DAHDAH v. ROCKET MORTGAGE LLC (2026)

OPINION

This case requires us to consider what actions qualify as valid “offers” and “acceptances” to form contracts in the online world. Websites often tell consumers that they will “agree to” or “accept” various contractual terms (including arbitration provisions) by engaging in certain conduct, such as clicking a button or registering an account. If consumers take the specified action, when do these online proposals create binding contracts to arbitrate? We must consider that question under California contract law.

LowerMyBills.com refers internet users who are interested in refinancing their home mortgages to affiliated lending partners, including Rocket Mortgage. The website tells users that they will agree to its hyperlinked “Terms of Use”—including a mandatory arbitration provision—if they click on a particular button. Michael Dahdah visited this website three times, inputted his information, and clicked the critical buttons. LowerMyBills referred him to Rocket. When Dahdah later received calls from Rocket that he did not want, he sued the company in federal court. Rocket responded by invoking LowerMyBills’ arbitration provision. But the district court held that Dahdah's “click” did not create an enforceable agreement. We disagree. Under the significant body of circuit precedent interpreting California law, LowerMyBills gave Dahdah sufficiently conspicuous notice that he would accept the proposed terms by clicking the button. So his decision to take this action qualified as a valid “acceptance” of LowerMyBills’ “offer” to contract. The district court thus should have granted Rocket's motion to compel arbitration. We reverse.

I

In February 2020, Dahdah bought a home in Washington, D.C. He financed this purchase by obtaining a mortgage loan from Embrace Home Loans.

Dahdah soon started to think about refinancing this mortgage. Three times over the next year, he used his cellphone to visit the website LowerMyBills.com. This website provides “free online financial” information about home mortgages and operates “a free referral service for consumers” who want to obtain an initial mortgage or to refinance an existing one. Viner Decl., R.11-2, PageID 72–73. LowerMyBills shares the same parent company as Rocket Mortgage (which called itself Quicken Loans at the relevant times). Rocket provides initial home loans and refinancing options to consumers.

Dahdah's first visit to LowerMyBills occurred on October 21, 2020. At that time, the website took consumers interested in refinancing through five webpages. The first few pages asked them questions about themselves, their house, and their existing loan.

The fourth page asked consumers to input (among other things) their age, their employment status, their number of late mortgage payments over the last 12 months, their property's address, and their email address. Below the box for an email address, this page included a large green button labeled “Calculate.” Viner Decl. Ex. A-1, R.11-3, PageID 96. And below this green button in much smaller font, the page stated: “By clicking the button above, you agree to be contacted by LowerMyBills, at the address entered above for promotional emails and consent, electronically via E-sign, to the LMB Lending Terms of Use, Privacy Policy, and Consent to Doing Business Electronically.” Id. If consumers clicked on the hyperlinked “Terms of Use” phrase, the website would take them to LowerMyBills’ Terms of Use. These terms included a provision in all caps that required consumers to participate in “ARBITRATION” for “ALL CLAIMS, DISPUTES OR CONTROVERSIES” that they might have with LowerMyBills or with any affiliate (including Rocket). Terms of Use, R.11-7, PageID 114. The bottom of this fourth page resembled the following image:

Viner Decl. Ex. A-1, R.11-3, PageID 96.

The fifth page asked consumers to list (among other things) their name, residence, and phone number. Below the boxes for phone numbers, the page contained another large green button labeled “Calculate your FREE results.” Id., PageID 97. And below this button in much smaller font, the page stated that “[b]y clicking the button above,” consumers would “express [their] understanding and consent” to several things, including the Terms of Use. Id. The bottom of this fifth page looked like the following image:

Id.

