OPINION
FELDMAN, Chief Justice.
Released after arrest and nine months’ incarceration, Michael J. Smith (“Smith”) brought a tort action seeking damages under various theories, including violation of his civil rights under 42 U.S.C. § 1983, against the City of Phoenix (“City”), two Phoenix Police Department detectives (“detectives”), and Maricopa County (“County”) (collectively “Defendants”). The trial court denied all of Defendants’ motions for summary judgment on Smith’s state tort law and federal claims. Defendants then petitioned the court of appeals for relief by special action.1 The court accepted jurisdiction and ordered the trial court to enter summary judgment for all Defendants on the claims asserted under 42 U.S.C. § 1983. City of Phoenix v. Superior Court, 180 Ariz. 472, 885 P.2d 160 (App.1994).
We granted review to consider the doctrine of qualified immunity, the issue of municipal liability, and when county officials’ actions may represent official municipal policy. We have jurisdiction under Ariz. Const, art. 6, § 5(3) and A.R.S. § 12-120.24. We affirm the order granting judgment in favor of the City and the detectives. Because the record reflects genuine issues of material fact concerning the County’s potential liability, however, we vacate the court of appeals’ opinion and its order directing entry of summary judgment in favor of the County.
We view the facts in the light most favorable to Smith, the party against whom summary judgment was granted. See Hill v. Chubb Life Am. Ins. Co., 182 Ariz. 158, 160, 894 P.2d 701, 703 (1995).2 This is especially true when the non-movant is a plaintiff alleging a civil rights violation. Gobel v. Maricopa County, 867 F.2d 1201, 1203 (9th Cir. 1989); see also Morris v. Alabama State Dept. of Indus. Relations, 435 F.2d 137 (5th Cir.1970).
FACTS AND PROCEDURAL HISTORY
Victim A, a white female, was sexually assaulted in her Phoenix apartment early on the morning of January 26,1991. The assailant demanded her ATM access number, an unusual act in such attacks, and stole her purse and other belongings. The victim described her attacker to detectives as a muscular black male, 28-34 years old, with fine features and bushy eyebrows. He wore gloves and cotton sweat-type pants and used profanity. That same day, detectives learned that a person resembling this description had been seen peeking in windows and following female tenants at a nearby apartment complex. Burglaries followed two such sight[313]*313ings. Smith previously lived in this complex, and the manager suspected him of these activities. Based on this information, the investigation focused on Smith. In addition, Victim A identified him in a photographic lineup, even though his physical description did not match the description she gave the police.3 Smith was arrested that day and later indicted for attempted sexual assault of Victim A and the burglary of her apartment.
One week later, Victim B, also a white female, was sexually assaulted in her home early in the morning by a black male. Victim B’s attacker also demanded her ATM access number and stole bank cards and other belongings. The striking similarity to the attack on Victim A led Phoenix Police Detective Edgcombe to believe that both crimes were committed by the same person. The detective was so convinced that he felt compelled to check whether Smith was still in jail, which he was. Police later arrested Warren Isaac, who matched Victim B’s description of her attacker.4 In Isaac’s apartment, the police found Victim A’s purse, credit cards, and other property, as well as property stolen from a number of other victims of similar crimes. Now surely convinced that Isaac, and not Smith, had attacked Victim A, a Phoenix police sergeant urged the deputy county attorney assigned to Smith’s case to drop the charges.
The county attorney’s office, however, did not drop the charges. Nor did it disclose this exculpatory information to Smith’s defense attorney. Instead, the ease deputy conferred with other prosecutors and arranged to meet with the police sergeant and the county attorney’s Criminal Trial Division Supervisor (“supervisor”). The deputy prosecutor’s log clearly indicates that he was unsure how to proceed when the police relayed the exculpatory information and asked that Smith be released; it also shows that the supervisor decided to continue prosecuting Smith on the basis of Victim A’s identification.5
In response to the exculpatory evidence, the supervisor arranged a live lineup with both Smith and Isaac to determine if Victim A would again select Smith. While the lineup participants were being prepared, the police sergeant told Smith that the purpose of the lineup was to prove that he was innocent. However, police did not tell Victim A that they believed another man was the perpetrator in her case or that her property was found in another man’s apartment. In fact, although she was shown Smith’s clothing to see if she could identify any of it as the clothes worn during the assault—she could not—Victim A was never shown any of Isaac’s clothing.
