After our decision in Augat, Inc. v. Aegis, Inc., 409 Mass. 165 (1991) (Augat I), in which we rejected all but one theory under which the trial judge had ruled that the defendants were liable to the plaintiffs (id. at 168), the case was tried on the damages question. The defendants have now appealed from a judgment awarding substantial damages to the plaintiffs. The plaintiffs have appealed challenging one ruling of the judge. We granted the defendants’ application for direct appellate review. We conclude that the judge erred in his determination of damages and, therefore, we remand the case for a new determination of damages.

In Augat I, we identified a theory of liability which we characterized as “important but substantially isolated.” Id. at 173. The defendants had wrongfully joined with Jay Greenspan, while he was the general manager of the plaintiff Isotronics, in a violation of his duty of loyalty to Isotronics. Id. at 172. Greenspan, a top managerial employee, secretly solicited the departure of certain key managerial personnel from Isotronics to the defendant Aegis while he had a “duty to maintain at least adequate managerial personnel” at Isotronics. Id. at 173-174.

We said that “[t]he plaintiffs’ damages relate to negative effects on operating results that would not have occurred but for the departure of the key managerial employees” solicited by Greenspan. Id. at 175. We added the following explanation of plaintiffs’ obligations: “The plaintiffs must prove that losses that Isotronics sustained would not have occurred but for Greenspan’s breach of his duty of loyalty. These would be losses that were caused by problems arising from the departure to Aegis of key managerial employees who were approached by Greenspan while he and they were still employed by Isotronics, provided that the losses were caused by events occurring before Isotronics reasonably should have replaced the departed managerial employees with competent people. See BBF, Inc. v. Germanium Power Devices Corp., 13 Mass. App. Ct. 166, 173 (1982).” Id. at 175-176.

The trial judge conducted a thirty-day nonjury trial on the damages question. He issued a memorandum of decision and *486order containing numerous findings of fact. He entered judgment against the defendants jointly and severally in the amount of approximately $40,617,000, based on compensatory damages of $14,140,000, interest on that amount, attorneys’ fees of approximately $1,216,000, costs of approximately $376,600, and a doubling of the compensatory damages under G. L. c. 93A (1992 ed.).

We reject the plaintiffs’ argument, which seeks judgment against each defendant severally for the noncompensatory damages awarded under G. L. c. 93A. We also reject the defendants’ argument that the judge erred in ruling that their conduct was a wilful and knowing violation of G. L. c. 93A. We agree with the defendants, however, that the judge’s calculation of damages was erroneous in several significant respects. Consequently, there must be a redetermination of the damages to which the plaintiffs are entitled.

1. We reject the plaintiffs’ argument that damages pursuant to G. L. c. 93A, §§ 2 and 11 (1992 ed.), should have been awarded severally. In International Fidelity Ins. Co. v. Wilson, 387 Mass. 841, 858 (1983), we said that where multiple “defendants have participated through their own individual acts in a single wrong” several liability is appropriate. Here Scherer and Aegis did not act individually. Aegis acted wrongfully only through its agent Scherer. See Pepsi-Cola Metro. Bottling Co. v. Checkers, Inc., 754 F.2d 10, 19 (1st Cir. 1985).

2. The defendants urge that we reconsider our determination in Augat I, supra at 177 n.7, that the judge properly ruled that they wilfully and knowingly violated G. L. c. 93A, §§ 2 and 11. They cite our opinion in Underwood v. Risman, 414 Mass. 96, 100 (1993), which was released after Augat I, in support .of their claim that Scherer’s acts were not wilful and knowing within the meaning of those words in § 11. The Underwood case, which concerned the nondisclosure of information of which the defendant was not aware, is not relevant here. The judge was warranted in concluding that Scherer and Aegis knowingly and intentionally acted in violation of G. L. c. 93A. Scherer knew that Greenspan held a responsi*487ble top managerial position at Isotronics and was soliciting the departure to Aegis of key employees while still an Isotronics employee.