LowerMyBills kept records indicating that Dahdah proceeded through these five pages on October 21. The records showed that he clicked the “Calculate” button on the fourth page after he entered his email address. And he clicked the “Calculate your FREE results” button on the fifth page after he entered his name and phone number. So if these clicks created a contract, Dahdah agreed to adhere to the Terms of Use and receive calls from LowerMyBills or an affiliate (even if he had registered on a Do Not Call registry). After he clicked the final button, LowerMyBills matched him with Rocket and identified refinancing terms that would lead to a “possible monthly payment decrease.” Viner Decl., R.11-2, PageID 81. But he chose not to pursue the refinancing.

Dahdah next visited LowerMyBills on February 18, 2021. This time, the website had a different setup. Dahdah would have submitted similar information in the initial pages. The final page asked for his name and phone number. It then contained a disclaimer in small font: “By clicking the button below, you acknowledge, consent, and agree to our terms at the bottom of th[is] page[.]” Viner Decl. Ex. A-2, R.11-4, PageID 105. A purple button stated “Calculate your results” in a large font. Id. Below the button, the website next suggested that Dahdah's clicking on the button would confirm his agreement to many things, including the hyperlinked Terms of Use. This final page looked like the following screenshot:

Id. LowerMyBills kept records showing that Dahdah inputted his phone number and clicked this button. The website's results again “matched” him with a refinancing option from Rocket at a certain interest rate and payment schedule. Viner Decl., R.11-2, PageID 84. But he again did not take any further action.

The next day, Dahdah visited LowerMyBills for a third time. The website's layout for this third visit resembled its layout for his first one. So Dahdah twice clicked buttons saying either “Calculate” or “Calculate your FREE results.” And language immediately below these buttons suggested that he would agree to (among other things) the Terms of Use if he clicked them. The results matched him to Rocket for a third time. And for a third time, he did not pursue refinancing with Rocket.

According to Dahdah's complaint, however, Rocket did not give up. He alleges that he put his phone number on a Do Not Call registry back in 2017. Between June 27 and June 30, 2022, though, Rocket called him at least eight times to sell him mortgage products or services. Rocket made some calls as early as 5:00 a.m. And when Dahdah answered other calls, nobody would respond for more than two seconds if at all. Although Dahdah told Rocket to stop calling, it continued to do so.

Finding these calls frustrating, Dahdah brought a class-action complaint against Rocket under the Telephone Consumer Protection Act. See 47 U.S.C. § 227. The complaint alleged that Rocket violated this Act by repeatedly calling people who had listed themselves on the Do Not Call registry. It also alleged that Rocket had violated the Act in other ways: by continuing to call people who opted out of the communications, by failing to follow certain rules for automated calls, and by calling at illegal hours.

In response, Rocket moved to compel Dahdah to arbitrate his dispute or, in the alternative, to dismiss his complaint for failure to state a claim. As relevant here, it asserted that Dahdah had agreed to LowerMyBills’ Terms of Use because he repeatedly clicked the buttons on the website. And those terms included a broad arbitration provision.

The district court resolved these motions across three orders. The court first dismissed the complaint on the merits and denied Rocket's request for arbitration as moot. See Dahdah v. Rocket Mortg., LLC, 2023 WL 5941730, at *5 (E.D. Mich. Sept. 12, 2023). But the court quickly realized that it should have decided the arbitration question ahead of the merits question. Dahdah v. Rocket Mortg., LLC, 2023 WL 11944898, at *1 (E.D. Mich. Nov. 17, 2023). So it reopened the case. See id. But it found that Dahdah and Rocket had not formed a binding contract to arbitrate this dispute and denied Rocket's motion to arbitrate. See id. at *2–9. Several months later, the court denied Rocket's request that it reconsider this ruling and granted Dahdah's motion to amend his complaint. See Dahdah v. Rocket Mortg., LLC, 2024 WL 4299564, at *8 (E.D. Mich. Sept. 26, 2024). Rocket immediately appealed the district court's denial of its request to arbitrate. See 9 U.S.C. § 16(a)(1)(A)–(B). We review the district court's refusal to compel arbitration de novo. See Floss v. Ryan's Fam. Steak Houses, Inc., 211 F.3d 306, 311 (6th Cir. 2000).

II

The Federal Arbitration Act makes arbitration agreements “valid, irrevocable, and enforceable, save upon such grounds as exist at law or in equity for the revocation of any contract[.]” 9 U.S.C. § 2. This text instructs courts to follow the same contract-law rules when analyzing an arbitration agreement that they would follow when analyzing any other agreement. See Morgan v. Sundance, Inc., 596 U.S. 411, 418, 142 S.Ct. 1708, 212 L.Ed.2d 753 (2022). And ever since Erie Railroad Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938), federal courts have relied on state law to identify these contract-law rules. See, e.g., Talley v. State Farm Fire & Cas. Co., 223 F.3d 323, 326 (6th Cir. 2000); Equitable Life Assurance Soc'y of U.S. v. Poe, 143 F.3d 1013, 1016 (6th Cir. 1998). So state law likewise governs whether the parties formed a binding contract to arbitrate. See Seawright v. Am. Gen. Fin. Servs., Inc., 507 F.3d 967, 972 (6th Cir. 2007); Berman v. Freedom Fin. Network, LLC, 30 F.4th 849, 855 (9th Cir. 2022). Because Dahdah and Rocket agree that California law governs their dispute, see Appellant's Br. 18, we may assume that choice of law. See AtriCure, Inc. v. Meng, 12 F.4th 516, 525 (6th Cir. 2021). We thus start by analyzing California law on this question.

A. California Contract Law

California law follows typical contract-formation principles. We will describe its general rules before turning to the specific standards that govern the kind of contract offer at issue here.

1. General “Offer” and “Acceptance” Principles

To enter a contract, both sides must show “[m]utual assent” to “the same thing in the same sense.” Sellers v. JustAnswer LLC, 73 Cal.App.5th 444, 289 Cal. Rptr. 3d 1, 12 (2021) (citation omitted); see Donovan v. RRL Corp., 26 Cal.4th 261, 109 Cal.Rptr.2d 807, 27 P.3d 702, 709 (2001). Parties often show this assent when one side makes an adequate “offer” and the other responds with an adequate “acceptance.” See Donovan, 26 Cal.4th 261, 109 Cal.Rptr.2d 807, 27 P.3d at 709; Restatement (Second) of Contracts § 22(1) (A.L.I. 1981). To decide whether parties made valid offers and acceptances, courts look to their “outward manifestations” or “expressions,” not their inward intents or motives. B.D. v. Blizzard Ent., Inc., 76 Cal.App.5th 931, 292 Cal. Rptr. 3d 47, 58 (2022) (citation omitted). In other words, contract creation generally depends on an objective review of the parties’ actions—not a subjective review of their beliefs. See Windsor Mills, Inc. v. Collins & Aikman Corp., 25 Cal.App.3d 987, 101 Cal. Rptr. 347, 350 (1972); Restatement (Second) of Contracts § 2 cmt. b.

This objective inquiry applies to both offers and acceptances. As for offers, a party's expression must contain conspicuous enough words that a “reasonable person” who heard them would believe that they revealed the offeror's assent to contract. Meyer v. Benko, 55 Cal.App.3d 937, 127 Cal. Rptr. 846, 848 (1976). As for acceptances, if a reasonable person would “objectively” view the offeree's response to the offer as assent to its terms, the parties have formed a contract even if the offeree did not intend that result. Blizzard Ent., 292 Cal. Rptr. 3d at 58. And while offerees “traditionally” show this assent through “written or spoken” language, they may “do so through conduct” too. Berman, 30 F.4th at 855; see Restatement (Second) of Contracts § 19. For example, the offer itself might propose that the offeree accept “by performing” a “specified act,” such as an employee's decision to remain employed after learning of new employment terms. Restatement (Second) of Contracts § 30(1); see Mar v. Perkins, 102 Cal.App.5th 201, 321 Cal. Rptr. 3d 268, 277–80 (2024).

The objective test can help or hurt offerors. On the one hand, it means that offerors can enforce contracts even if the offerees do not read an offer's terms or understand its contents, so long as an offeree knows an offer has been made. See Windsor Mills, 101 Cal. Rptr. at 350; Cunningham v. Int'l Comm. of Y.M.C.A., 51 Cal.App. 487, 197 P. 140, 141 (1921). Suppose, for example, that a customer at a car dealer signs a document entitled “contract” that states the terms governing a car purchase. Even if the customer does not read the document before signing it, a court would likely conclude that the customer accepted its terms. See Marin Storage & Trucking, Inc. v. Benco Contracting & Eng'g, Inc., 89 Cal.App.4th 1042, 107 Cal. Rptr. 2d 645, 651 (2001). On the other hand, the objective test means that offerors cannot attempt to create a contract by disguising “inconspicuous” terms in “a document whose contractual nature is not obvious.” Windsor Mills, 101 Cal. Rptr. at 351. After all, an offeree's conduct does not objectively show acceptance of an offer if no reasonable person would realize that an offer had “been made” in the first place. Id.; see Meyer v. Uber Techs., Inc., 868 F.3d 66, 74 (2d Cir. 2017).

The same basic rules continue to apply when parties switch from the real world to the virtual one. See Blizzard Ent., 292 Cal. Rptr. 3d at 58–59; Long v. Provide Commerce, Inc., 245 Cal.App.4th 855, 200 Cal. Rptr. 3d 117, 122 (2016). Parties can form agreements over the internet if an offeror who runs a website adequately conveys contractual terms and an offeree who uses the website adequately conveys an acceptance. See Berman, 30 F.4th at 856. To be sure, just like parties who fail to read documents before signing them, website users often claim to lack “actual knowledge” of the terms after the fact. See id. So the validity of an online offer and acceptance in that circumstance will still turn on whether a reasonable person would objectively view both sides’ conduct as a “manifestation of assent” to the contract's terms. Sellers, 289 Cal. Rptr. 3d at 13.

Websites try to propose cognizable offers in different ways. Which types are “sufficiently conspicuous” to create enforceable contracts? Blizzard Ent., 292 Cal. Rptr. 3d at 65. California law gives clear answers at the extremes. On one side, any reasonable person would conclude that so-called scrollwrap or clickwrap offers objectively convey the website operator's “manifestation of [a] willingness to enter into a bargain” with website users. Restatement (Second) of Contracts § 24; see Keebaugh v. Warner Bros. Ent., Inc., 100 F.4th 1005, 1014 (9th Cir. 2024). To create a scrollwrap agreement, users “must scroll through all the Terms of Service before they can click the mandatory ‘I agree’ box.” Keebaugh, 100 F.4th at 1014. To create a clickwrap agreement, users must “click on an ‘I agree’ box after being presented with a list of terms and conditions of use.” Id. Both types of offers thus make the terms “readily available” for users and tell the users to click an “I agree” or “I accept” button if they assent to them. See Blizzard Ent., 292 Cal. Rptr. 3d at 65; Sellers, 289 Cal. Rptr. 3d at 15. Given this design, a website user “knows or has reason to know that [the operator] may infer from his conduct that he assents” by clicking the button. Berman, 30 F.4th at 856 (quoting Restatement (Second) of Contracts § 19(2)). Courts thus have generally found that these offers can create valid contracts because the operator explicitly discloses the offer and requires users to take a specific action exclusively to convey their acceptance. See Keebaugh, 100 F.4th at 1014.

So-called browsewrap proposals fall on “the other end” of potential offers. Oberstein v. Live Nation Ent., Inc., 60 F.4th 505, 514 (9th Cir. 2023) (quoting Berman, 30 F.4th at 856). These proposals place a hyperlink to the operator's terms somewhere on its website. See Long, 200 Cal. Rptr. 3d at 123. The website also might contain a notice somewhere asserting that users will accept those terms merely by using that site. See id. Yet website operators often bury the link to these terms, so reasonable users often will lack any “reason to know” that their use of the site signals their acceptance of those hidden terms. Restatement (Second) of Contracts § 19(2); see Nguyen v. Barnes & Noble Inc., 763 F.3d 1171, 1178–79 (9th Cir. 2014). Indeed, this type of offer might represent the virtual equivalent of hiding “inconspicuous contractual” terms in a “document” that does not look like a contract. Windsor Mills, 101 Cal. Rptr. at 351. So courts often hold that these offers cannot create valid agreements because they leave users “unaware” that the operator has even proposed an offer. Berman, 30 F.4th at 856; see Long, 200 Cal. Rptr. 3d at 123–27.

Yet many proposals fall in between these extremes. These “hybrid” offers (what some courts have called “sign-in wrap” offers) present the trickiest cases. Blizzard Ent., 292 Cal. Rptr. 3d at 60 (citation omitted); see Keebaugh, 100 F.4th at 1014. Like clickwrap offers, hybrid offers include a hyperlink to a proposal's terms. Sellers, 289 Cal. Rptr. 3d at 15. Yet unlike clickwrap offers, the offers do not call the user's specific attention to the terms through a pop-up box or similar feature that requires them to click an explicit “I agree” button. Blizzard Ent., 292 Cal. Rptr. 3d at 60. Rather, the websites include language that the users will manifest their acceptance if they take other action (such as creating an account or clicking an entry button). See, e.g., Keebaugh, 100 F.4th at 1014; Oberstein, 60 F.4th at 515–16. To determine the validity of these proposals, courts have relied on the same general offer and acceptance rule that they require elsewhere. To qualify as a valid offer, the website's proposal must give “reasonably conspicuous” notice that a user will accept the terms by clicking a button, signing into an account, or taking a similar action. Berman, 30 F.4th at 856. And to qualify as a valid acceptance, the user must objectively show assent to the terms by (for example) taking the specified action. See id.

2. Specific Standards for “Hybrid” Offers

Courts from around the country have analyzed these “hybrid” offers under California law. See Dhruva v. CuriosityStream, Inc., 131 F.4th 146, 151–55 & n.1 (4th Cir. 2025); Chabolla v. ClassPass Inc., 129 F.4th 1147, 1155–58 (9th Cir. 2025); Selden v. Airbnb, Inc., 4 F.4th 148, 155–58 (D.C. Cir. 2021); Meyer, 868 F.3d at 74–80. These courts have held that they (rather than a jury) may decide whether a specific website has made a sufficiently “conspicuous” offer when the parties do not dispute the historical facts (e.g., the facts about the layout of the website, the font size, or the like). See Meyer, 868 F.3d at 80; Oberstein, 60 F.4th at 517. At the same time, courts have resolved this conspicuousness question using a “fact-intensive” approach that turns on all the circumstances of each website. Oberstein, 60 F.4th at 515; see Sellers, 289 Cal. Rptr. 3d at 23. Ultimately, website owners have the duty to “put users on notice of the terms for which they wish to bind consumers[.]” Chabolla 129 F.4th at 1156 (quoting Sellers, 289 Cal. Rptr. 3d at 20).

Courts largely ask the same general questions when deciding whether a “reasonably prudent Internet user” would have found a website's proposed terms “conspicuous.” Berman, 30 F.4th at 856. We will highlight four of those common questions, although the nature of a “totality of the circumstances” test limits the weight courts should ascribe to any particular question.

Question One: Did the website display the offer on an “uncluttered” page, Meyer, 868 F.3d at 78, or on a page filled with items that will “draw the user's attention away from” the proposal, Berman, 30 F.4th at 857? Courts are more likely to find an offer conspicuous if a page uses a “simple, streamlined design” to convey it. Selden, 4 F.4th at 157; see Patrick v. Running Warehouse, LLC, 93 F.4th 468, 477 (9th Cir. 2024); Dohrmann v. Intuit, Inc., 823 F. App'x 482, 484-85 (9th Cir. 2020) (memorandum). And they are less likely to find it conspicuous if a page has a lot of distractions. See Berman, 30 F.4th at 857.

A comparison of Meyer and Berman illustrates this point. Meyer addressed the registration process for Uber. 868 F.3d at 70–71. The screen for entering credit-card information included a large gray “REGISTER” button followed by links to alternative payment methods. Id. at 71. Under those links, it then notified users in a smaller font: “By creating an Uber account, you agree to the TERMS OF SERVICE & PRIVACY POLICY.” Id. If users clicked on the hyperlinked terms-of-service phrase, they would get sent those terms (including an arbitration provision). Id. The screen looked like the following image:

Id. at 81. Applying California law, the Second Circuit held that this screen provided sufficiently conspicuous notice of its offer's terms (including the arbitration provision). See id. at 77–78. Although Uber used a small font for its proposal, the court emphasized that the “uncluttered” page did not contain a lot of other information that might distract users. Id.

Now compare that page to the one in Berman. There, Fluent (a marketing company) offered gift cards and free samples on its webpages to get consumers to give their contact information and take surveys. Berman, 30 F.4th at 853. One page looked like the following screenshot:

Id. at 853, 859. Below the box to confirm a zip code and above the button that read “This is correct, Continue!”, the page included the following text in “tiny gray font”: “I understand and agree to the Terms & Conditions which includes mandatory arbitration and Privacy Policy.” Id. at 853-54. The Ninth Circuit held that this page did not include a conspicuous offer of the terms (including the arbitration provision). Id. at 856–57. The court reasoned (among other things) that the page's other “visual elements” would take a user's attention away from the terms. Id.

Question Two: Did the website operator place the proposed offer close to—or away from—the button that a user must click to signal the user's acceptance of the proposal? The closer that the operator places the offer to the critical action button, the more conspicuous courts will find it. See Oberstein, 60 F.4th at 515–16. And the farther away that the operator places the offer from this button, the less conspicuous courts will find it. See Chabolla, 129 F.4th at 1157.

A comparison of two Ninth Circuit opinions (Oberstein and Chabolla) demonstrates this second point. Oberstein addressed Ticketmaster's website. 60 F.4th at 515. Ticketmaster presented its offer to users at “three independent stages” during visits to its site: when they created an account, when they signed into the account, and when they completed a purchase. Id. Consider the sign-in page. Immediately above the blue “Sign in” button, the site told users that “[b]y continuing past this page, you agree to the Terms of Use” (which included an arbitration provision):

Oberstein v. Live Nation Ent., Inc., 2021 WL 4772885, at *2 (C.D. Cal. Sept. 20, 2021). The Ninth Circuit found the offer reasonably conspicuous. Oberstein, 60 F.4th at 515–16 (Boggs, J.). As relevant now, it explained that Ticketmaster placed the proposal to agree to the terms “directly above or below the action button” in each of the three relevant pages. Id. at 516.

The Ninth Circuit reached the opposite result in Chabolla. There, a company called ClassPass sold packages that gave consumers access to gyms and fitness classes. Chabolla, 129 F.4th at 1151–52. The sign-up process included three screens. Id. at 1152. The court found the proposal on the first of these screens insufficiently conspicuous (and rejected the proposal on the other two screens on other grounds). See id. at 1155–60. The right of that first screen allowed users to sign up either by using an email address and clicking “Continue” or by clicking on a link to Facebook. Id. at 1152. Underneath the Facebook button, the screen told users: “By clicking ‘Sign up with Facebook’ or ‘Continue,’ I agree to the Terms of Use and Privacy Policy”:

Id. at 1152, 1174. A divided court reasoned that a user who signed up via email (rather than Facebook) would immediately click “Continue” and “read no further if she had no intention of using Facebook.” