[5]*5OPINION
DIETZEN, Justice.
Appellant Kenneth Eugene Andersen was found guilty by a Becker County jury of first-degree premeditated murder, Minn.Stat. § 609.185(a)(1) (2012), arising out of the shooting death of Chad Swed-berg. The district court entered a judgment of conviction, and sentence was imposed. On direct appeal, we affirmed Andersen’s conviction. State v. Andersen, 784 N.W.2d 320, 323 (Minn.2010). After his direct appeal, Andersen filed a petition for postconviction relief alleging seven separate grounds that he contends warrant a new trial. The postconviction court summarily denied Andersen’s petition, explaining that Andersen’s claims are either meritless on their face or barred by State v. Knaffla, 309 Minn. 246, 252, 243 N.W.2d 737, 741 (1976). In this appeal, Andersen argues the postcon-viction court erred in denying his petition for postconviction relief. Because we conclude the postconviction court did not err in summarily denying Andersen’s petition for postconviction relief, we affirm.
A grand jury indicted Andersen of first-degree premeditated murder arising out of the April 13, 2007 shooting death of Chad Swedberg.1 Andersen demanded a jury trial. While awaiting trial, Andersen was held at a local jail that monitored and recorded most phone calls, including some of Andersen’s calls with his attorneys and private investigator.
At trial, the State presented the following evidence. Chad Swedberg lived with, among others, his wife, Leslie Fain, and her son, Jesse Fain, in rural Becker County. On the morning of the murder, Swed-berg, his friend Albert Baker, and Jesse Fain planned to process maple syrup at a nearby syruping camp. Swedberg left his home at about 8:00 a.m. to get everything ready before the others arrived. Shortly after Swedberg left, Leslie Fain heard two gunshots from the direction of the camp. Since it was not hunting season she became concerned, and called Swedberg’s cell phone at 8:13 a.m. After trying to reach Swedberg several times without an answer, she walked to the camp. At the camp, she found Swedberg’s body and called 911.
During the investigation, police discovered that Andersen and Swedberg were friends and worked together, but the relationship had recently deteriorated. Based on interviews of Andersen and other witnesses, the investigation quickly focused on Andersen. The police determined that Andersen called Swedberg at 7:46 a.m. on the morning of the murder. Six minutes later, at 7:52 a.m., Andersen called Baker and asked him to stop by Andersen’s house on the way to the syruping camp, ensuring that Baker would not be at the syruping camp until later that morning. Moreover, Andersen was familiar with the area surrounding the camp, having previously visited the camp and hunted in the area.
Pursuant to a search warrant, police found bullets in Andersen’s house that had similar characteristics to the bullets recovered from Swedberg’s body. The police also found a Tikka T3 Lite .300 Winchester short magnum rifle concealed under the insulation of an outbuilding near Andersen’s house. Andersen attempted to con[6]*6ceal his ownership and possession of the Tikka rifle, but made no attempt to conceal his ownership or possession of other guns that did not fire .30-caliber bullets. The State’s palm-print expert, Patrick Warrick, testified that he had issued a report stating that a palm print identified as Andersen’s was found on the rifle. Warrick further explained that he had sent images of the rifle and print to a second examiner, and that pursuant to his office policy, if a second analyst disagrees with a palm-print identification, the result is deemed inconclusive.
The medical examiner testified that Swedberg was shot twice — once in the back of the right shoulder and once in the left buttock. As a result, he bled to death. Based on the physical evidence, the examiner concluded that Swedberg was not shot at close range. The State’s firearm expert, Nathaniel Pearlson, opined that the bullets recovered from Swedberg’s body were from a .30 caliber weapon and that he was “reasonably certain” the bullets were Winchester Supreme Ballistic Silvertip bullets. In reaching his opinion, the firearm expert relied on marketing material he received in an email from a Winchester representative, who did not testify at trial. Defense counsel explicitly declined to object to the introduction of the marketing material at trial.
The jury found Andersen guilty of first-degree premeditated murder. At the sentencing hearing, defense counsel moved for a mistrial, arguing that the State violated Minn. R.Crim. P. 9.01 by failing to timely disclose a police report in which Baker admitted to possessing a .30-caliber rifle. The district court denied the motion and sentenced Andersen to life in prison without the possibility of release.
In his direct appeal, Andersen argued: (1) there were misrepresentations in the search warrant application used to obtain his gun that invalidated the warrant; (2) there was insufficient evidence to support his conviction for first-degree premeditated murder; (3) his right to counsel was violated because calls to his attorney’s cell phone were monitored and recorded; and (4) the district court erroneously questioned jurors, bailiffs, and witnesses about incidents that occurred outside of the trial. We rejected all of his arguments and affirmed Andersen’s first-degree murder conviction. Andersen, 784 N.W.2d at 323.
In December 2010, Andersen filed the present petition for postconviction relief alleging that he is entitled to a new trial because: (1) there was newly discovered evidence that was both material and exculpatory; (2) recordings of phone calls he made while in jail were admitted at trial in violation of his constitutional right to counsel; (3) testimony by firearm and palm-print experts and an exhibit displaying several types of Winchester bullets were admitted at trial in violation of his Sixth Amendment right of confrontation; (4) the State committed misconduct during closing argument; (5) the State withheld certain evidence until after trial; (6) his trial counsel was ineffective; and (7) his appellate counsel was ineffective. The postconviction court summarily dismissed Andersen’s claims. Because Andersen’s claims are either meritless on their face or barred by the Knaffla rule, we affirm.
I.
Andersen first argues that the postconviction court erred in concluding that the newly-discovered-evidence claim in his petition lacked merit. We review a postconviction court’s legal conclusions de novo. Davis v. State, 784 N.W.2d 387, 390 (Minn.2010). But we will not reverse the court’s factual findings unless they are clearly erroneous. See Doppler v. State, 771 N.W.2d 867, 875 (Minn.2009).
[7]*7A person convicted of a crime who claims that his or her conviction was obtained in violation of the person’s constitutional or statutory rights may file a petition for postconviction relief under Minn. Stat. § 590.01, subd. 1(1) (2012). The postconviction court must hold an eviden-tiary hearing “[u]nless the petition and the files and records of the proceeding conclusively show that the petitioner is entitled to no relief.” Minn.Stat. § 590.04, subd. 1 (2012).
To obtain a new trial on the basis of newly discovered evidence, a petitioner must establish: “(1) that the evidence was not known to the defendant or his/her counsel at the time of the trial; (2) that the evidence could not have been discovered through due diligence before trial; (3) that the evidence is not cumulative, impeaching, or doubtful; and (4) that the evidence would probably produce an acquittal or a more favorable- result.”2 Rainer v. State, 566 N.W.2d 692, 695 (Minn.1997). But “[a]n evidentiary hearing is unnecessary if the substance of the' affidavit purporting to contain newly discovered evidence, when taken at face value, is insufficient to entitle the petitioner to the relief requested.” Scherf v. State, 788 N.W.2d 504, 508 (Minn.2010). Andersen alleges that an affidavit from his mother, Geraldine Bellanger, constitutes newly discovered evidence that entitles him to a new trial or an evidentiary hearing. Bel-langer claims that Ken Swedberg, Chad Swedberg’s brother, told her and Lisa Swedberg that he saw a blue pick-up truck leave Chad Swedberg’s property at 8:45 a.m. on the morning of the murder, and that the driver of the blue pick-up truck was “Brian,” who was the boyfriend of one of the occupants of Swedberg’s house. The record reflects that Ken Swedberg told police he did not see a blue pick-up truck leave Chad Swedberg’s residence on the morning of the murder. Ken Swed-berg, however, later clarified that he told Bellanger that he wondered where the blue pick-up truck was on the morning of. the murder. The postconviction court concluded that the Bellanger affidavit did not satisfy the third prong of the Rainer test. We affirm on a different ground.
We conclude Andersen failed to satisfy the fourth prong of Rainer that the evidence would probably produce an acquittal or more favorable result. It is improbable that Bellanger’s testimony would produce an acquittal or more favorable result, especially when the murder did'not occur at the Swedberg house and there is no evidence that “Brian” knew the specific site of the maple syruping operation, or knew Chad Swedberg would be there on- the morning of the murder. Consequently, the postconviction court did not err when it summarily denied Andersen’s newly discovered evidence claim.
II.
We next address Andersen’s argument that the postconviction court erred when it [8]*8determined that three of his claims were known or should have been known at the time of his direct appeal, and are therefore barred by the rule announced in State v. Knaffla, 309 Minn. 246, 252, 248 N.W.2d 737, 741 (1976). More specifically, Andersen argues that: (1) recordings of the phone calls Andersen made while in jail were admitted at trial in violation of his constitutional right to counsel; (2) the prosecutor committed misconduct during closing argument, specifically by arguing that Andersen was the only possessor of a “secret” gun when the prosecutor knew that Baker also had such a weapon; and (3) the State withheld certain evidence until after trial. We conclude that the post-conviction court properly determined that these three claims were barred by the Knaffla rule.
Under the Knaffla rule, a petition for postconviction relief raising claims that were raised on direct appeal, or were known or should have been known but were not raised at the time of the direct appeal, are procedurally barred. See id. at 252, 243 N.W.2d at 741; see also King v. State, 649 N.W.2d 149, 156 (Minn.2002). But a claim is not Knaffla-barred if (1) the claim is novel or (2) the interests of fairness and justice warrant relief. Reed v. State, 793 N.W.2d 725, 730 (Minn.2010) (quoting Quick v. State, 692 N.W.2d 438, 439 (Minn.2005)); Powers v. State, 688 N.W.2d 559, 561 (Minn.2004). Petitions allowed under the second exception must have substantive merit and must be asserted without deliberate or inexcusable delay. Wright v. State, 765 N.W.2d 85, 90 (Minn.2009); Deegan v. State, 711 N.W.2d 89, 94 (Minn.2006).
At the time of his direct appeal, Andersen knew that (1) the jail had recorded his telephone calls; (2) under his theory of the case, the prosecutor had improperly argued to the jury that only Andersen had a secret .30-caliber gun; and (3) the State had not disclosed the Baker report in its pretrial discovery. See State v. Andersen, 784 N.W.2d 320, 323 (Minn.2010). None of these claims are novel, nor has Andersen established that these claims have substantial merit and are being asserted without deliberate or inexcusable delay.3 We therefore conclude that the postconvietion court did not err when it concluded that the claims in question were barred by the Knaffla rule.
III.
Andersen next argues that the State improperly introduced evidence through expert testimony from two Bureau of Criminal Apprehension (BCA) experts in violation of the Confrontation Clause. Assuming without deciding that Andersen’s Confrontation Clause claim is not Knaffla-barred, we conclude that the claim is without merit because the disputed evidence was either not testimonial or its admission did not constitute reversible error.
The Confrontation Clause of the United States Constitution provides that a criminal defendant shall have the right “to be [9]*9confronted with the witnesses against him.” U.S. Const, amend. VI; see also Minn. Const, art. I, § 6; State v. Hull, 788 N.W.2d 91, 100 (Minn.2010) (noting that Confrontation Clause claims are analyzed identically under the federal and Minnesota constitutions).
In Crawford v. Washington, the United States Supreme Court considered whether out-of-court statements made to police officers were testimonial and thus could not be admitted at trial. 541 U.S. 36, 68, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004). The Court held that the Confrontation Clause bars the admission of testimonial out-of-court statements unless the declarant is unavailable and the defendant had a prior opportunity to cross-examine the declarant. Id. The Confrontation Clause, however, provides no separate bar to the admission of nontestimonial statements. Id. at 59, 68, 124 S.Ct. 1354; see also Davis v. Washington, 547 U.S. 813, 823-26, 126 S.Ct. 2266, 165 L.Ed.2d 224 (2006). Moreover, the State bears the burden of proving that a statement is not testimonial. See State v. Burrell, 697 N.W.2d 579, 600 (Minn.2005). A successful Confrontation Clause claim has three prerequisites: the statement in question was testimonial, the statement was admitted for the truth of the matter asserted, and the defendant was unable to cross-examine the declarant. Crawford, 541 U.S. at 59 & n. 9, 124 S.Ct. 1354.
The disputed expert testimony was given by two BCA scientists — Nathaniel Pearlson, a firearm and bullet expert, and Patrick Warrick, a finger- and palm-print expert. Pearlson testified that the bullets recovered from the victim were “characteristic of some of the bullets that Winchester uses,” and identified an exhibit displaying several types of Winchester bullets to help explain the differences between the bullets. The witness explained that the exhibit was not prepared for litigation; instead it was marketing material he received as part of an email from a representative of Winchester Owen Corporation. When the prosecutor offered the exhibit, Andersen’s counsel explicitly declined to object to its introduction. Pearl-son then used the exhibit to compare the bullets in the exhibit to the bullets recovered from the victim’s body. He concluded that the bullets recovered from the victim were consistent with .30-caliber Ballistic Silver Tip bullets.
We conclude that the exhibit was not testimonial under Crawford. The exhibit did not constitute a testimonial statement or the functional equivalent of live in-court testimony. See id. at 68, 124 S.Ct. 1354. Instead, the exhibit was used by the witness to explain the different types of Winchester bullets sold by the manufacturer.
The second piece of disputed evidence was Warrick’s palm-print testimony. Warrick testified that he examined a palm print found on Andersen’s Tikka rifle and concluded that the print belonged to Andersen. Subsequently, he sent images of the Tikka rifle and Andersen’s print to a second BCA examiner to conduct a second examination and comparison. The disputed portion of his testimony was that BCA procedure provides that if the second examiner disagrees "with the first examiner’s palm identification, or it is inconclusive, no report may be issued. The logical inference from his testimony was that the second examiner verified Warrick’s test results.
Assuming without deciding that the second examiner’s verification — introduced through Warrick’s expert testimony — was testimonial, we conclude that the error was harmless beyond a reasonable doubt. See State v. Swaney, 787 N.W.2d 541, 555 (Minn.2010) (applying harmless error stan[10]*10dard to violation of the Confrontation Clause to conclude error was harmless beyond a reasonable doubt). The BCA verification procedure testimony was a very small portion of Warrick’s testimony, and created only an inference of agreement by the second examiner. The verification process also was not mentioned in the State’s closing argument. Moreover, Andersen’s counsel extensively cross-examined Warrick on the limitations of his opinion testimony that the palm print belonged to Andersen. In sum, when the evidence is viewed as a whole, the testimony about the BCA verification process was harmless beyond a reasonable doubt.
IV.
Finally, Andersen argues that the postconviction court erred when it denied his claims that his trial and appellate counsel were ineffective. When a claim of ineffective assistance of trial counsel can be determined on the basis of the trial record, the claim must be brought on direct appeal or it is Knaffla-harred. Sanchez-Diaz v. State, 758 N.W.2d 843, 847 (Minn.2008); Torres v. State, 688 N.W.2d 569, 572 (Minn.2004). But such a claim is not Knaffla-harred when the claim requires examination of evidence outside the trial record or additional fact-finding by the postconviction court, because the claim is not based solely on the briefs and trial court transcript. Robinson v. State, 567 N.W.2d 491, 494-95 (Minn.1997).
When an ineffective-assistance-of-counsel claim is properly before us, we examine the claim under the two-prong test set forth in Strickland v. Washington, 466 U.S. 668, 687, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984); State v. Rhodes, 657 N.W.2d 823, 842 (Minn.2003). Under the Strickland test, Andersen must demonstrate that (1) his counsel’s performance fell below an objective standard of reasonableness, and (2) that a reasonable probability exists that the outcome would have been different but for counsel’s errors. Rhodes, 657 N.W.2d at 842 (quoting State v. Lahue, 585 N.W.2d 785, 789 (Minn.1998)). We need not address both the performance and prejudice prongs if one is dispositive. Id. Under the second prong, a defendant must show that, but for counsel’s errors, the result of the proceeding probably would have been different. Id. The reviewing court considers the totality of the evidence before the judge or- jury in making a determination of prejudice. Id.
We will generally not review an ineffective-assistance-of-counsel claim that is based on trial strategy. State v. Bobo, 770 N.W.2d 129, 138 (Minn.2009); Opsahl v. State, 677 N.W.2d 414, 421 (Minn.2004). The extent of trial counsel’s investigation is considered part of trial strategy. Opsahl, 677 N.W.2d at 421. Trial strategy also includes the selection of evidence presented to the jury. White v. State, 711 N.W.2d 106, 111 (Minn.2006) (citing State v. Miller, 666 N.W.2d 703, 716 (Minn.2003)). In evaluating claims of ineffective assistance of counsel, there is a strong presumption that counsel’s performance was reasonable. Boitnott v. State, 631 N.W.2d 362, 370 (Minn.2001).
With these principles in mind, we consider Andersen’s ineffective-assistance-of-counsel claims. We first discuss the ineffective-assistance-of-trial-counsel claims and then consider the ineffective-assistance-of-appellate-eounsel claims.
A.
Andersen argues that his trial counsel was ineffective in three ways. First, Andersen argues before this court that his trial counsel was ineffective because Andersen “wanted to testify” but “could not because [his trial counsel] had not prepared a direct examination” or provided [11]*11him copies of discovery that would have enabled Andersen to make a rational decision about whether to testify.4 Andersen asserts that if trial counsel had prepared a direct examination and provided him with copies of the discovery information, he would have testified. The State responds that even if Andersen proved trial counsel’s alleged deficiencies at an evidentiary hearing, the deficiencies did not prevent Andersen from testifying.
A defendant’s right to testify is protected by the Due Process Clause of the United States Constitution and Minnesota law. State v. Ihnot, 575 N.W.2d 581, 587 (Minn.1998); see Rock v. Arkansas, 483 U.S. 44, 51-53, 107 S.Ct. 2704, 97 L.Ed.2d 37 (1987). The right is personal and may be waived only by the defendant. See State v. Rosillo, 281 N.W.2d 877, 878 (Minn.1979). The defendant’s waiver must be knowingly and voluntarily made. State v. Walen, 563 N.W.2d 742, 751 (Minn.1997). When a defendant knows and understands his right to testify, a claim that his attorneys’ actions denied him the right to testify must fail “absent some indication in the record that [his] lawyers coerced [him] into not testifying by applying undue pressure, using illegitimate means, or otherwise depriving [him] of [his] free will.” State v. Berkovitz, 705 N.W.2d 399, 407 (Minn.2005). The defendant has the burden of proving that he or she did not voluntarily and knowingly waive the right to testify. Walen, 563 N.W.2d at 751. Absent a finding to the contrary, we presume that the defendant waived the right to testify for the reasons stated on the record. See Berkovitz, 705 N.W.2d at 405-08 & n. 1 (noting that we prefer the district court conduct a colloquy on the record). “Solemn declarations in open court carry a strong presumption of verity” and “subsequent presentation of conclusory allegations unsupported by specifics is subject to summary dismissal.” Blackledge v. Allison, 431 U.S. 63, 74, 97 S.Ct. 1621, 52 L.Ed.2d 136 (1977).
At Andersen’s trial, the district court presided over a detailed colloquy between Andersen and his counsel in which Andersen expressly waived under oath his right to testify. The colloquy is as follows:
Defense Counsel [DC]: And we discussed it and you have made the decision that you wish not to testify on your own behalf. Do you understand that?
Andersen [A]: Yes.
DC: And that’s your decision that you have made, and you absolutely want to waive your right, meaning give wp your right to testify on your own behalf?
A: Yes;
DC: And that’s because you feel that the evidence has come in pretty well or pretty good?
A: Without a doubt.
DC: And you would rather go to the jury after closing arguments without you testifying; is that correct?
A: Yes.
DC: Are you sure? We have talked about it at great length; correct?
A: Yes.
DC: And it is a tough decision. I understand that.
A: I stick with my decision.
DC: And that’s the decision not to testify?
[12]*12A: Yes.
(Emphasis added.)
Based on the trial record, Andersen plainly entered a valid waiver of his right to testify on the record, and explicitly stated that he elected not to testify because he believed the State’s evidence was insufficient. When a detailed waiver colloquy is made on the record, the defendant must present more than conclusory allegations unsupported by specific facts to obtain an evidentiary hearing. Andersen does not explain, nor can we discern, why trial counsel should be required to prepare a direct examination before the defendant makes the decision to testify at trial. Likewise, Andersen does not explain, nor can we discern, how obtaining copies of the discovery in question would have affected his decision not to testify, especially when Andersen has failed to establish that the disputed discovery could have affected his decision not to testify. Our case law requires a greater showing than Andersen has alleged in this case. See Berkovitz, 705 N.W.2d at 405-07. Thus, even if Andersen proved at an evidentiary hearing that trial counsel failed to prepare a direct examination and/or provide him copies of the discovery in question, the alleged deficiencies fail to satisfy the prejudice prong of the Strickland test because, according to his sworn statements made at trial, Andersen chose not to testify based on his belief that the State had failed to present sufficient evidence, not the alleged deficiencies of trial counsel. We therefore conclude that the postconviction court did not err when it summarily denied Andersen’s first ineffective-assistance-of-trial-counsel claim.
Second, Andersen argues that trial counsel was ineffective because Andersen wanted to aid in his defense but could not because trial counsel failed to provide him copies of documents that counsel obtained from the State.5 At the time of Andersen’s trial, Minn. R.Crim. P. 9.03, subd. 4 read, “Any materials furnished to an attorney under discovery rules or orders shall remain in the custody of and be used by the attorney only for the purpose of conducting that attorney’s side of the case, and shall be subject to such other terms and conditions as the court may prescribe.” Because the rule prohibited disclosure of discovery to Andersen, trial counsel’s conduct did not fall below an objective standard of reasonableness. In the absence of any specific allegations that would satisfy the first prong of the Strickland test, the petition, files, and records conclusively show that Andersen is not entitled to relief based on his second inef[13]*13fective-assistance-of-trial-counsel claim. Moreover, we conclude that Andersen’s petition for postconviction relief fails to allege facts that would satisfy the prejudice prong of the Strickland test. Andersen has not presented any argument or facts that would support a determination that the disputed discovery probably would have changed the outcome of the case. See Leake v. State, 767 N.W.2d 5, 11 (Minn.2009). We therefore affirm the postconviction court’s summary denial of Andersen’s second ineffective-assistance-of-trial-counsel claim.6
Third, Andersen argues that trial counsel was ineffective because he was unprepared at trial, did not read the discovery, failed to investigate the case, did not call exculpatory witnesses or impeach adverse witnesses, and failed to stay in touch with a prior attorney of the firm and witnesses he subpoenaed. But these claims were known to Andersen at the time of his direct appeal, and therefore are barred by State v. Knaffla, 309 Minn. 246, 252, 243 N.W.2d 737, 741 (1976). Moreover, the claims are not novel and lack merit. Specifically, the conduct at issue arguably falls within trial strategy and therefore is not reviewable. See State v. Davis, 820 N.W.2d 525, 539 n. 10 (Minn.2012) (noting that decisions about which witnesses to interview are typically matters of trial strategy that we will not review); Francis v. State, 781 N.W.2d 892, 898 (Minn.2010) (noting that whether to cross-examine the State’s expert witness constitutes trial strategy); Bobo, 770 N.W.2d at 138-39; Williams v. State, 764 N.W.2d 21, 31 (Minn.2009) (“When determining whether alleged failure to investigate constitutes ineffective assistance of counsel, we consider whether the decision was based on trial strategy or whether it demonstrated that counsel’s performance fell below an objective standard of reasonableness”); Hodgson v. State, 540 N.W.2d 515, 518 (Minn.1995) (noting that defense counsel’s decision not to present evidence that someone else may have committed the murder and not to investigate leads was trial strategy). We therefore affirm the postconviction court’s summary denial of Andersen’s third ineffective-assistance-of-trial-counsel claim.
B.
We next consider Andersen’s ineffective-assistance-of-appellate-counsel claims. Andersen argues that his appellate counsel was ineffective in failing to provide the discovery to Andersen before the direct appeal, and in refusing to request a stay of Andersen’s direct appeal to develop the necessary record to support his ineffective-assistance-of-trial-counsel claims.
The postconviction petition does not allege that Andersen asked his appellate counsel to stay his direct appeal in order to prepare an ineffective-assistance-of-counsel claim on the ground that trial counsel failed to disclose relevant discovery to him that prevented him from making an informed decision about whether to testify. Thus, we conclude the postconviction court did not abuse its discretion in denying Andersen’s claim without a hearing. See Buckingham v. State, 799 N.W.2d 229, 233-34 (Minn.2011).
[14]*14In sum, the postconviction court properly determined that the petition and the files and records of the proceeding conclusively showed that Andersen was not entitled to postconviction relief because his claims are either meritless on their face or barred by the Knaffla rule. Therefore, we affirm the postconviction court’s summary denial of Andersen’s petition for postcon-viction relief.
Affirmed.