3. We come to the damages question and start with a brief description of the judge’s conclusions that led him to the compensatory damage award of $14,140,000.

The judge found that, in the period from January 1, 1985, through March 31, 1987, Isotronics sustained a loss of profits because of “the disruption of its entire business caused by the departures” of the three key employees in late 1984 and early 1985. Isotronics had held “a dominant market position and a virtually unbroken track record of profitability.” In 1984, however, Isotronics had not had a good year, but that was due to extraordinary conditions.

The judge assumed that Isotronics’s sales would have increased at an annual rate of 20% from 1983 into 1987. He further concluded that, except for adjustments to reflect changes in the cost of gold,3 Isotronics’s profits would have been 17% of sales. The judge then adjusted the resulting profit calculation by subtracting various identifiable expenses that were not caused by the departure of the three key employees. The judge made no downward adjustments in his hypothetical profit calculation to reflect either Isotronics’s unsuccessful attempt to operate a plant in San Antonio or the effect of Aegis’s competitive impact on Isotronics’s actual sales. The judge then calculated the differences between his projected before-tax profits for 1985, 1986, and the first quarter of 1987, and Isotronics’s actual operating results during the same periods, and came to damages attributable to the departure of the three key employees of $5,629,000 in 1985, $6,914,000 in 1986, and $1,597,000 in the first quarter of 1987. The judge cut off damages as of April 1, 1987, because “the effect of other events unrelated to the departures began to be felt after [March 31, 1987].” The judge made *488no finding that Isotronics had lost a customer or a particular sale or incurred a measurable extra expense as a result of the departure of the three key employees.

The plaintiffs had to prove that Isotronics sustained monetary losses due to the defendants’ wrongdoing. Jet Spray Cooler, Inc. v. Crampton, 377 Mass. 159, 180 (1979). It was not sufficient simply to show Isotronics’s projection of its sales and the historic return on total sales. Id. Many factors bear on the financial performance of a company. The plaintiffs had to show the portion of Isotronics’s losses, at least in general terms, that was attributable to the defendants’ misconduct. See BBF, Inc. v. Germanium Power Devices Corp., 13 Mass. App. Ct. 166, 175-176 (1982). Damages for lost profits are recoverable only when proof is made “with sufficient certainty.” Jet Spray Cooler, Inc. v. Crampton, supra at 181. BBF, Inc. v. Germanium Power Devices Corp., supra at 176-177 (plaintiff must establish that “harm had a reasonably ascertainable monetary value”). The problem with the judge’s decision is that his conclusion concerning damages is not supported by his underlying findings.4

*489The judge erred in concluding that, but for the departure of the three employees whom Greenspan wrongfully solicited, Augat’s sales would have increased in 1985, 1986, and part of 1987 at an annual rate of 20%. If this conclusion is a finding of fact, it is clearly erroneous. Mass. R. Civ. P. 52 (a), 365 Mass, 816 (1974). If it is an inference purportedly drawn from underlying findings, it is not warranted. In 1984, Isotronics already had two-thirds of the sales in the industry. The potential of a 20% annual growth in sales by such a company in 1985, 1986, and 1987 is not supported by the record or the judge’s findings. There is no evidence that the market expanded sufficiently to justify such a projection.5 Moreover, the judge may have misled himself on this matter because he erroneously attributed to Greenspan the judgment that sales would increase annually by 20% in 1985, 1986, and 1987. The views of one affiliated with the defendants, such as Greenspan, might be entitled to special weight. The estimate of 20% growth was made, however, by an Augat employee after this litigation had begun.

The implausibility of the judge’s projection of Isotronics’s growth figures is further demonstrated by his failure to recognize'the growing share of the market that the defendant Aegis obtained in 1985 and particularly in 1986 and 1987, when Aegis had significant sales. Aegis’s share of the market in these years was, respectively, 3 %, 20 % to 25 %, and 20 % to 25%. The judge’s decision to disregard Aegis’s market share was based on his conclusion that, in the twenty-seven month period for which he calculated Isotronics’s lost profits, Aegis would not have diverted any sales from Isotronics if the three key employees had remained at Isotronics.

The employees’ departures were wholly lawful, and Aegis committed no wrong in hiring them. The harm for which the plaintiffs are entitled to recover is the injury to Isotronics’s profits because the three employees were not at Isotronics be*490cause Greenspan and the defendants induced their departure to Aegis in breach of Greenspan’s duty to Isotronics. Any effect of the fact that these employees were at Aegis should not be ignored in determining Isotronics’s projected sales. In their original calculation of lost sales, even the plaintiffs reduced Isotronics’s projected sales to reflect sales made by Aegis.

There are further problems in the methodology advanced by Augat and accepted by the judge. The three managers whom Greenspan improperly solicited were just a few of the important people who left Isotronics to join Aegis in the relevant period. Greenspan, whom the judge found to be a highly competent and charismatic manager, also left Isotronics to join Aegis. His departure to Aegis was lawful. Augat I, supra at 175 n.5. The judge assigned no weight to the impact on Isotronics of the departure of Greenspan.

The judge erroneously attributed Isotronics’s expenses in connection with the operation and closing of its unsuccessful San Antonio plant to the departure of the three key employees. There is no causal link between the departures and the closing of the San Antonio plant which opened in September, 1984. None of the three worked in that plant, nor was any of them responsible for its operations. That plant’s operating costs and the loss that resulted from its closing are not properly attributable to the departure of the three key employees.

The judge’s consideration of the decline in the value of Isotronics as a check against his calculation of lost profits is not justified. Losses in earnings caused by the departure of the three employees states the measure of damages that this court fixed in Augat I. The judge’s attention to the decline in the value of Isotronics as a company dramatizes his willingness to attribute all Isotronics’s losses, not otherwise clearly proven to be caused by another identifiable event, to the departure of the three employees. This approach is inappropriate conceptually because it does not measure Isotronics’s loss of earnings caused by the departures of the three employees. It is, moreover, speculative, and increasingly so in each successive month after the three key employees left.

*491The remaining consideration is what should be done now. The trial judge has retired. The damages question was fought out after many days of trial. We are in as good a position to outline an answer to the damages question as a trial judge would be. Our involvement would not impinge on any parties’ right to trial by jury.

It is apparent that the judge allowed the causation claim to run out too far into the future; that he assumed a continuing growth in Isotronics’s sales that had no basis on the evidence; and that he allocated all Isotronics’s net income problems, not otherwise quantifiable, for a period of twenty-seven months to the departure of the three employees without making findings, supported by evidence, that justified that conclusion. On the other hand, although proof of the precise amount of loss is impossible, the defendants should not be permitted to escape the consequences of their wrongful conduct that caused harm to the plaintiffs if some reasonable damages calculation can be made. See Rombola v. Cosindas, 351 Mass. 382, 385 (1966); Air Technology Corp. v. General Elec. Co., 347 Mass. 613, 627 (1964); Bigelow v. RKO Radio Pictures, Inc., 327 U.S. 251, 264-266 (1946); Story Parchment Co. v. Paterson Parchment Paper Co., 282 U.S. 555, 563 (1931); Ricky Smith Pontiac, Inc. v. Subaru of New England, Inc., 14 Mass. App. Ct. 396, 426-427 (1982); W.A. Cerillo, Proving Business Damages § 112 (2d ed. 1991). The defendants should pay something for the disruptions in Isotronics’s operations that were caused by the departure of the three key employees.

Because the plaintiffs have been unable to point to a particular customer or sale that was lost or to a particular expense that Isotronics incurred because the three key employees left Isotronics, they must accept a generalized determination of damages. The impact of the departures would have been most significantly disruptive immediately after the three key employees left. We think that six months fairly represents the outside period for which damages should be awarded, that is, the period from January 1, 1985, to June

*49230, 1985.6 Sales by Aegis were not a factor because it made no significant sales during this period. We would assume that monthly sales during that period were 10%' above 1984 monthly sales and that the “historical” return of 17% on sales continued during that time. The indicated profit would have to be modified to reflect relevant adjustments for expenses not attributable to the departures that the trial judge identified as applicable to that six-month period and further adjusted to relieve the defendants of responsibility for losses attributable to the operation of the San Antonio plant. The indicated net profit should then be compared to the actual operating results for the first half of 1985 (taking one-half of the 1985 operating loss of $3,347,000). The difference would represent the compensable damages.

The only aspects of the calculation that need further analysis are adjustments for changes in the cost of gold7 and for expenses not attributable to the departures, including the ef*493fect on earnings in the first six months of 1985 of the operation of the San Antonio plant. Two of the judge’s 1985 adjustments, concerning the adverse effect on earnings of defective glass beads8 and the overhead expenses of Augat’s Reliable Plating facility, appear to be entirely allocable to the first six months of 1985. The judge who considers this case on remand may determine that other adjustments, recognized by the trial judge, should be applied. equally throughout 1985.9

The judgment is vacated, and the case is remanded for a determination of compensatory damages consistent with this opinion and for the entry of a new judgment awarding (a) interest and G. L. c. 93A damages based on the new determination of compensatory damages and (b) the attorneys’ fees previously allowed and any that may be awarded with respect to services rendered on remand. No counsel fees are awarded to the plaintiffs with respect to this appeal.

So ordered.

(concurring in part and dissenting in part, with whom Liacos, C.J., joins). I agree with the court that the judge erred in his method of determining damages and that *494the matter should be remanded to the trial court. I dissent from the court’s determination that it should depart from its traditional role of reviewing for errors of law and become the fact finder as to the time period for the assessment of damages. “We defer to the trial court’s special role as the primary fact finder in our judicial system; such obligatory deference is embodied in the ‘clearly erroneous’ standard of review.” Kendall v. Selvaggio, 413 Mass. 619, 625 (1992). “ [I]t is the duty of this court to examine the evidence and to decide the case on its own judgment, giving due weight to the findings of the judge, which will not be reversed on oral testimony unless plainly wrong.” Seder v. Gibbs, 333 Mass. 445, 446 (1956). Where, as here, the judge misapplied the law governing damages, the usual remedy is a remand to the trial court for a determination of damages based on correct principles of law. See Coggins v. New England Patriots Football Club, Inc., 397 Mass. 525, 536 (1986).

The court states, “We think that six months fairly represents the outside period for which damages should be awarded.” Ante at 491. The court bases this conclusion on the fact that “[s]ix months was an adequate time within which to adjust to the departures.” Supra at 492 n.6. The court does not suggest that six months was the only proper date to use in assessing damages. That should be a matter of evidence and fact-finding. “[Ajppellate courts must constantly have in mind that their function is not to decide factual issues de nova.” First Pa. Mortgage Trust v. Dorchester Sav. Bank, 395 Mass. 614, 621 (1985). Rapp v. Barry, 398 Mass. 1004, 1005 (1986). Therefore, I would remand this matter for further proceedings “to the tribunal charged with the task of factfinding in the first instance.” Pullman-Standard v. Swint, 456 U.S. 273, 293 (1982). The retirement of the original trial judge is irrelevant. I think it a grave error for this court to act both as an appellate court and as the trial court.

(dissenting). Today, the majority takes an unorthodox approach in reviewing a judgment; after considering the record, it determines that the trial judge’s findings are unwarranted, and it then finds facts on its own to support an outcome which, in its judgment, is warranted. This is inappropriate. I comment briefly on several aspects of the majority opinion.

1. Trial judge’s award of damages. I acknowledge that prospective profits “cannot be recovered when they are remote, speculative, hypothetical, and not within the realm of reasonable certainty.” Lowrie v. Castle, 225 Mass. 37, 51 (1916). However, the trial judge’s ultimate findings cannot be deemed speculative or hypothetical, as they are supported sufficiently by subsidiary findings, each of which stands without error. In the circumstances, the plaintiffs reasonably could not have been expected to prove damages to the penny. This should not preclude a full recovery: “[Prospective profits] need not be susceptible of calculation with mathematical exactness, provided there is a sufficient foundation for a rational conclusion.” Id. The record supports the trial judge’s subsidiary findings, and these findings constitute a “sufficient foundation” for his ultimate findings and judgment.

I see no clear error in the judge’s finding that Isotronics’s sales would have grown at a rate of 20 %. Indeed, the majority, while contending that the figure is without support in the record, acknowledges that such an estimate was made at trial. It refuses to accept the estimate because it was prof: fered by an employee of Augat. This fact, however, goes only to the weight of the evidence, and does not warrant a ruling of error. In fact, contrary to the majority’s implicit assertion, Greenspan himself had estimated Isotronics’s sales growth to be greater than 20 %. There was evidence in the record sufficient to support the finding.

I agree with the majority that the trial judge’s calculation of Isotronics’s loss in value was not justified. However, the judge did not rely on the figure in calculating damages (except to check that the actual damages award was not excessive), and thus, it should not be of concern.

*496With respect to the period of two and one-quarter years for measurement of damages, I disagree with the majority’s determination that “the judge concluded that a competent replacement had to be, in effect, a clone of a departed employee,” ante at 492 n.6. The judge’s ultimate finding is consistent with his preliminary findings regarding the complex hybrid microcircuit packaging industry, Isotronics’s former position in that industry, and the positions and experience of the three departed employees.

2. Majority’s award of damages. The majority states early in its opinion that “[djamages for lost profits are recoverable only when proof is made ‘with sufficient certainty.’ ” It goes on to set forth how, in its judgment, the trial judge’s award of damages was not supported to a sufficient certainty by his subsidiary findings and the record. The majority then finds facts on its own, contradicting those found by the trial judge, and attempts to justify doing so by stating: “ [Although proof of the precise amount of loss is impossible, the defendants should not be permitted to escape the consequences of their wrongful conduct that caused harm to the plaintiffs. . . . The defendants should pay something . . . .” This is ironic; the majority contradicts its initial stance that damages must be proven to a sufficient certainty. If the majority examined its own award of damages with the same standard it used to examine that of the trial judge, without a doubt it would be unable to conclude that it should stand.

The majority’s cut-off date of June 30, 1985, is arbitrary. That date contradicts the trial judge’s numerous preliminary findings which support his ultimate finding that causation ended at or about April 1, 1987. The majority offers no support for the date, either from the trial judge’s findings of fact or from the record. Similarly, the “assumed” sales growth figure used by the majority, as well as its profit margin figure, also contradict the trial judge’s findings of fact, and again the majority offers no support for them. Indeed, the “assumed” sales growth figure is pulled out of thin air.

If the court finds error in the trial judge’s findings of fact, or if it determines that a conclusion or ultimate finding of the *497trial judge is not supported by the preliminary findings, it should so state. If it is able to reach alternative ultimate findings based on the trial judge’s preliminary findings, it rightly may do so. See Simon v. Weymouth Agric & Indus. Soc’y, 389 Mass. 146, 148 (1983) (“[inferences from the basic facts . . . are open for our decision . . . .” [emphasis added]). It may not, however, ignore both the record and the trial judge’s findings of fact by baldly “assuming” facts on its own to reach an alternative conclusion which it feels is appropriate.

Damages must be — and were, without error — measured in accordance with the facts as found by the appropriate fact-finding body, not by this court.