RIPPLE, Circuit Judge.
On September 12, 1999, Friday Gardner was shot to death in front of an apartment building on the south side of Chicago. The State of Illinois charged three men, including Michael Carter, with Gardner’s murder. Mr. Carter was tried alongside his brother, Michael Stone, in a single trial. Both were convicted of murder; Mr. Carter was sentenced to thirty years’ imprisonment. Following an unsuccessful state postconviction proceeding, Mr. Carter filed a pro se petition for habeas corpus in the district court under 28 U.S.C. § 2254. The district court denied relief on each of the eight grounds presented in his petition and also denied a certificate of appealability, see 28 U.S.C. § 2253(c). We granted a certificate as to a single claim — -whether Mr. Carter received effective assistance of counsel. We also appointed appellate counsel.
Mr. Carter brings to us an ineffective assistance claim. His claim turns on the potential effect" of the testimony of two witnesses who were not called in his defense at trial. The Illinois Appellate Court determined that'the proffered testimony would not have changed the outcome of the trial. Although the state court’s analysis stumbles in some respects, we nevertheless must conclude that its decision was not unreasonable. Accordingly, given our deferential standard of review, we affirm the district court’s judgment denying habeas relief to Mr. Carter;
I
BACKGROUND
A.
Gardner was murdered in the course of a heated argument about the robbery of an apartment occupied by Stone and other of Mr. Carter’s relatives. Mr. Carter and Stone believed Gardner to have been responsible.
At the time of Gardner’s murder, Mr. Carter and Gardner. each had relatives living on the second floor of a building at [934]*93461st and May in Chicago. In one apartment, Gardner’s cousin,, Antonio Phillips, lived with his mother, Rena Phillips, and her other children. Stone lived next door, in an apartment he shared with his cousin (and Mr. . Carter’s cousin) Felicia Anderson,- her fiancé, Corey Grant, and their children. Both .Gardner and Mr. Carter — neither of whom lived in the building — at times visited their- relatives there.
On the afternoon of September 12,1999, the events culminating in Gardner’s murder later that night began to unfold: a robbery, a search for the robber, a larceny in. retaliation, and, eventually, a heated argument about the robbery that ended in gunfire. First, two men broke into the apartment occupied by Mr. Carter’s relatives and robbed Grant at gunpoint. The robbers took money, jewelry, and marijuana packaged for sale. Grant was not alone at-the time of the robbery; another of Mr. Carter and Stone’s cousins, Michella Anderson, was present, as were others, although the apartment’s other occupants, Felicia Anderson and Stone, were not. One of the robbers had a gun, and, while no one was shot, Grant was struck with a baseball bat on his head in the course of the robbery. After the perpetrators fled, Grant ran next door and began pounding on the door and shouting about the robbery. Antonio Phillips emerged with Gardner, who was visiting at the time, and, according to eyewitnesses, both' joined the unsuccessful effort by Grant and'Michella Anderson to chase down the robbers.
Later, Stone and Felicia Anderson returned home, accompanied by Felicia’s sister, LaTonya Cheeks. At some point after learning of the robbery, Stone called Mr. Carter for help in determining the identity of the robbers. Mr. Carter came to the building with a friend, Cortez Jones. Jones stated that he had heard from a friend that Gardner'had been selling packaged marijuana at - another location and bragging about robbing someone at 61st ■and May. Grant, the robbery victim, denied Gardner’s involvement,, stating that Gardner in fact had tried to chase down the assailants with him. Grant was unable to convince Stone, Cortez, and Mr. Carter. At some point during the day, Stone acquired a gun and stashed it in a locked basement storage area in the apartment building.
According to witnesses at Mr. Carter’s murder trial,'later that evening, Mr. Carter, Stone, and Jones broke into Gardner’s van, which was parked outside the apartment building, stole his radio, and left. Gardner saw them through a window and yelled at them to stop and that he had nothing to do with the earlier robbery. Gardner called his friend, Tommy Gaston, who arrived sometime later. Gaston and Gardner met downstairs and went to the van, where they observed that the radio had been stolen. Gardner’s relative, Rena Phillips, and her boyfriend, Paul Cálmese, arrived and began talking with Gardner and Gaston outside.
While Gardner and the others were still outside, Mr. Carter and Jones returned, and a heated argument ensued. Stone witnessed the beginning of the exchange from upstairs and went to retrieve the gun from the storage space. Now armed, he emerged from the adjacent alley and headed towards the argument on the street. In the ensuing minutes, Gardner was shot fatally, struck by two bullets in the abdomen. Stone, Jones, and Mr. Carter all fled the scene.
Police officers patrolling the area were in close enough proximity to view flashes and hear gunshots. Officer Cedric Taylor ran to the scene, arriving within seconds. He attempted to chase the assailants but lost sight of them and returned to the [935]*935scene. Officers immediately began taking statements from the numerous eyewitnesses.
B.
The State charged Mr. Carter, Stone, and Jones with the murder of Gardner. In 2002, Mr. Carter and Stone proceeded to trial together, represented by separate counsel. Jones was tried in a separate proceeding.
