OPINION

COLE, Circuit Judge.

Demetrius Foster was convicted of second-degree murder in connection with the shooting death of Bobby Morris. After trial, the Michigan trial court conducted a Ginther hearing and concluded that Foster's trial counsel was ineffective for failing to raise an alibi defense, stemming primarily from the testimony of a witness who said that Foster was with him at the time of the killing. The Michigan Court of Appeals, however, concluded that the failure to raise the alibi defense was an exercise of sound trial strategy by Foster's trial counsel. On habeas review, the district court held that Foster's trial attorney was deficient for failing to raise the alibi defense, but that the deficiency did not prejudice Foster. We AFFIRM the district court with respect to the deficient performance finding, REVERSE with respect to the prejudice determination, and GRANT a conditional writ of habeas corpus giving the State of Michigan 180 days to retry Foster or, failing that, to release him.

I. BACKGROUND

Around 11 p.m. on the evening of November 16, 1998, Foster went to the home of Monique Brassell in Detroit, Michigan. People v. Foster, No. 226311, 2003 WL 1558238, at *1 (Mich.Ct.App. Mar. 25, 2003) (unpublished). Brassell testified that she had never met Foster prior to that evening. Id. At approximately 1 a.m. on November 17, Brassell and Foster decided to have sex in an upstairs room. An hour later, Bobby Morris arrived at the Brassell home, accompanied by three friends, Deborah Hollins, Javan Carter, and Jerry Wallace. Id. Morris was the former long-term boyfriend of Brassell and father of Brassell's children. Morris began banging on the outside door, and pushed past Brassell when she opened it. Morris assaulted Brassell, and, after leaving the Brassell home, smashed the back window of Foster's late-model Ford Bronco truck with a sledgehammer. Id. Brassell later testified that Foster was angry over the damaged window. Brassell also testified that Foster overheard her telling police officers where Morris lived. Id.

After leaving the Brassell home, Hollins began driving Morris, Carter, and Wallace home. As Hollins drove down Flanders Street to Morris's home at approximately 2:45 a.m., she observed an individual walking alongside the car on the sidewalk, wearing a green jacket. Id. When Hollins stopped the car in front of Morris's home, the man approached the car on the driver's side. The man had a gun, and Hollins identified him at trial as Foster.1 Id. The assailant fired multiple shots at Morris as he tried to flee the car. Hollins, Carter, and Wallace fled the scene. Police officers later found Morris, unconscious, with at least one gunshot wound. Morris was pronounced dead upon arrival at the hospital. Id.

Foster was charged with first-degree murder. Foster's primary defense at trial was mistaken identification by Hollins. Prior to trial, Arthur Daniels sent a letter to Foster's trial counsel, claiming that Foster was at his home from 2:15 to 6 a.m. on the day of the shooting. Foster's trial counsel eventually spoke to Daniels over the telephone for roughly fifteen to twenty minutes. Daniels was not called as a witness during trial. After initially deadlocking, the jury eventually returned a guilty verdict on the lesser included charge of second-degree murder.

On appeal, Foster requested a remand to the trial court, pursuant to People v. Ginther, 390 Mich. 436, 212 N.W.2d 922 (1973), for consideration of whether his trial counsel was ineffective for failing to call Daniels as a witness. The Michigan Court of Appeals granted the request for a Ginther hearing. At the hearing, the trial court took the testimony of Daniels and Foster's trial counsel. The court concluded that Foster's counsel was deficient for failing to raise an alibi defense, that this deficiency prejudiced Foster, and that Foster should have a new trial. The Michigan Court of Appeals vacated the new trial order on procedural grounds, and ruled that the decision not to call Daniels was a function of sound trial strategy on the part of Foster's counsel, and thus Foster's counsel was not ineffective. Foster, 2003 WL 1558238, at *6-7. The Michigan Supreme Court denied leave to appeal. People v. Foster, 471 Mich. 895, 688 N.W.2d 83 (2004) (table decision). Foster unsuccessfully sought post-conviction relief under Michigan law. See People v. Foster, No. 264647 (Mich.Ct.App. Mar. 23, 2006) (unpublished); People v. Foster, 476 Mich. 865, 720 N.W.2d 281 (2006) (table decision).

Foster filed his amended petition for habeas corpus in federal court on September 21, 2006, raising a number of claims including ineffective assistance of trial counsel for the failure to mount an alibi defense. On March 3, 2009, a magistrate judge issued a Report and Recommendation, recommending denial of the petition on all counts. The district court modified the magistrate judge's Report, holding that Foster's trial counsel was deficient in failing to mount an alibi defense, but that this deficiency did not prejudice Foster in light of the "substantial" evidence of his guilt. Therefore, the district court denied the petition, but issued a Certificate of Appealability on the ineffective assistance of counsel question. This appeal followed.

II. ANALYSIS

A. Standard of Review

"In a habeas proceeding, we review de novo the district court's legal conclusions, including its ultimate decision to grant or deny the writ, and we review for clear error its factual findings." Satterlee v. Wolfenbarger, 453 F.3d 362, 365 (6th Cir.2006). A factual finding by the district court "is clearly erroneous when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed." Id. at 366 (quoting Norris v. Schotten, 146 F.3d 314, 323 (6th Cir.1998)) (internal quotation marks omitted).

However, as Foster's conviction became final after April 24, 1996, his petition is governed by the Antiterrorism and Effective Death Penalty Act of 1996 ("AEDPA"):

Murphy v. Ohio, 551 F.3d 485, 493 (6th Cir.2009) (quoting 28 U.S.C. § 2254(d)(1)-(2)). "A state-court decision is contrary to clearly established federal law if the state court applies a rule that contradicts the governing law set forth in [the Supreme Court's] cases or if the state court confronts a set of facts that are materially indistinguishable from a decision of [the Supreme] Court and nevertheless arrives at a result different from [that] precedent." Id. at 493-94 (quoting Williams v. Taylor, 529 U.S. 362, 405, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000)) (alterations in original) (internal quotation marks omitted). An unreasonable application of federal law is one where the state court either "correctly identifies the governing legal rule but applies it unreasonably to the facts of a particular prisoner's case," id. at 494 (quoting Williams, 529 U.S. at 407-08, 120 S.Ct. 1495), or "either unreasonably extends or unreasonably refuses to extend a legal principle from Supreme Court precedent to a new context," id. (quoting Seymour v. Walker, 224 F.3d 542, 549 (6th Cir.2000)).

The only issue before us is whether Foster's counsel was ineffective for failing to raise an alibi defense during trial. For ineffective assistance of counsel claims, the Supreme Court explained in Strickland v. Washington, "[t]he benchmark for judging any claim of ineffectiveness must be whether counsel's conduct so undermined the proper functioning of the adversarial process that the trial cannot be relied on as having produced a just result." 466 U.S. 668, 686, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984); see also Matthews v. Parker, 651 F.3d 489, 507 (6th Cir.2011) (quoting Strickland). The Matthews court stated:

Id. at 507-08 (internal citations and quotation marks omitted). Thus, a petitioner seeking a writ of habeas corpus based on ineffective assistance of trial counsel must show: (1) that trial counsel's performance was deficient to a constitutionally significant degree; and (2) that the deficient performance had an impact on the outcome of the petitioner's trial. Strickland determinations are reviewed de novo, while any underlying factual findings are reviewed for clear error. See Lovell v. Duffey, 629 F.3d 587, 596 (6th Cir.2011) ("Although at first blush [the ultimate] conclusion might seem purely factual, it actually involves an application of Strickland to the facts of this case, which we review de novo."), vacated on other grounds by Cullen v. Pinholster, ___ U.S. ___, 131 S.Ct. 1388, 179 L.Ed.2d 557 (2011).

B. Deficient Performance

Following a Ginther hearing, the trial court held that trial counsel was deficient for failing to investigate and present an alibi defense. The Michigan Court of Appeals reversed, holding that the decision not to put on an alibi defense was a strategic decision on the part of Foster's trial counsel. The Michigan Court of Appeals relied on testimony from trial counsel during the Ginther hearing, where trial counsel stated that he decided not to pursue the alibi defense based on his interview of Daniels. Trial counsel testified that he found Daniels's testimony to be "vague," and was concerned that Daniels would be a poor witness.

The district court agreed with the trial court and found that Foster's trial counsel was deficient in failing to raise an alibi defense. The district court concluded that counsel's ineffectiveness turned not on whether trial counsel's conclusion regarding Daniels was a reasonable strategic judgment, but rather on whether it was reasonable not to further investigate possible support for an alibi defense in light of Daniels's testimony. Foster v. Wolfenbarger, No. 04-73794, 2010 WL 3341534, at *1 (E.D.Mich. Aug. 24, 2010) (unpublished). "If trial counsel truly believed that the witness's testimony sounded `vague,' he should have done more to investigate the alibi." Id. Since it is undisputed that trial counsel did not perform any additional investigation beyond briefly interviewing Daniels, the district court concluded that trial counsel was deficient:

Id.

"The duty to investigate derives from counsel's basic function, which is to make the adversarial testing process work in the particular case." Towns v. Smith, 395 F.3d 251, 258 (6th Cir.2005) (quoting Kimmelman v. Morrison, 477 U.S. 365, 384, 106 S.Ct. 2574, 91 L.Ed.2d 305 (1986)) (internal quotation marks omitted). "This duty includes the obligation to investigate all witnesses who may have information concerning his or her client's guilt or innocence." Id. at 258. "A purportedly strategic decision is not objectively reasonable when the attorney has failed to investigate his options and make a reasonable choice between them." Id. (internal quotation marks omitted). In fact, "strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation." Wiggins v. Smith, 539 U.S. 510, 521, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003) (quoting Strickland, 466 U.S. at 690-91, 104 S.Ct. 2052).

Here, trial counsel's only investigation with regard to the alibi defense was to question Daniels, over the telephone, for approximately fifteen to twenty minutes. Even crediting trial counsel's conclusion that Daniels would make a poor witness, that does not represent a full investigation of the alibi defense under the circumstances. While Daniels might have been a poor witness, trial counsel did nothing to investigate the other two individuals at the home with Daniels who might have information concerning Foster's alibi theory.2 Either of these two individuals, or all three of the witnesses collectively, might have overcome any deficiencies in Daniels's testimony and demeanor. Trial counsel did nothing to explore these possibilities.

Moreover, the alibi defense was completely consistent with, and in fact complimentary to, trial counsel's theory of mistaken identification. The fact that one or more individuals would place Foster in a different location at the time of the shooting supports the premise that Hollins misidentified Foster. In fact, trial counsel admitted this during the Ginther hearing. Thus, this is not a situation where defense counsel is required to choose between two different avenues and pick the theory of the case that counsel believed was stronger. Instead, trial counsel chose not to investigate an avenue that potentially could have bolstered the defense that counsel was already pursuing.

Finally, we note that the trial court concluded that Foster's counsel was deficient. We give due deference to the conclusions of the trial judge on the effectiveness of counsel, because "[t]he judge, having observed the earlier trial, should have an advantageous perspective for determining the effectiveness of counsel's conduct and whether any deficiencies were prejudicial." Massaro v. United States, 538 U.S. 500, 506, 123 S.Ct. 1690, 155 L.Ed.2d 714 (2003). The trial judge here heard the testimony of both Foster's counsel and Daniels, and based on that testimony found that Foster's counsel should have pursued the alibi theory. While the trial court's ruling is not a dispositive basis for granting relief, it provides support for the district court's rationale and our own conclusion. See Leonard v. Michigan, 256 F.Supp.2d 723, 732 (W.D.Mich. 2003) (giving deference to the state trial court's findings on trial counsel's deficiencies and deeming the state appellate court "unreasonable" for not doing the same).

