Guy Martino appeals from the district court’s grant of summary judgment to his former employer, MCI Communications Services, Inc., and then, following a merger, Verizon Business Network Services, Inc. (Unless specificity is needed, we’ll refer to both as the “company.”) Martino alleges that the company selected him for termination during a nationwide reduction in force (RIF) because he was nearing his 56th birthday. The district court decided that no reasonable jury could reach that conclusion. Martino appeals and we re*450view the district court’s decision de novo, starting with the facts viewed in the light most favorable to Mr. Martino.
Martino wasn’t exactly a spring chicken when he started working for MCI. He was 54. But apparently that didn’t trouble MCI or the manager who interviewed him, Bob Gross. From the time he began his employment in February 2005 to the day he was released in July 2006, Martino reported to Gross in his capacity as a “business solutions consultant,” or BSC. In that position, Martino assisted “core sales teams,” which in turn had a list of client accounts. To use the company’s lingo, Martino was in an “overlay” position — he provided support to sales teams but never actually spearheaded sales. On the other hand, Martino had a financial stake in the process. In addition to a salary, he received commissions on sales to which he was assigned. That would bring quite the benefit when Martino had the good fortune to work on a deal with British Petroleum (BP).
MCI began negotiations with BP over the summer of 2005 and closed the deal in October. Although Martino was technically part of the team working the deal, his involvement was peripheral. Even at that, some of the more important people involved — David Schiffman, the core sales rep who “quarterbacked the deal,” and Steve Rumstein, the director of IT Hosting Solutions — thought Martino came up short. One of the key phases of the transaction was drafting a response to BP’s “request for proposal” (RFP). (In plain English, this was MCI’s chance to make its sales pitch.) Schiffman took the lead on the project, but he assigned Martino the task of “gathering and providing technical information regarding data centers and hosting-related products and services.” From Schiffman’s perspective, Martino failed in that task. He “did not do an adequate job in assisting the core sales team”; he “provided inaccurate and/or incomplete information”; he was “frequently inaccessible”; and he was not a team player. Rumstein’s impression was much the same. He expected Martino to “take a much stronger leadership role in driving the sale process.” Naturally, Martino had a different view of things. He said his colleagues relied “very heavily on [him] from a hosting standpoint,” and he “would like to think” of himself as a “key person” on the deal.
In any event, Martino got credit for the BP deal after it closed, boosting his sales figures well over quota and turning a handsome commission. That was, however, the sole bright spot in an otherwise unremarkable tenure. Nearly 85 percent of Martino’s sales during his employment resulted from the BP deal and, as we noted, Martino’s credit for those sales was more a consequence of the nature of MCI’s crediting policy — to acknowledge everyone involved, no matter how weak their performance — than of hard work. When the BP deal wasn’t in play, Martino’s numbers suffered. In 2005, he failed to reach quota during March, April, May, June, July, August, September, November, and December. And in the absence of the BP deal, Martino would have hit only 47 percent of his quota for 2006. Martino therefore puts it rather mildly when he says the BP transaction was just a “significant contributor to [his] overall performance.”
But apart from the fact that the BP deal was Martino’s only big success — -and he was hardly instrumental to the deal at that — Martino’s skill set was soon to be obsolete. Following MCI’s merger with Verizon in January 2006, sweeping changes took place. Verizon did a “redundancy analysis” to identify duplicative positions, a common practice when companies in the same industry merge. Naturally, *451redundancies were found, and that led to a RIF. But in Martino’s department at least, something else drove the RIF. Verizon wanted to shift the focus of that department, IT Hosting Solutions, to the provision of more complex services. (Bear with us now as we march through the associated jargon.)
During Martino’s employment with the company, the data services sold were either “colocation” services or “managed hosting” services. Colocation involved a client purchasing space, power, and cooling for its servers in the company’s data centers. The client managed the servers. In managed services, on the other hand, the client purchased space, power, and cooling, but the company managed the servers and applications. So colocation services were more basic (and thus cheaper), while managed services were more elaborate, involved more work on the company’s end, and presumably generated more revenue. When Verizon took over it decided to change the role of the BSCs in all of this. Going forward, BSCs would no longer be responsible for selling basic colocation services. Instead, they would only receive credit and commissions for the sale of managed services, and they would be expected to take a more active role in the sales process. This was problematic for Martino because he had limited success selling managed services; the majority of his qualified sales — including the BP transaction in which he did not meaningfully participate — consisted of basic colocation services. Martino was thus a prime candidate for termination.
And terminated he was. Word came down in June 2006 that 35 employees needed to be discharged from IT Services (of which Martino’s group, IT Hosting Solutions, was a subunit). Ed Franklin, the vice-president of IT Services, in turn asked Rumstein to submit a list of individuals from IT Hosting Solutions least likely to contribute to the company moving forward. In addition to Martino, Rumstein identified five others, all of whom were significantly younger than Martino (ranging in age from 33 to 45). He considered several factors to come up with the names, including geographical coverage (because some areas were staffed more adequately than others); the employee’s “demonstrated ability” to sell the complete product line (with a particular emphasis on managed hosting services); credibility with core sales teams; and “actual sales performance.” Rumstein placed limited weight on this last factor, however, since, as we have explained, the sales numbers could be skewed by virtue of the crediting system. Considering all the factors, Rumstein believed Martino looked good on paper but brought little to the table beyond that, especially in light of the shifted focus to managed hosting services.
Franklin then decided to discharge Martino along similar lines:
After receiving Mr. Rumstein’s list, I selected Mr. Martino for the RIF based on his performance. Although Mr. Martino’s qualified sales numbers were very good on their face, a deeper inquiry into his performance revealed that he was not as strong as his numbers suggested. First, the overwhelming majority of his qualified sales were a result of one sale to BP that occurred in 2005. Based on information I received from my reports, most notably [from] Mr. Rumstein, I was aware that Mr. Martino did not participate in the BP sale in a meaningful way. Second, Mr. Martino’s track record as a Business Solutions Consultant demonstrated that he had a limited ability to sell managed services or application-based services and, instead, the majority of his qualified sales consisted of basic colocation services. At the time *452of the RIF, Verizon ... was planning to change the BSC position such that BSCs would be responsible for selling only managed services, and no longer be responsible for selling colocation services, in the second half of 2006 or the beginning of 2007.
In other words, Franklin’s decision was “based entirely on the needs of Verizon ... and nothing else.” Indeed, from the looks of it, Franklin was unaware at the time of Martino’s age.
So, where is the evidence of age discrimination? Martino concedes that there is none when it comes to either the actual decisionmaker (Franklin) or the manager who offered him up as a RIF candidate (Rumstein). Invoking the “cat’s paw” theory, however, he contends that he has a viable claim for discriminatory discharge through the alleged animus of a nondecisionmaker — Gross, his immediate supervisor. According to Martino, Gross sometimes called him an “oldtimer.” That doesn’t strike us as inherently offensive— between friends it’s often a term of endearment — but Martino sees it as evidence of bias in favor of the young. Before we test that claim, though, let’s review the basics of age discrimination in employment and the more specialized doctrine known as the cat’s paw.
The Age Discrimination in Employment Act (ADEA) prohibits employers from firing workers who are 40 or older on the basis of their age. 29 U.S.C. § 623(a)(1), 631(a); Hemsworth v. Quotesmith.Com, Inc., 476 F.3d 487, 490 (7th Cir.2007). A plaintiff suing under the ADEA may show discrimination directly or indirectly, in the latter instance through the approach established in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). Because both methods allow the use of circumstantial evidence, however, the distinction is often fleeting. See Faas v. Sears, Roebuck & Co., 532 F.3d 633, 641 (7th Cir.2008) (“We have noted that because a plaintiff may utilize circumstantial evidence under both methods of proof, the distinction between the two avenues of proof is vague, and the terms ‘direct’ and ‘indirect’ themselves are some-what misleading in the present context.”) (internal quotation marks and alterations omitted). In either case, the bottom-line question is whether the plaintiff has proved intentional discrimination, Olson v. Northern FS, Inc., 387 F.3d 632, 635 (7th Cir.2004), and therefore much — if not all — of the same evidence is at play, see, e.g., Venturelli v. ARC Cmty. Servs., Inc., 350 F.3d 592, 601 (7th Cir.2003) (describing one category of overlapping evidence).
Here, Martino attempts to show intentional discrimination through both methods of proof. We start with the direct method, which requires Martino to offer direct or circumstantial evidence that Verizon’s decision to terminate was motivated by age. Tubergen v. St. Vincent Hosp. & Health Care Ctr., Inc., 517 F.3d 470, 473 (7th Cir.2008). Martino focuses on Gross’s “oldtimer” comments, but because Gross was not a decisionmaker, these comments are only relevant if Gross had “singular influence” over the decision-maker (Franklin and, to a lesser extent, Rumstein) and “use[d] that influence to cause the adverse employment action.” Staub v. Proctor Hosp., 560 F.3d 647, 651 (7th Cir.2009); see also Lucas v. Chicago Transit Auth., 367 F.3d 714, 730 (7th Cir.2004) (“Generally speaking, comments by a non-decision maker do not suffice as evidence of discriminatory intent.”). The company can defeat this tack — what we call the cat’s paw theory — by showing that, even if Gross was biased and attempted to get Martino terminated for this reason, the decisionmaker did an inde*453pendent analysis and came to his own conclusion. See Staub, 560 F.3d at 656 (“If the decisionmaker wasn’t used as a cat’s paw — if she didn’t just take the monkey’s word for it, as it were — then of course the theory is not in play.”). This is where Martino gets tripped up. This is a particularly weak cat’s paw case because, even if we assume Gross was prejudiced, there were not one but two layers of bias-free analysis. Rumstein, who knew Martino’s work well (having been the director of IT Hosting Solutions), considered several legitimate factors in choosing Martino as a RIF candidate. And Franklin did the same when he made the final decision to adopt Rumstein’s recommendation. The fact that Rumstein conferred with Gross about Martino’s performance on the BP deal does not mean that Gross’s supposed animus should be imputed to Rumstein (and ultimately to Franklin). As we noted in Staub, a decision-maker is not required
to be a paragon of independence. It is enough that the decisionmaker “is not wholly dependent on a single source of information” and conducts her “own investigation into the facts relevant to the decision.” To require much more than that would be to ignore the realities of the workplace. Decisionmakers usually have to rely on others’ opinions to some extent because they are removed from the underlying situation. But to be a cat’s paw requires more; true to the fable, it requires a blind reliance, the stuff of “singular influence.”
Staub, 560 F.3d at 659 (quoting Brewer v. Bd. of Trs. of Univ. of Ill., 479 F.3d 908, 918 (7th Cir.2007)) (citations omitted). We don’t have anything close to “singular influence” in this case (to say nothing of the fact that the “oldtimer” remark just isn’t that egregious). A reasonable jury could not have found in favor of Martino under the direct method of proof.1
The same is true for the indirect method. To establish a prima facie case of age discrimination under the indirect method, Martino must prove that (1) he is a member of a protected class (which he is, being 40 or older); (2) his performance met the company’s legitimate expectations; (3) despite his performance he was subject to an adverse employment action (again, not at issue); and (4) the company treated similarly situated employees under 40 more favorably. Faas, 532 F.3d at 641. If Martino satisfies these criteria, the company may provide a legitimate, nondiscriminatory reason for the termination. Id. at 641-42. Assuming the company offers as much, Martino may challenge the stated reason as a pretext for discrimination. Id. at 642. Again, however, the ultimate burden to prove intentional discrimination always remains with Martino. Greene v. Potter, 557 F.3d 765, 769 (7th Cir.2009).
Martino cannot make out the second and fourth prongs of a prima facie *454case. Quite simply, a reasonable jury could find neither that he met the company’s expectations nor that the company treated younger workers any better. As to business expectations, the evidence shows that Martino only excelled in one instance — the BP deal' — and even then that success was limited to paper. He had the good foi'tune of being staffed on an historic deal — which earned him sales credit and commissions in abundance — but it was more of a windfall than anything else. In his supervisors’ eyes, he was not a team player, he was not available, and he didn’t take a very active role in the process. Perhaps he helped out, but making some contribution isn’t the same as making the expected contribution. And, as we have emphasized, Martino’s importance to the company was waning. After the merger, the company strived to focus on managed hosting services rather than colocation services. But Martino showed little promise in that arena. Choosing to terminate someone on the basis of old age is impermissible; choosing to let someone go because they have an obsolete skill set, on the other hand, is completely kosher.
Martino doesn’t fare any better when it comes to showing preferential treatment for similarly situated younger employees. First off, a number of younger BSCs were let go right along with Martino. Though that’s not the dispositive issue here — the issue is whether younger workers were favored, i.e., kept on the payroll, because of their age — it’s still relevant. It’s hard to make out a case for age discrimination when younger workers are also being shown the door. Nevertheless, it is true that the company retained six BSCs who reported to Gross and were younger than Martino at the time of his discharge. See Amrhein v. Health Care Serv. Corp., 546 F.3d 854, 860 (7th Cir.2008) (explaining that “similarly situated” employees are those with the same manager and same responsibilities). We agree with the district court that two of the six BSCs — Greg Smidt and Ray Yaeger — provide little aid to Martino because they were 47 and 49, respectively. See Tubergen, 517 F.3d at 475 n. 4 (“Under the ADEA, in the case of younger employees that fall above the age of forty, the age difference must be ten years or greater in order to be presumptively substantial.”). Yet we admit that the other four BSCs, those under 40, are more difficult cases. They were far from stellar performers; it would not be surprising if the company had discharged them as well. But on the other hand, it was Martino uniquely who had lost the confidence of the core sales teams, and he apparently stood out as poorly equipped to assist in the sale of managed hosting services. RIF decisions often involve splitting hairs, and sometimes employers make mistakes, retaining an inferior worker for lack of omniscience. See Testerman v. EDS Technical Prods. Corp., 98 F.3d 297, 304 (7th Cir.1996) (in RIF cases, “we deal with small gradations, with an employer’s subjective comparison of one employee to another, and it is incumbent upon us to remember that what is at issue is not the wisdom of an employer’s decision, but the genuineness of the employer’s motives.”). Maybe that happened here. Maybe not. What is clear is that the company didn’t RIF Martino because he was “over the hill.” Because we concur with the district court that Martino failed to make out a prima facie case of age discrimination, we need not address the other parts of the indirect framework.
A pair of parting remarks. We mentioned at the outset that Martino was interviewed and hired by the same man, Bob Gross, who supposedly sought to see him fired because of his age less than two years later. Common sense tells us that it’s unlikely for a person to suddenly devel*455op a strong bias against older folks. That’s not to say it’s impossible, just that the much more likely scenario involves a person harboring prejudice from the beginning, particularly when we’re talking about a relatively short time frame. Nevertheless, a pair of our recent decisions suggests that this logic — called the “same-actor” inference — may be flawed in some cases. Petts v. Rockledge Furniture, LLC, 534 F.3d 715, 724-25 (7th Cir.2008); Filar v. Bd. of Educ. of City of Chicago, 526 F.3d 1054, 1065 n. 4 (7th Cir.2008). We do not, however, believe these decisions make the same-actor inference completely off-limits. Instead, we simply cautioned courts not to place “too strong a reliance” on the inference of nondiscrimination. Filar, 526 F.3d at 1065 n. 4. We can heed that advice in this case — and still adhere to common sense — by concluding that while the same-actor inference doesn’t defeat Martino’s claim, it’s one more thing stacked against him.
And if there were any doubt that Martino cannot survive summary judgment, it evaporates completely in the wake of the Supreme Court’s decision in Gross v. FBL Financial Services, Inc., — U.S. -, 129 S.Ct. 2343, 174 L.Ed.2d 119 (2009). The Court held that in the ADEA context, it’s not enough to show that age was á motivating factor. The plaintiff must prove that, but for his age, the adverse action would not have occurred. Martino cannot handle that. At best, he has done no more than show that his age possibly solidified the decision to include him in the RIF. But a reasonable jury could only conclude that he would have been fired anyway; age was not a but-for cause.
The judgment of the district court is Affirmed.