In Mr. Carter and Stone’s trial, the State’s theory was that two sets of shots were fired:' an initial set by Stone and a second set from either Jones or Mr. Carter. It argued that all of the defendants were armed and that they fled after the fatal shots were fired and discarded their weapons. The State told the jury that, under an accountability theory, it was-irrelevant which individual actually had fired the fatal shots.
Stone admitted to shooting Gardnér and presented both self-defense and defense of others as a justification. Stone contended that he had seen Gardner point a gun, or attempt to point a gun, at Mr.' Carter, and that only then did Stone fire. Mr! Carter argued that he was unarmed and was not responsible for the actions of Stone'or Jones.
The eyewitness statements, both to the police and at trial, varied widely, and several witnesses testified inconsistently with their prior statements to. law enforcement in the days following Gardner’s murder. Based on those initial statements given to police and testimony before the grand jury, the State assembled a witness list to support its case, consisting of law enforcement officers, Lenisha Pearson (Gardner’s then-girlfriend), Grant, Felicia Anderson, Rena and Antonio Phillips, and Cheeks.1
Except for the officers, each of the witnesses who testified at trial had some preexisting relationship either to Gardner or to the alleged perpetrators. The first group had a relationship to the victim, Gardner. At trial, Gardner’s relatives, Rena and Antonio Phillips, testified that the shots fired at Gardner came from the direction of Mr. Carter and Jones, not Stone. Both testified that they saw Jones fire shots first, and both testified that they saw Mr. Carter with a gun. Rena Phillips testified that she saw Mr. Carter shoot as well. Antonio Phillips testified that by the time the second set of shots were fired, he was running toward the scene from his prior vantage point. Both testified that they saw Mr. Carter and Jones fiee. Neither witness said anything about Stone. Pearson,. Gardner’s then-girlfriend, testified that she had heard two sets of shots, the first of which came from Stone, and the second of which came from either Mr.. Carter or Jones. She also testified that she saw Gardner throughout the exchange and that he did not have a gun. Gaston, a friend of Gardner’s, testified that he heard, but did not see, an initial shot. He then saw Jones pull a gun from his pocket and shoot at Gardner four times, these shots being louder than the initial single shot. He testified that he then saw Mr. Carter and Jones flee. He did not see Gardner with a gun that night.
The second set of State’s witnesses were Mr. Carter’s relatives and friends — Felicia Anderson, Grant and Cheeks. Although their testimony was part of the State’s case-in-chief,'it was more problematic for the prosecution. Felicia Anderson, Mr. Carter’s cousin, testified that she did not [936]*936see the shooting, but that she did see Gardner with a gun and also saw Gaston remove it from the scene. She was impeached, however, with her prior signed statement to police and her grand jury testimony. In those earlier statements, she had said that she was standing near the shooting, heard shots, turned to .face the scene, and witnessed Mr. Carter pointing a gun at Gardner. She left the scene shouting that Mr. Carter had “shot him.” 2 At trial, she attempted to explain these discrepancies by claiming that her earlier statements were based on what others had told her rather than what she personally-had observed.
Grant, Felicia’s fiancé, claimed to have been inside the apartment during the shooting. Grant testified only that he saw-people running from the scene after hearing shots. He stated on cross-examination that he had seen Gardner with a gun earlier in the day when he had tried to chase down the people who had robbed Grant.
Cheeks, Felicia’s sister, also testified at trial that she saw Gardner pull a gun and point it at Mr. Carter and that Stone shot in defense. She later contradicted that testimony and stated that she had seen Gardner point his gun in the air rather than at Mr. Carter. Also, Cheeks initially claimed that only Stone shot at Gardner, but on cross she stated that she heard shots from the area where Jones and Mr. Carter were standing. Her prior signed statement and grand jury testimony were published to the jury. In them, Cheeks stated that she did not see anything in Gardner’s hands, that Stone had shot first and fired three times, and that a fourth shot was fired by either Jones or Mr. Carter.
The physical evidence presented at trial showed that Gardner was shot twice with .380 caliber bullets, and the parties stipulated that they had been fired from the same gun. Three .380 cartridge casings were also recovered from the scene within a few feet of the blood stain from Gardner’s body, but it could not be determined if they were from the same weapon that killed Gardner. No other evidence linked the bullets or the casings to any particular defendant. More than one witness, including a police officer stationed nearby who heard the shots, testified that there was an initial set of shots and a second set, and the two sounded somewhat different. Gardner’s autopsy report showed no evidence of close-range firing, although multiple witnesses had stated that Jones and Mr. Carter were within a few feet of Gardner at the time of his murder.
The attorneys representing Stone' and Mr. Carter, respectively, called only two defense witnesses: Michella Anderson, a cousin of Stone and Mr. Carter, and Stone himself. Michella Anderson testified that she saw Gardner pull a gun and point it in the air. A shooter, whom she could not identify, entered from the alley and fired. She also testified that, after Gardner was shot, Gaston retrieved Gardner’s gun, put it in. his. car, and drove away.. Unlike Felicia Anderson and Cheeks, Michella Anderson had not made a statement to the police immediately after the incident and was "not called to appear before the grand jury. She did testify that she had met with attorneys for the State and for the defendants at various points during the investigation and had informed each of them that she had seen Gardner with a gun.