The requirement that trial counsel fully investigate a possible defense is clearly established Supreme Court precedent under Strickland and its progeny. The evidence also shows that trial counsel did not properly investigate Foster's alibi defense. These two facts make the Michigan Court of Appeals's decision an "an unreasonable application of ... clearly established Federal law, as determined by the Supreme Court of the United States" as required under 28 U.S.C. § 2254(d)(1).

C. Prejudice

Having concluded that Foster's trial counsel's performance was deficient, we must now consider whether this deficiency was prejudicial to Foster. At the Ginther hearing, the trial court held that Foster was prejudiced upon finding that "there probably could have been a different outcome." Likening the alibi testimony to Barry Bonds's (at the time) recent historic home run performance, the court stated: "it would put this one in the seats and possibly get the record." The Michigan Court of Appeals did not reach the question of prejudice, relying on its conclusion that there was no deficiency. We have recently reiterated that, where the state court ruled on one prong of the Strickland test, but not the other, we apply a de novo standard of review to the prong that the state court did not consider. See Rayner v. Mills, 685 F.3d 631, 636-39 (6th Cir.2012) (applying de novo standard of review stated in Wiggins v. Smith, 539 U.S. 510, 534, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003), and distinguishing Harrington v. Richter, ___ U.S.___, 131 S.Ct. 770, 784, 178 L.Ed.2d 624 (2011)). When evaluating prejudice to the defendant, a reviewing court must "[take] into account the totality of the circumstances, as well as the relative strength of the case proffered by the prosecution...." Campbell v. Coyle, 260 F.3d 531, 551 (6th Cir.2001).

The district court characterized the evidence against Foster as "substantial." This conclusion is belied by the trial record. Hollins is the only witness who could identify Foster at the scene of the crime. Hollins told the police that the assailant was 5'6" or 5'7" and weighed 180 pounds, while Foster is 6'0" and approximately 240 pounds. In addition, while Hollins testified at trial that the assailant was wearing a green jacket of the same type as the man she saw at Brassell's home, she was successfully impeached with her preliminary hearing testimony that the assailant was wearing a blue jacket. By any standard this is a weak identification. Moreover, Hollins's identification is the only direct link of Foster to the crime — there was no forensic evidence recovered at the scene. In that light, there is a reasonable probability that, had defense counsel further investigated and called Daniels or the other people with him, the jury would have found Foster not guilty.

Moreover, as with the deficiency inquiry, we must give consideration to the trial judge's conclusions made at the Ginther hearing. The trial court concluded that an alibi witness would have buttressed the other arguments made by trial counsel, and could have potentially tipped the jury to vote for acquittal.3 The trial judge also noted that the jury initially deadlocked and reached their verdict only after further instructions from the trial judge. In other words, the jury's verdict rested on a narrow ground, and additional exculpatory evidence, such as the alibi defense, had a strong likelihood of tipping the scales in the other direction.

Contrary to the district court's conclusion, the evidence of Foster's guilt was not "substantial." Instead, we find the relative weakness in the state's case and strength of the potential alibi defense creates "a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." Matthews, 651 F.3d at 508 (citation omitted). In light of these facts, we conclude that Foster was prejudiced by the failure of his trial counsel to investigate and raise an alibi defense. Because Foster has met both prongs of the Strickland test, we conclude that his trial counsel provided ineffective assistance when he failed to investigate fully an alibi defense at trial, and as such is entitled to a conditional writ of habeas corpus.

III. CONCLUSION

For the foregoing reasons, we AFFIRM the district court's determination that Foster's trial counsel was deficient in failing to raise an alibi defense, REVERSE the district court's conclusion that such deficiency did not prejudice Foster, and GRANT a conditional writ of habeas corpus giving the State of Michigan 180 days to retry Foster or, failing that, to release him from custody.