission has statutory authority to make adjustments to test-period revenues. R.C. 4909.15(D)(2) allows the commission to “smooth out anomalies” in test-period data. Columbus S. Power Co. v. Pub. Util. Comm. (1993), 67 Ohio St.3d 535, 539, 620 N.E.2d 835, 839. The reasoning behind that policy was explained in Consumers’ Counsel v. Pub. Util. Comm. (1981), 67 Ohio St.2d 153, 166, 21 O.O.3d 96, 104, 423 N.E.2d 820, 828:
“It is our view that R.C. 4909.15(D)(2)(b) is designed to allow the commission to make minor adjustments to rates ascertained by the statutory formula when the criteria upon which rates are based are skewed for one reason or another. Thus, under R.C. 4909.15(D)(2)(b), the commission may smooth out anomalies in the ratemaking equation that tend to make the test year data unrepresentative for ratemaking purposes.”
The rates paid for gas transport by AK Steel and Ford during the test period were not unrepresentative of the price they had paid previously or would continue *63to pay. The test-period rates were those rates that AK Steel and Ford would continue to pay in the long term, contract rates that were approved by the commission pursuant to R.C. 4905.31. Those contracts required CG&E to transport gas to AK Steel and Ford for $0.24/Mcf and $0.23/Mcf respectively, rather than the tariff rate of $0.537/Mcf.
This court has held that the commission has the power to “smooth out anomalies” when something occurs out of the ordinary in the test period. But in Dayton Power & Light Co. v. Pub. Util. Comm. (1983), 4 Ohio St.3d 91, 95, 4 OBR 341, 344, 447 N.E.2d 733, 736, this court stated that such “exceptions remain exceptions, and ad hoc tinkering with the statutory formula is not to become the rule.” For instance, if the volume transported spiked one way or other because of an unlikely demand shift, the commission could adjust the figures to more appropriately reflect what they would be in a more normal time period. But in this case, only the ordinary happened. The rate stayed as it had been and the way it would continue to be, and the way the commission had earlier approved. The contract revenues are representative of normal operations and do not represent “anomalies in the ratemaking equation that tend to make the test year data unrepresentative for ratemaking purposes.” Consumers’ Counsel, 67 Ohio St.2d at 166, 21 O.O.3d at 104, 423 N.E.2d at 828.
Therefore, it was unlawful for the commission to substitute the maximum tariff rate of $0.537/Mcf for the contractual rates of $0.24/Mcf and $0.23/Mcf in calculating CG&E’s test-period revenues. We reverse the commission on this issue.
Order affirmed in part and reversed in part.
Resnick and F.E. Sweeney, JJ., concur. Moyer, C.J., Cook and Lundberg Stratton, JJ., concur in judgment only. Douglas, J., dissents.