Stone testified that he alone shot Gardner and that he did so only after Gardner pointed a gun at Mr. Carter. He claimed that he was the only shooter and that he [937]*937never saw Mr, Carter with a gun that evening. Mr. Carter did not testify.
In closing arguments, Stone’s attorney focused on his self-defense theory and relied in significant measure on Michella’s testimony that Gardner, the victim, was visibly armed. Mr. Carter’s attorney focused on the lack of evidence of any close-range firing and the evidence that Mr. Carter was within a few feet of Gardner at the time of the shooting. He contended that the evidence could support that Jones and Stone had fired their weapons, ■ but that there had been no evidence that Mr. Carter knew that Jones was armed or that Stone was there at all. The State’s attorney countered that it did not matter who did the shooting because the evidence demonstrated that all three men were armed. ■ Further, the State noted that Mr. Carter was responsible for bringing, Jones to the scene, and, following the shooting, Mr. Carter, Stone, and Jones fled the scene while the other witnesses remained.
The jury found both Stone and Mr. Carter guilty of first-degree murder. Jones also was convicted of first-degree murder in a separate proceeding. Each was sentenced to thirty years’ imprisonment.
Mr. Carter and Stone filed a joint direct appeal alleging multiple points of error, including that trial counsel was ineffective for failing to preserve objections to evidence about the marijuana stolen from the home or to otherwise request a limiting instruction, as well as other claims not relevant to the present petition. No other ineffective assistance claims were raised. In the course of affirming the trial court’s judgment, the Appellate Court of Illinois noted “that the evidence in this case was not closely balanced.”3 The Illinois Supreme Court denied leave to appeal.
C.
In 2005, Mr. Carter and Stone filed a joint pro se petition for postconviction relief in the Circuit Court of Cook County. In the petition, they asserted that trial and appellate counsel had been ineffective. They contended specifically that trial counsel had been ineffective for failing to call two additional witnesses, Jeremiah McReynolds and Paul Cálmese, failing to impeach certain witnesses, and .emphasizing Stone’s self-defense theory over Mr. Carter’s mere presence theory. They further contended that counsel had failed to present available evidence to support the claim of self-defense. Finally, they claimed that appellate counsel was ineffective for failing to raise the ineffective assistance claims based on trial counsel’s performance.
Attached to the petition were several affidavits, including one from McReynolds. McReynolds , stated that he also lived at 61st and May and that, on the night of the murder, had heard a commotion, looked outside, and observed the argument across the street. During the argument, McRey-nolds “observed Friday Gardner pull an object out from behind his back and then ... heard several shots ring out from the alleyway.”4 He identified the shooter as “Man,” a nickname trial witnesses had indicated belonged to Stone.5 McReynolds stated that he “did not personally observe anyone else doing any shooting,” and that he observed Mr. Carter and Jones “scatter in an effort to avoid being shot.”6 He indicated that he had shared his account with Cheeks and offered to testify and that [938]*938he eventually learned that he had been placed on Mr. Carter’s witness list. McReynolds’s affidavit concluded: “Although I was available and in Chicago, Illinois at all times in which the trial was going on, no lawyer or anyone else from the court contacted me or called me as a witness about the facts that happened on September 12,1999.”7
In support of the assertion regarding Calmese’s probable testimony, Mr. Carter attached a police report to his petition. Following the shooting, Cálmese told police that he had been talking with Gardner when Mr. Carter and Jones arrived and the argument began. He saw that Jones had a gun during the argument. He also saw that someone came from the alley and shot at Gardner, and Jones subsequently also shot at Gardner.'
The Illinois circuit court denied the petition and a subsequent motion to reconsider. Mr. Carter appealed to the Illinois Appellate Court and was represented by the Office of the State Appellate Defender. The State appellate court affirmed.8 It reached the merits of Mr. Carter’s ineffective assistance arguments and* rejected them. It first set forth the familiar twor prong deficiency-and-prejudiee standard under Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), citing both Strickland and a state case, People v. Coulter, 352 Ill.App.3d 151, 287 Ill.Dec. 255, 815 N.E.2d 899, 905 (2004). The court then continued, again citing Coulter: “To show prejudice, defendant must show that counsel’s deficient performance rendered the result of the proceeding unreliable or fundamentally unfair.” 9
In its analysis, the court skipped the deficiency prong and stated that “even if trial counsel’s failure to call McReynolds and Cálmese to'testify fell below an objective standard of reasonableness, defendant’s claim fails because he is unable to show resulting prejudice.”10 The court noted' that the evidence in the case was “not close,” and the ¡testimony of McRey-nolds and Cálmese “would not have been exculpatory and would have merely been cumulative of the testimony presented by Felicia Anderson, LaTonya Cheeks, Mic-hell[a] Anderson, and co-defendant Stone. Defendant’s theory of défense,” it concluded, “was presented at trial and corroborated where these witnesses testified that the victim had a gun and defendant did not.”11 The court further noted that McReynolds’s affidavit was “insufficient and unsupportive of defendant’s defense” because it did not establish that he was unarmed or uninvolved, .only that McReynolds had not seen a gun.12 Finally, the court concluded that “the result of defendant’s trial would have been the same even if Cálmese and McReynolds had .testified that .defendant was unarmed where the jury was given” an accountability instruction that focused on whether a defendant “solicits, aids, abets, agrees to aid, or attempts to aid the other person in the planning or commission of the offense.”13 Accordingly, “[a]ny possible testimony from McReynolds and Cálmese that defendant was unarmed would not have illuminated whether defendant was legally accountable for the ac[939]*939tions of his co-defendants.”14 The court closed by noting that, on direct appeal, it had concluded “that the evidence here was not closely balanced,” and that Mr. Carter “is unable to show resulting prejudice where the alleged deficient performance of counsel did not render the proceeding unreliable or fundamentally unfair.”15
,D.