Because Victim A indicated that she could recognize the perpetrator by voice as well as sight, the lineup participants repeated statements made during the crime, such as “What’s your PIN number?” and “I’m gonna kill you bitch.” Victim A again identified Smith, saying that “he looks like the man that came into my house.” Based on Victim A’s subsequent identification, Smith remained in jail charged with her sexual assault. Smith argues that the lineup identification was tainted by the Victim’s prior identification of him in the photographic lineup, and that even though part of the identification was based on a voice exemplar, the visual identification tainted the voice identification. Smith further complains that the lineup was flawed because police neither .showed Isaac’s picture to Victim A prior to [314]*314the lineup, nor told her that there was a second suspect.
From March until late September, police conducted no further investigation of Smith’s or Isaac’s potential involvement in the assault on Victim A. Police found no fingerprints or other physical evidence tying Smith to Victim A, and she could not identify any of Smith’s clothing as the type worn by her attacker. Thus, the only evidence implicating Smith was Vietim A’s identification.
In September, police assigned a new lieutenant in the sex crime division, Alton Thiele, to Smith’s case. He learned of the Isaac crime, its similarity to the one for which Smith was charged, the general belief of the police detectives that Smith was not guilty, and that the police had relayed this information to the county attorney’s office with a request that the charges be dropped. Thiele had a meeting with the deputy county attorney and the police sergeant on September 30 and again requested that the charges against Smith be dropped. After the meeting, the case prosecutor wrote in his log:
Meeting with [sergeant] + his lieutenant A1 Thiele. They requested that the charges be dismissed since they believe we have the wrong guy and them captain sent them + concurs—also Det. Marco Ling + Mike Sechez both believe Defendant is the wrong guy—they all feel its (sic) Warren Isaac____ We agreed to let defendant take polygraph—if he passed we would dismiss.
The results of the polygraph administered to Smith approximately nine months after his initial arrest were inconclusive. Nevertheless, the deputy county attorney moved to dismiss the charges against Smith based on new evidence. City of Phoenix, 180 Ariz. at 478, 885 P.2d at 166.6 On October 8, 1991, the County dropped the charges against Smith and he was released from jail. His release occurred after nine months’ incarceration, the last eight served after the police informed the county attorney’s office of Isaac’s arrest and probable involvement in the assault on Vietim A. Only after releasing Smith from jail did the County provide Smith’s defense counsel with the exculpatory reports of Isaac’s arrest.
Smith then filed an action against the City of Phoenix, Phoenix Detectives Edgcombe and Sechez, and Maricopa County alleging that his civil rights were violated under 42 U.S.C. § 1983; he also asserted state tort claims seeking damages for negligence, intentional infliction of emotional distress, breach of contract, malicious prosecution, false arrest, and false imprisonment. Defendants moved for summary judgment on all claims on the basis of absolute immunity.
The trial court analyzed the claims against Defendants by dividing the alleged torts into two time frames: the initial arrest and detention, and the detention and other acts occurring after February 19, 1991, when Victim A’s property was discovered in Isaac’s apartment. As to the initial arrest and detention, the trial court granted summary judgment for the City and the detectives on the basis of absolute immunity. Likewise, the trial court concluded that the County had absolute immunity for the initial charging decision and detention. However, the court found that all Defendants may be liable for the acts that occurred after February 19, 1991. Thus the false arrest, malicious prosecution, false imprisonment, and civil rights claims remained.7
Subsequently, the City and County each moved for summary judgment on the civil rights claims on the grounds of qualified [315]*315immunity from liability. The trial court denied the motions, finding that qualified immunity applied to the actions taken by Defendants, but concluded that a jury issue remained on the question of whether Defendants’ acts or omissions violated Smith’s constitutional rights. Defendants then filed a special action in the court of appeals to review the denials of summary judgment. Accepting jurisdiction, that court ordered the grant of summary judgment to the City and the detectives, finding no possible liability for their actions under § 1983. The court also directed the grant of summary judgment for the County, finding no link between the County’s acts or omissions and a policymaker with final decisionmaking authority to establish municipal liability.8 This petition for review followed.
SPECIAL ACTION JURISDICTION
As we have stated, we disapprove of appellate special action proceedings to review trial court denials of partial summary judgment. Ft. Lowell-NSS Ltd. Partnership v. Kelly, 166 Ariz. 96, 99, 800 P.2d 962, 965 (1990); see also Alhambra School Dist. v. Superior Court, 165 Ariz. 38, 40 n. 3, 796 P.2d 470, 472 n. 3 (1990). This practice often frustrates the expeditious resolution of claims, unnecessarily increases both appellate court caseload and interference with trial judges, harasses litigants with prolonged and costly appeals, and provides piecemeal review.9 See State ex rel. LaSota v. Corcoran, 119 Ariz. 573, 575, 583 P.2d 229, 231 (1978); Charles H. Googe, Jr., Qualified Immunity and Interlocutory Appeal: Is the Protection Lost when Legal and Equitable Claims are Joined?, 87 Colum.L.Rev. 161, 164 (1987).
13, 4] The court of appeals noted correctly that special action review is appropriate in § 1983 immunity claims to resolve immunity issues at the earliest possible stage. City of Phoenix, 180 Ariz. at 476, 885 P.2d at 164 (“Defendant in a section 1983 action should not be required to await final judgment in order to challenge a trial court’s adverse ruling on the issue of immunity from suit.”) (citing Henke v. Superior Court, 161 Ariz. 96, 100, 775 P.2d 1160, 1164 (App.1989)); see also Mitchell v. Forsyth, 472 U.S. 511, 524-25, 105 S.Ct. 2806, 2814, 86 L.Ed.2d 411 (1985). In fact, the very purpose of qualified immunity encourages this type of review. Harlow v. Fitzgerald, 457 U.S. 800, 102 S.Ct. 2727, 73 L.Ed.2d 396 (1982) (purpose of qualified immunity is to keep public officials out of court); see also Googe, supra at 164.10
If the civil rights violation was the only claim against Defendants, the special action proceeding might have fulfilled its purpose. But the goal of expeditious resolution is not served when state tort claims remain pending against the same defendants regardless of the court of appeals’ decision on immunity. In this case resolution of the parties’ rights and liabilities was further delayed, not resolved, by the special action because the court of appeals evidently did not find it appropriate to accept special action jurisdiction over the state tort claims against Defendants. City of Phoenix 180 Ariz. at 475 n. 1, 885 P.2d at 163 n. 1.
DISCUSSION
A. Liability under § 1983 for the City and the individual detectives
The Civil Rights Act of 1871 provides:
[316]*316Every person who, under color of any statute, ordinance, regulation, custom, or usage of any state ... subjects, or causes to be subjected, any citizen of the United States ... to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws, shall be liable to the party injured in an action at law____
42 U.S.C. § 1983. A “person” includes public officials in their individual capacity, Neal v. Georgia, 469 F.2d 446 (5th Cir.1972), and municipalities, Monell v. Department of Soc. Serv., 436 U.S. 658, 98 S.Ct. 2018, 56 L.Ed.2d 611 (1978). A different test applies to establishing official liability and municipal liability, and we address the liability of the detectives, the City, and the County accordingly.
1. The individual Phoenix police detectives
Individual officials face liability under § 1983 when they commit a constitutional tort or are personally involved in the alleged unconstitutional act. Guy v. City of Phoenix, 668 F.Supp. 1342, 1351 (D.Ariz.1987); Hinshaw v. Doffer, 785 F.2d 1260, 1263 (5th Cir.1986). It is well settled that government officials may be immune from liability for certain conduct performed in their official roles, despite the United States Supreme Court’s pronouncement that “[n]o man ... is so high that he is above the law.” United States v. Lee, 106 U.S. 196, 220, 1 S.Ct. 240, 261, 27 L.Ed. 171 (1882). See, e.g., Pulliam v. Allen, 466 U.S. 522, 539-40, 104 S.Ct. 1970, 1979-80, 80 L.Ed.2d 565 (1984). Official immunity exists to protect government officials from personal liability for them official functions by securing an early dismissal of claims. See Scheuer v. Rhodes, 416 U.S. 232, 240, 94 S.Ct. 1683, 1688, 40 L.Ed.2d 90 (1974); C. Antieau and M. Mecham, Tort Liability of Government Officers and Employees 2 (1990).
In determining whether the detectives are liable under § 1983 and before reaching the question of immunity, the court must first evaluate whether the officials’ conduct “violate[d] clearly established statutory or constitutional rights of which a reasonable person would have known.” Harlow, 457 U.S. at 818, 102 S.Ct. at 2738.11 “If the law at that time was not clearly established,” the defendant may avoid liability altogether through a dismissal on summary judgment. Id. The trial court determined that qualified immunity applied to Defendants’ acts and omissions but held that a jury issue precluded summary judgment.
The court of appeals reversed. Without ever reaching the issue of the detectives’ immunity, the court concluded that the detectives did not violate Smith’s clearly established constitutional rights. City of Phoenix, 180 Ariz. at 480, 885 P.2d at 168.12 Smith claimed that because Lieutenant Thiele ultimately convinced the County to dismiss the charges, the individual detectives could have and should have accomplished the same thing earlier. In effect, the detectives should be liable to Smith because they did not lobby more vigorously.
We disagree with Smith and think the court of appeals correctly characterized the detectives’ behavior. Contrary to Smith’s portrayal, the detectives did what the law required of them: they promptly disclosed [317]*317the exculpatory evidence and conveyed their view of the case to the prosecutor. See Walker v. City of New York, 974 F.2d 293, 298-99 (2d Cir.1992) {Brady obligations of the police require no more than disclosure to the prosecutor; police need not disclose directly to the defendant or his defense attorney), cert. denied, 507 U.S. 961, 113 S.Ct. 1387, 122 L.Ed.2d 762 (1993). Even assuming that the detectives had a duty to recommend dismissal, they fulfilled it. The detectives, however, had neither the authority nor the duty to personally effectuate Smith’s release. On the contrary, the prosecutor has sole discretion to determine whether to continue prosecution. See A.R.S. § 11-532(A)(1); State v. Murphy, 113 Ariz. 416, 418, 555 P.2d 1110, 1112 (1976). The record is quite clear that the detectives acted in a professional and proper manner. Consequently, the detectives cannot be liable under § 1983 for any injury resulting from the fact of prosecution or the manner in which it was conducted.
2. The City’s liability
If a municipality’s agents have not caused the deprivation of any constitutional or statutory rights, the entity itself is not liable for the conduct. City of Los Angeles v. Heller, 475 U.S. 796, 799, 106 S.Ct. 1571, 1573, 89 L.Ed.2d 806 (1986). Because the detectives did not violate any of Smith’s constitutional rights, their employer, the City, cannot be liable for any injury to Smith under § 1983. We therefore agree with the court of appeals’ conclusion that as a matter of law, neither the detectives nor the City is liable for the § 1983 violation alleged in this ease. The trial court should grant summary judgment to the City of Phoenix and Detectives Edgcombe and Sechez.
B. The County’s liability under § 1983
1. The doctrine of municipal liability
Not until 1978 were municipalities recognized as persons for purposes of § 1983 liability. Monell, 436 U.S. 658, 98 S.Ct. 2018, overruling Monroe v. Pape, 365 U.S. 167, 81 S.Ct. 473, 5 L.Ed.2d 492 (1961). Counties, like municipalities, are considered persons for purposes of § 1983 liability. Smallwood, v. Jefferson County Gov’t, 753 F.Supp. 657 (W.D.Ky.1991).
However, municipal liability cannot be imposed on the basis of respondeat superior or vicarious liability. Monell, 436 U.S. at 691, 98 S.Ct. at 2036. Thus, a municipality is not necessarily liable even if its agent committed a constitutional tort. On the other hand, a municipality can be liable even if the accountable employee is held to have immunity from liability. See Owen v. City of Independence, 445 U.S. 622, 647-650 and n. 18, 100 S.Ct. 1398, 1413-15 and n. 18, 63 L.Ed.2d 673 (1980) (municipality has no immunity for immune acts of its officers); Haynesworth v. Miller, 820 F.2d 1245, 1273 n. 227 (D.C.Cir.1987) (“[T]he personal immunity accorded to city officials militates in favor of municipal liability, since a contrary ruling would leave victims of unconstitutional conduct without any remedy.”) (citing Owen, 445 U.S. at 651, 100 S.Ct. at 1413); Greene v. Polk County, 406 N.W.2d 433, 435 (Iowa 1987) (“[I]mmunity enjoyed by the employee is not material as to the liability of the government employer.”); see Leatherman v. Tarrant County Narcotics Intelligence & Coordination Unit, 507 U.S. 163, 166, 113 S.Ct. 1160, 1162, 122 L.Ed.2d 517 (1993), (municipality’s protection from liability under § 1983 does not encompass immunity “either absolute or qualified”); see also Mark R. Brown, Correlating Municipal Liability and Official Immunity Under Section 1983, 1989 Ill.L.Rev. 625, 671 (“A municipality can be held liable where municipal officials, entitled to absolute immunity, violate the Constitution.”).
Although vicarious liability is inapplicable when a municipality is sued for a § 1983 violation, direct municipal liability13 is available if the municipality is “alleged to have caused a constitutional tort through ‘a policy statement, ordinance, regulation or decision officially adopted and promulgated by that body’s officers.’ ” City of St. Louis v. Praprotnik, 485 U.S. 112, 121, 108 S.Ct. 915, [318]*318923, 99 L.Ed.2d 107 (1988) (quoting Monell, 436 U.S. at 690, 98 S.Ct. at 2036). It is the plaintiff’s burden to prove the existence of such a municipal act. In the present case, the direct liability issue raised against the County involves only the policy statement prong.
A municipal policy can be established by the single act of a policymaker. Pembaur v. City of Cincinnati, 475 U.S. 469, 480, 106 S.Ct. 1292, 1298, 89 L.Ed.2d 452 (1986) (plurality). The official must have final policymaking authority to bind the municipality to his actions. Id. at 483, 106 S.Ct. at 1300. A policymaker is one who, on the relevant issue, has the final authority to make a decision from several available alternatives. Id. at 483-84, 106 S.Ct. at 1300. A municipality can be liable for the harm caused by an official with this authority because the municipal agent’s status cloaks him with the municipality’s authority. Pembaur, 475 U.S. at 481, 106 S.Ct. at 1299 (“If the decision to adopt that particular course of action is properly made by that government’s authorized decisionmakers, it surely represents an act of affirmative government ‘policy’ as that term is commonly understood.”); Praprotnik, 485 U.S. at 122, 108 S.Ct. at 923; Brown, supra at 658.