Mr. Carter next filed a pro se petition for a writ of habeas corpus in federal district court, seeking relief on eight separate grounds. - Included among those claims were a free-standing actual innocence claim, claims related to a denial of a fair trial, claims related to excessiveness of sentence, and the claim regarding ineffectiveness of trial ’counsel for failure to call McReynolds and Cálmese. The district court determined that all of the claims, save for ineffective assistance, were either procedurally defaulted or not cognizable on federal habeas review and rejected them. Turning to the only non-defaulted, cognizable claim, the court held that Mr. Carter had failed to “overcome the presumption that counsels decision not to call these witnesses was reasonable.”16 The record made clear that counsel was aware of McReynolds and may have decided not to call him because of his significant criminal record. Counsel may also have decided that the testimony of , either witness was duplicative of other testimony, or was “unavailing because Carter was charged under an accountability theory, -meaning that he could be found legally accountable for his co-defendant’s actions even if he did not fire at Gardner.”17 The court then determined that, even if performance was deficient, Mr. Carter could not demonstrate prejudice because a self-defense theory had been presented to the jury and rejected, and, because at least six witnesses had testified that Mr. Carter, Jones, or Stone fired at Gardner, which would support the State’s accountability theory. The district court denied relief and denied a certificate of appealability.
We issued a certificate of appealability, limited to the ineffective assistance of counsel claim. We also recruited counsel and specifically directed briefing on trial counsel’s failure to call McReynolds and Cálmese.
II
DISCUSSION
Mr. Carter contends that he was denied his Sixth Amendment right to counsel. Specifically, he argues that his trial counsel rendered ineffective assistance by failing to investigate and to calí McReynolds and Cálmese. He also contends that the Illinois Appellate Court applied an incorrect legal framework to this claim by requiring him to demonstrate that counsel’s alleged errors rendered his trial “unreliable or fundamentally unfair.”- The State counters that the state courts reasonably and appropriately applied the Strickland standard. In any event, the State adds, Mr. Carter’s claims fail on de novo review. It argues that Mr. Carter waived the portion of his claim relating to a failure to investigate McReynolds and Cálmese, rather than a failure* to call. Finally, it contends that there is no reasonablé probability that the outcome of Mr. Carter’s trial would have been different with their testimony. We address these issues in turn.
[940]*940A.
Our standards of review in this context are complex but familiar. We review the district court’s decision denying habeas relief de novo. Smith v. Gaetz, 565 F.3d 346, 351 (7th Cir.2009). Under the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), if a claim “was adjudicated on the merits in State court proceedings,”, federal review of the conviction is highly, circumscribed. 28 U.S.C. § 2254(d). Focusing on the decision of the last state court to address a given claim on the merits, Ylst v. Nunnemaker, 501 U.S. 797, 803, 111 S.Ct. 2590, 115 L.Ed.2d 706 (1991), we ask only whether the state adjudication “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States” or “resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding,” 28 U.S.C. § 2254(d)(1) — (2).
“Under the ‘contrary to’. clause, a federal habeas court may grant the writ if the state court arrives at a conclusion opposite to that reached by [the Supreme] Court on a question of law or if the state court decides a case differently than [the Supreme] Court has on a set of materially indistinguishable' facts.” Williams v. Taylor, 529 U.S. 362, 412-13, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000); see also Bell v. Cone, 535 U.S. 685, 694, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002) (“A federal habeas court may issue the writ under the ‘contrary to’ clause if the state court applies a rule different from the governing law set forth in [Supreme Court] cases____”).
When, by contrast, the state court has articulated properly the governing legal standard, a petitioner still may succeed by showing that the State’s application of that standard was “unreasonable.” Williams; 529 U.S. at 411, 120 S.Ct. 1495. Under this standard,
We may not issue a writ “simply because [we] .conclude^] ... that the relevant state-court decision applied clearly established federal law erroneously or incorrectly. ' Rather, that application must also be 'unreasonable.” Williams v. Taylor, 529 U.S. 362, 411 [120 S.Ct. 1495] (2000); Rastafari v. Anderson, 278 F.3d 673, 688 (7th. Cir.2002). This demanding standard allows us to issue a writ only in cases “where there is no possibility fair-minded jurists could disagree that the state court’s decision conflicts with .[Supreme Court] precedents. It goes no farther.” Harrington [v. Richter, 562 U.S. 86, 102, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011) ].
Carter v. Butts, 760 F.3d 631, 635 (7th Cir.2014) (alterations in original) (parallel citations omitted).
With these rules in mind, we now turn to Mr. Carter’s arguments on the merits.
Mr. Carter now raises a single claim of ineffective assistance based' on his attorney’s failure to investigate -and call McRey-nolds and Cálmese to testify at his trial. In his view, both possible witnesses offered testimony in his favor that was' unique in substance, quality, or source, i.e., because it came from a witness without a previous tie to Mr. Carter or his associates. In the case of McReynolds, the proffered testimony both came from a totally disinterested witness and undermined the State’s case. And unlike the other witnesses who gave testimony favorable to Mr. Carter, McRey-nolds was not vulnerable to impeachment with prior inconsistent statements to law enforcement. In the case of Cálmese, he was connected to Gardner, and would have been the only witness so situated who [941]*941could have corroborated unequivocally Mr. Carter’s claims that he was not a shooter.18
The framework of our analysis of Mr. Carter’s 'Sixth Amendment claim is Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052 (1984). Under Strickland’s familiar two-prong test, we begin with the issue of deficiency, i.e., whether “counsel’s representation fell below an objective standard of reasonableness,” id. at 688, 104 S.Ct. 2052, and then consider the issue' of prejudice, i.e., whether “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceéding would have been different,” id. at 694,104 S.Ct. 2052.
1.
As we already have noted, the state court did not address the deficiency prong of the Strickland analysis. Accordingly, “we must ‘dispose of the matter as law and justice require,’ 28 U.S.C. § 2243, which is essentially de novo review.” Eichwedel v. Chandler, 696 F.3d 660, 671 (7th Cir.2012).19
We begin with a preliminary issue concerning Mr. Carter’s precise contentions on deficient performance; In the current briefing, Mr. Carter repeatedly frames his claim as a claim that his trial counsel’s deficiency was a failure to investigate and call the proffered witnesses. In his petition in the district court, • however, Mr. Carter principally argued that the failure to call the witnesses was counsel’s deficient performance. The State seizes on the discrepancy and contends that' Mr. Carter has waived any claims relating to counsel’s failure to investigate these witnesses.20
We are not persuaded that Mr. Carter’s marginally different characterizations of his claim are consequential. In the first place, Mr. Carter’s pro se petition in the district court must be construed liberally. Bennett v. Gaetz, 592 F.3d 786, 790 (7th Cir.2010); cf. Ward v. Jenkins, 613 F.3d 692, 696-97 (7th Cir.2010) (employing a liberal construction to a pro se state petition to determine if claims were fairly presented to state courts). More importantly, however, a review of our substantive standards for evaluating Mr. Carter’s failure-to-call claims demonstrates that we have regarded such claims as closely tied to what the record tells us about the nature of counsel’s investigation. Specifically, in applying Strickland’s first prong generally, we have stated that
we presume that counsel’s actions fall within the wide range of reasonable professional assistance, and defer to strategic decision-making by a trial attorney. Despite this weighty deference, we nonetheless must carefully cohsider whether the attorney brought to bear the skill and knowledge that allows for a proper adversarial testing process, considering all the circumstances.
[942]*942Adams v. Bertrand, 453 F.3d 428, 434-35 (7th Cir.2006) (emphasis added) (citations omitted) (internal quotation marks omitted). In elucidating this standard in the context of a claim of failure to present certain potentially favorable testimony, we haye stated:
“[A] lawyer’s decision to call or not to call a witness is a strategic decision generally, not subject to review. The Constitution does not oblige counsel to present each and every witness that is suggested to him.” United States v. Williams, 106 F.3d 1362, 1367 (7th Cir.1997) (internal citation and quotation marks omitted). If counsel has investigated witnesses and consciously decided not to call them, the decision is probably strategic. An outright failure to investigate witnesses, however, is more likely to be a sign of deficient performance ....
____, New decisions not to present testimony can be considered “strategic” before some investigation has taken place. As we explained in United States ex rel. Hampton v. Leibach, 347 F.3d 219 (7th Cir.2003), “strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation.”
United States v. Best, 426 F.3d 937, 945-46 (7th Cir.2005) (emphasis in original); see also Mosley v. Atchison, 689 F.3d 838, 848 (7th Cir.2012) (“To avoid the inevitable temptation to evaluate a lawyer’s performance through the distorting lens of hindsight, Strickland establishes a deferential presumption that strategic judgments made by defense counsel are reasonable. But the presumption applies only if the lawyer actually exercised judgment.” (citation omitted)). That is, al-’ though we defer to strategic decisions, we first assure ourselves that.a strategic decision was made, because “[t]he consequences of inattention rather than reasoned strategic decisions are not entitled to the presumption of reasonableness.” Mosley, 689 F.3d at 848. In sum, although “[i]t would be a rare case where counsel’s conscious decision not to call a witness would amount to, constitutionally ineffective assistance,” United States v. Weaver, 882 F.2d 1128, 1139 (7th Cir.1989) (emphasis added), “strategic choices made after less than complete investigation are reasonable only to the extent that reasonable professional judgments support the limitations on investigation.” Wiggins v. Smith, 539 U.S. 510, 533, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003) (internal quotation marks omitted); see also Strickland, 466 U.S. at 691, 104 S.Ct. 2052 (“In other words, counsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigation's unnecessary,”).
The record does not .reveal anything about the scope of counsel’s investigation of Cálmese and McReynolds.- Counsel was aware of both . witnesses and initially placed McReynolds on the witness list and took steps to. secure his testimony.21 For reasons that are not disclosed by the record, counsel did not actually contact or call either potential witness. On the record before us, therefore, we cannot determine that counsel conducted the, investigation necessary to conclude that 'pursuing McReynolds’s or Calmese’s testimony would be fruitless.22
[943]*943Potential reasons'not-to call either witness are, of course, conceivable. McRey-nolds was incarcerated during the lead-up to Mr. Carter’s trial, and counsel may have concluded that his criminal history (the details of which are not part of the present record) were sufficiently problematic that his testimony would have been of little value. Counsel may have believed that the testimony of either witness would have been cumulative and unhelpful, a point we shall examine in our prejudice analysis. On the other hand, we have found deficiency in situations where counsel has failed to call witnesses even when testimony is cumulative, if the missing witness is disinterested in a case in which other witnesses have a relationship, to the defendant. See Montgomery v. Petersen, 846 F.2d 407, 414 (7th Cir.1988) (“The jury was presented with a straightforward credibility choice. Every one of these witnesses had a reason to be biased. Given the standoff between two factions in this family, one group supporting Wayne Montgomery and the other group supporting the petitioner, independent corroboration by a neutral, disinterested witness would perforce be extremely significant.”).
Under these circumstances and considering the potential value of the testimony of Cálmese and McReynolds — the merits of which we shortly shall examine — it may well have been that “counsel could not have made a reasonable strategic decision not to call [either witness] without interviewing [them] in order to evaluate [their] proposed testimony, [their] credibility or [their] demeanor.” Toliver, 539 F.3d at 775. Based on the allegations contained in the affidavit of McReynolds,. which we must take as- true at this stage of the proceeding, such an investigation did not occur. Remand therefore, might well be appropriate to address the first Strickland prong, unless the claim was properly denied on the prejudice prong. Cf. id at 782 (“The state courts never resolved, under the first prong of the Strickland analysis, whether Mr. Toliver’s counsel was ineffective in not interviewing Harvey and in not calling Angeal. Therefore, issues of fact Concerning counsel’s competence were never resolved.”).23 We therefore proceed to tlie prejudice prong of Strickland.
[944]*9442.
The parties dispute the appropriate standard of review on the issue of prejudice. The Illinois Appellate Court discussed and resolved this issue on the merits against Mr. Carter. Ordinarily, this state court determination would require us to apply AEDPA deference to the state court’s decision. See 28 U.S.C. § 2254(d)(2); supra section II.A. Mr. Carter, nevertheless, contends that our review of the prejudice prong is also de novo, because the state court analyzed the case under People v. Coulter, 352 Ill.App.3d 151, 287 Ill.Dec. 255, 815 N.E.2d 899 (2004), a case which stated that outcome-determinative prejudice is insufficient unless counsel’s performance resulted in a trial that was unreliable dr fundamentally unfair.
At first blush, Mr. Carter’s point is well taken, and one that we have noted before. In Strickland, the Supreme Court identified the now familiar prejudice standard as whether “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” 466 U.S. at 694, 104 S.Ct. 2052. Mr. Carter is correct that the State court recited a standard that placed an additional burden on him, specifically, one that required him to demonstrate not only a reasonable probability of a different outcome, but that the result of counsel’s errors rendered his trial unreliable or fundamentally unfair. The state court’s actual decision rested, however, on his failure to satisfy even the lower outcome burden under Strickland. Put simply, although the state court’s decision is bookended by an articulation of the fundamentally-unfair-or-unreliable standard, its analysis is focused on whether the proffered testimony could have affected the outcome, which is the correct inquiry under Strickland. See R.14-2 at 169-70 (noting that the testimony “would not have been exculpatory” and was “cumulative,” that McReynolds’ testimony did not accomplish what Mr. Carter alleged that it did, and that, because of the State’s accountability theory, “the result of defendant’s trial would have been the same”).
We faced a similar situation in Floyd v. Hanks, 364 F.3d 847 (7th Cir.2004). There, the Court of Appeals of Indiana examined prejudice under Strickland but recited that “when errors do not make the result of the trial unreliable, they do not cause prejudice.” Id. at 852. Nevertheless, we noted that
a fuller view of the appellate court’s discussion reveals that while the term “reliability” was employed, the actual analysis of Floyd’s counsel’s conduct properly considered whether the coun[945]*945sel’s actions affected the outcome of the trial. As noted above, the Indiana Appellate Court considered the potential effect of the [favorable evidence] against the weight of the other evidence heard by the jury. In reaching its decision that there was no prejudice, the'court found that the inculpating evidence was overwhelming and had Floyd’s 'counsel taken the steps that Floyd now demands, the result would have been the same; this is the very analysis that is required by Strickland and Williams.
Id. at 852-53 (footnote omitted); cf. Goodman v. Bertrand, 467 F.3d 1022, 1028 (7th Cir.2006) (finding state court applied an incorrect prejudice analysis where “it repeatedly reasoned that Goodman failed to show that his second trial was ‘fundamentally unfair’ or ‘unreliable’ ”).
The analysis in Floyd applies here. Even though the Illinois court noted that counsel’s performance “did not render the proceeding unreliable or fundamentally unfair,”24 its analysis focused on the probable impact of the proffered testimony on the outcome. The court evaluated the evidence piece by piece and concluded explicitly that, because the jury was instructed on an accountability theory,'“the result of defendant’s trial would have been the same even if Cálmese and McReynolds had testified that defendant was unarmed.”25 “Any possible testimony from McReynolds and Cálmese that defendant was unarmed would not have illuminated whether defendant was legally accountable for the actions of his co-defendants.”26 This is precisely the analysis' demanded by Strickland. Accordingly, the Illinois court’s analysis is not “contrary to” settled law, and we therefore apply AEDPA deference to the state court’s resolution of the issue.
We next turn to the question whether the State unreasonably applied Strickland on the facts before it. See Cullen v. Pinholster, 563 U.S. 170, 181, 131 S.Ct. 1388, 179 L.Ed.2d 557 (2011) (“[Rjeview under § 2254(d)(1) is limited to-the record that was before the state court that adjudicated the claim on the merits.”). Specifically, we must determine whether the state court’s assessment of the probable value of the proffered testimony was unreasonable.27
We begin with Cálmese. If he had testified consistently with his statement to po[946]*946lice, Cálmese would have stated that Stone fired first, and then Jones fired.28, He made • no statement about Mr. Carter’s possession or use of a weapon. It is true that, unlike- the testimony of Rena and Antonio Phillips, Calmese’s testimony would not have inculpated directly Mr. Carter. And unlike Pearson (who indicated that Stone shot and either Jones or Mr. Carter followed) and Gaston (who identified Jones as a shooter and was unclear if anyone else shot), Calmese’s two-shooter scenario involves only Stone and Jones. It is perhaps most consistent with Cheeks’s testimony that Stone shot and -then additional shots came from the direction of Mr. Carter and Jones..
According to his affidavit, McReynolds would have testified that he heard the argument, “observed Friday Gardner pull an object out from behind his back and . then ... heard several shots ring out.”29- He continues that “[t]he individual [947]*947firing the shots” was Stone.30. Further, McReynolds states: “I did not personally observe anyone else doing any shooting, and when the shooting was occurring the reaction of Junior (Michael Carter) and Cortez Jones was to scatter in an effort to avoid being shot.”31 If he had been called to testify, therefore, McReynolds ■ would have echoed the testimony of Felicia and Michella Anderson, Cheeks, and Stone with respect to Gardner’s drawing a weapon, and with Felicia and Michella Anderson, and. Stone, in identifying Stone as the only shooter- .
In assessing the testimony of Cálmese and McReynolds, the state court concluded that it would have had little probable impact on the outcome of Mr. Carter’s trial. Specifically, it said that the testimony
would not have been exculpatory and would have merely been cumulative of the testimony presented by Felicia Anderson, LaTonya Cheeks, Michell[a] Anderson, and co-defendant Stone. Defendant’s theory of defense was presented at trial and corroborated where these witnesses testified that .the victim had a gun and defendant did not. The jury rejected this defense.
Moreover, McReynolds’ affidavit is. insufficient and unsupportive of defendant’s defense. Specifically, defendant incorrectly relies on the affidavit to establish that he was unarmed and did not participate in shooting the victim. McReynold[s’] affidavit merely states that he only saw co-defendant Stone shoot a gun, and that defendant “scattered” to avoid being shot. He does not specify that defendant was unarmed or somehow uninvolved in the shooting. The same holds true for the police report regarding Cálmese, which merely documents that the police interviewed Cálmese, who saw .co-defendants Stone and Jones shoot the victim. It does not state that defendant was unarmed or somehow uninvolved in the shooting.
Furthermore, the result of defendant’s trial would have been the same even if Cálmese and McReynolds had testified that defendant was unarmed where the jury was given the following accountability instruction:
“A person is legally responsible for the conduct of another person when either before or during the commission of an offense with the intent to promote or facilitate the commission of an offense he knowingly solicits, aids, abets, agrees to aid, or attempts to aid the other person in the. planning or commission of the offense. [ ] A person who is legally responsible for the conduct of another may be convicted for the offense committed by the other person even though the other person, who it is claimed committed the offense, has not been convicted.”
The jury heard the evidence, received the instructions, and found defendant guilty. Any possible testimony from McReynolds and Cálmese that defendant was unarmed would not have illuminated whether defendant was legally [948]*948accountable for the actions of his co-defendants.[32]
It concluded its analysis with a final reference to its determination on direct appeal “that the evidence here was not closely balanced.”33
The court’s analysis, while flawed, is ultimately not so far off the mark as to be unreasonable. See 28 U.S.C. § 2254(d)(1); Hardaway v. Young, 302 F.3d 757, 762 (7th Cir.2002) (“A state court decision must be more than incorrect from the point of view of the federal court; AEDPA requires that it be ‘unreasonable,’ which means something like lying well outside the boundaries of permissible differences of opinion.”). We are uncomfortable with the state court’s conclusion that the testimony is “cumulative” and that the theory of defense was adequately presented at trial and rejected by the jury. In reaching this conclusion and noting specifically whose testimony McReynolds and Cálmese would echo, the state court makes no mention of the fact that all of the witnesses it is referencing are witnesses with a preexisting relationship to Mr. Carter and therefore have a potential bias that would have been clear to the jury. By contrast, Cálmese was a witness with a relationship to Gardner, and McReynolds appears to be totally disinterested. This is a significant and noteworthy difference that deserved to factor in to the state court’s analysis, Cf. Montgomery, 846 F.2d at 414 (“Every one of these witnesses had a reason to be biased. Given the standoff between two factions ... independent corroboration by a neutral, disinterested witness would perforce be extremely significant.”).
Nevertheless, the remainder of- the state court’s analysis is sufficient to assure us that, despite this significant oversight, the conclusion that the new testimony did not create a substantial probability of acquittal is not unreasonable. In reaching this conclusion, the most significant factors are the competing theories of the case presented at trial: Stone argued self-defense, Mr. Carter argued mere presence, arid the State argued accountability. Calmese’s testimony is not valuable under these circumstances, because it identifies not just Stone but also Jones as shooters, undermining the sole-shooter scenario Stone had presented to the jury. Moreover, anything that confirms Jones’s involvement is factually even more problematic for Mr. Carter under the accountability theory, because Mr. Carter’s ties to Jones that day and at the moment of the shooting make a claim that Jones acted independently significantly less credible: Mr. Carter had gotten Jones involved in the dispute over the robbery; they spent the afternoon together; they drove to the building to begin the confrontation together; they fled together; and, outside of backing up Mr. Carter, Jones did not have a proverbial “dog in the fight” stemming from the original burglary or-the people involved. Accordingly, to truly undercut the State’s case, Mr. Carter’s defense needed to distance him from the shooters, whether that be Stone alone or Jones and Stone. Calmese’s testimony does not accomplish that result.
McReynolds’s proffered testimony, for the reasons noted by the state court, also ultimately fails to furnish the linchpin of Mr. Carter’s claim. Had the proffer in-[949]*949eluded an unequivocal statement from McReynolds that he had watched the entire scene, that Stone was the only shooter, and that Mr. Carter and Jones were unarmed, we might well conclude, given McReynolds’s disinterested status, that the. state court’s conclusion on. prejudice was unreasonable. But as the State notes, McReynolds says nothing about Mr. Carter’s activities .prior , to “scattering].”34 He does not say that he watched Mr. Carter and saw no gun and no shooting; he says only that he “heard several shots ring out from the alleyway,” Stone’s location, and “did not personally observe” any other shooting.35 McReynolds does not specifically say that he saw Stone shoot. The phraseology is notably weak, and all of the potentially relevant factual assertions about the shooting comprise only three sentences, woefully lacking in detail. His further testimony that Mr. Carter and Jones scattered after Stone’s shots to avoid being shot themselves is fairly characterized as adding somewhat of a gloss on the testimony of essentially every other witness that day, all of whom agree that all three of the defendants fled the scene. The most striking factual claim McRey-nolds makes in the affidavit is that he saw Gardner draw “an object out from behind his back,” but this fact, while potentially significant to Stone’s self-defense claims, is irrelevant to Mr. Carter’s mere presence defense.36
The state court indicated that this case was not a close one. That characterization •may be a matter of some legitimate debate. The issue before us today is, however, a narrow one concerning only two pieces of additional evidence. Considering the limited potential value, of that evidence, we conclude that the .state court’s assessment of the prejudice prong of Strickland cannot be characterized as unreasonable.37
Conclusion
The district court’s resolution of the sole claim presented in,. Mr. Carter’s federal [950]*950habeas petition was correct. Although trial counsel’s performance may have been deficient in failing to investigate potential witnesses,, the state court’s resolution ■ of the prejudice analysis was not unreasonable within the meaning of 28 U.S.C. § 2254(d)(1). Accordingly, we affirm the judgment of the district court.
AFFIRMED