Whether an official has final policymaking authority to bind the municipality is a question of state law. Praprotnik, 485 U.S. at 123, 124, 108 S.Ct. at 924. Final policymaking authority is a question “to be resolved by the trial judge before the case is submitted to the jury.” Jett v. Dallas Indep. Sch. Dist, 491 U.S. 701, 737, 109 S.Ct. 2702, 2724, 105 L.Ed.2d 598 (1989) (emphasis added). The judge’s decision should be based on state and local law and customs and usages that have the force of law. Id.
In making his decision on the County’s qualified immunity summary judgment motion, the trial judge did not attempt to decide whether any of the County’s agents had final policymaking authority because the motions did not raise this issue. The County’s motion did not address or disclaim liability on the basis of direct municipal liability; instead it erroneously focused on the individual prosecutors’ alleged qualified immunity. The County argued that their alleged qualified immunity necessarily frees both the prosecutors and the County from liability. The County, and even the trial court, characterized the County’s motions as qualified immunity motions in numerous pleadings. This inquiry into the individual prosecutors’ qualified immunity proceeded despite its irrelevance to the direct liability claim against the County and despite the fact that the individual prosecutors were not even named as parties to the action.
Because the prosecutors were not defendants and the County could not benefit from then’ immunity, the whole inquiry was an exercise in futility. Instead, the County should have focused its summary judgment motion on direct municipal liability, the only theory on which it could be held liable. Because the County’s failure to raise the issue of direct municipal liability in its motions not only confuses the record but is central to our disposition, we will briefly address the County’s pleadings.
2. The County’s motion for qualified immunity—the real issue before the court of appeals
So far as the record shows, the County’s motion for summary judgment did not touch on the issue of direct liability. In that motion, the County did not argue that there was insufficient evidence to show that a policymaker decided to continue prosecuting Smith without disclosing the Isaac exculpatory evidence. Instead, the County argued that it was entitled to summary judgment entirely on the basis of the qualified immunity allegedly enjoyed by the individual prosecutors. Smith’s response to the County’s qualified immunity motion is not a part of the record.14 In its reply, the County merely [319]*319repeated the qualified immunity argument made in its motion. The sufficiency or existence of evidence on the policymaker issue pertaining to the County’s direct liability was thus not an issue raised in the County’s motion.15 Nor did the trial court consider or decide the issue.16
In its special action, the only argument the County presented to the court of appeals was whether the actions of the Maricopa County Attorney’s Office are entitled to qualified immunity. In his response to the petition for special action, Smith again argued that the actions at issue were taken in accordance with a municipal policy, and that the County’s disclaimer of vicarious liability is inapplicable to its direct liability. Smith also correctly noted that under Owen the County has no qualified immunity defense. The County finally addressed direct municipal liability in its reply to Smith’s response. Thus, the County addressed direct municipal liability for the first time in the court of appeals, in a reply brief, having avoided the issue altogether in the trial court and in its initial argument to the court of appeals.17
The basis for the County’s denied motion for summary judgment, and consequently the issue before the court of appeals, was qualified immunity. Assuming the question was relevant, the court of appeals should have addressed only the propriety of the trial court’s ruling on the County’s qualified immunity motion because this was the only issue developed on the record. The court should not have reached the issue of direct liability because direct liability is a question of law that turns on the facts. Jett, 491 U.S. at 737, 109 S.Ct. at 2723-24; Fountain Hills Civic Ass’n, Inc. v. City of Scottsdale, 152 Ariz. 569, 575, 733 P.2d 1152, 1158 (App. 1986). In the present case, those facts have not yet been put in issue by a properly focused motion.18
[320]*320We must therefore determine whether, on the record before it, the court of appeals could conclude that the trial judge erred in denying the County’s motion for summary judgment because the County could not be directly liable as a matter of law. Resolution of this question must be made with appropriate deference to the fact that the issue was not raised in the County’s motion for summary judgment.
C. Was there a genuine issue of material fact regarding direct municipal liability?
Having stated the framework applicable to the County’s direct liability, we must determine whether, under state law, there was sufficient evidence on the record that a policymaker with final decisionmaking authority made the decision to pursue prosecution and withhold evidence.
As correctly noted by the court of appeals, a county attorney can be a policymaker for § 1983 purposes. City of Phoenix, 180 Ariz. at 485, 885 P.2d at 173 (citing Gobel, 867 F.2d at 1207-08). The court of appeals assumed without explanation that only the county attorney maintains final policymaking status. The court labeled the deputy county attorneys subordinate county officials and found no evidence on the record that the county attorney supervised or directed the alleged civil rights violations. Id. The court in effect granted summary judgment for the County on an issue not raised and considered in the trial court because Smith failed to “link the act or omission of the county to a policy maker with final decision making authority,” the county attorney. Id. at 485, 885 P.2d at 173. The court of appeals made this decision without Smith’s response to the County’s motion in the record.
We believe it is incorrect to direct entry of summary judgment on issues not raised by the movant in the trial court and on which the parties have therefore not had an opportunity to marshal and present evidence.19 This is especially true when the court of appeals bases its decision on evidence sufficiency and one of the party’s motion papers are not part of the record. We also disagree with the legal conclusion that only the county attorney can be the final policymaker in the county attorney’s office. The Maricopa County Attorney’s office is the largest in the state and employs approximately 200 lawyers. The county attorney surely does not personally make decisions in each case being prosecuted.
The United States Supreme Court has rejected arguments similar to the view taken by the court of appeals. In Jett, a teacher sued a public school district after his principal transferred him for allegedly discriminatory reasons. After concluding that the lower courts applied the wrong standard, the Court remanded the case to determine whether final policymaking authority lay with [321]*321the individual school principal, the district superintendent, or the district itself. 491 U.S. at 737, 109 S.Ct. at 2724 (remand necessary because “the trial judge must identify those officials or governmental bodies who speak with final policymaking authority for the local governmental actor”) (emphasis added). The Court’s inquiry in Jett would have ended if the highest rank or “name” official did not cause the alleged injury. See also Praprotnik, 485 U.S. at 124-26, 108 S.Ct. at 925-27 (a policymaker can be the highest official in the department, or the person given the decisionmaking responsibility in matters of the kind at issue).
In Praprotnik, the Supreme Court expressly recognized that the determination of who has final policymaking authority is a fact-intensive inquiry and a decision left for the trial judge. 485 U.S. at 125, 108 S.Ct. at 925.20 And although Jett said that the court should make this determination before “the case is submitted to the jury,” this does not necessarily mean in summary judgment proceedings directed to other issues. 491 U.S. at 737,109 S.Ct. at 2724.
Had the County’s motion been directed to the proper analysis of direct municipal liability, and if Smith had produced no more than is in the record now, a judgment for the County may have been appropriate. Given the issues raised—and not raised—in the County’s motion, however, it is no surprise that the court of appeals found little evidence on the issue of municipal liability. But this finding puts the cart before the horse. Had Smith been on notice that the issue would be the subject of summary judgment proceedings, he may have produced evidence or conducted discovery establishing that the decision not to disclose the exculpatory facts was made by a policymaker with final decision-making authority. See Ariz.R.Civ.P. 56(c). The County’s failure to recognize its own basis for liability at the trial court cannot be saved by the appellate court’s recognition of the issue when the deficiency results in unexplored factual issues.
Although the record is sparse, it undisputedly shows that the question regarding the manner of prosecuting Smith’s charges went up the chain of command in the county attorney’s office and that a supervisor ordered that the prosecution be continued, without ordering disclosure of the Brady material. Given the procedural posture, and consequently the few facts available, this court cannot safely resolve the issue of whether the County’s criminal trial supervisor was expressly or by custom delegated the necessary decisionmaking authority to impose liability on the County, thus prohibiting dismissal of the claim against the County. See Jett, 491 U.S. at 738, 109 S.Ct. at 2724. Because appellate courts must review in a light most favorable to the non-moving parties and civil rights actions should be read broadly, the court of appeals should not have ordered summary judgment. See, e.g., Gobel, 867 F.2d at 1203. This case must be remanded to provide Smith with an opportunity to discover and marshal the relevant facts.
D. The dissent
The dissent argues that Smith did not raise the policymaker issue in either the trial court or the court of appeals. We disagree. Smith raised the issue. The County did not, arguing instead in both courts only that it could not be liable because its employees had qualified immunity. This issue, of course, was irrelevant because the County could not be held vicariously liable and therefore could not be protected from direct liability under the theory that its employees had qualified immunity. Thus the dissent’s comment that “Smith never properly raised the [policy] issue in the trial court ... and thus there was no occasion for the County to raise the issue in the court of appeals” is both not relevant and incorrect. Post at 325, 909 P.2d at 392. Smith had no occasion to argue policymaker liability in response to a motion based on qualified immunity, but the question of direct, policymaker liability was in the case and had been raised in prior motion papers. See ante notes 14r-16.
[322]*322The dissent examined the complaint to conclude that “Smith never pleaded a ‘direct liability claim against the County.’ ” Post at 323, 909 P.2d at 390. As noted by the dissent, the complaint is not a part of the record. We do not base our decision on a document outside of the record. Even so, the dissent’s reference to the complaint in no way supports the court of appeals’ decision because that court acted without the complaint in departing from both the record and the motion to reach the conclusion that the County could not be directly liable.
The dissent uses the complaint to support the court of appeals’ decision, arguing that Smith “failed to allege the violation of a policy.” Post at 323, 909 P.2d at 390. But Arizona is a notice-pleading state. All of the cases cited by the dissent for the need for so specific an allegation are cases in which the municipal defendant moved to dismiss on the plaintiffs failure to state a claim upon which direct relief could be granted. Post at 324 n. 2, 909 P.2d at 391 n. 2. The County did not raise that argument in its motion. It based its motion for summary judgment only on qualified immunity.
There is no real disagreement between the majority and the dissent regarding the law of municipal liability under § 1983. The case law clearly states that “a municipality may be sued directly if it is alleged to have caused a constitutional tort through ‘a policy statement, ordinance, regulation, or decision officially adopted and promulgated by that body’s officers.’ ” Praprotnik, 485 U.S. at 121, 108 S.Ct. at 923 (quoting Monell, 436 U.S. at 690, 98 S.Ct. at 2036). We might agree with the dissent that nothing “of record ties what happened here to a policy,” but this merely reiterates our point. Post at 325, 909 P.2d at 392 (emphasis added). The insufficiency of the evidence in the record on the policy issue prohibits granting summary judgment to the County because the County’s motion would not put Smith on notice to marshal facts on that issue.
The dissent argues that we have intervened in a case with an incomplete record: “the majority’s opinion chiefly discusses Maricopa County’s liability on the basis of policy or custom. Yet that issue is not presented for review.” Post at 323, 909 P.2d at 390. This is precisely the criticism that should be directed at the court of appeals. Once we discover that the court of appeals incorrectly decided the case on an issue not before it, this court must either ignore the error or act to correct it. The dissent chooses the first course, but we believe the second is a better choice. Disagreement with this court’s intervention does not square with affirming the same act taken by the appellate court. When the court of appeals accepted jurisdiction and went beyond the motion presented before it, this court has an obligation to intervene when the appealed decision was made on an insufficient record.
DISPOSITION
The procedural posture and issues of material fact preclude summary judgment. Smith is entitled to a factual determination whether the decision to continue his prosecution and to withhold the exculpatory Brady material was made by an official with final policymaking authority, as that term has been defined by the United States Supreme Court.
We vacate the court of appeals’ opinion and remand the issue of the County’s direct liability for further proceedings consistent with this opinion, including dismissal of the § 1983 claims against the detectives and the City.
MOELLER, V.C.J., and CORCORAN and ZLAKET, JJ., concur.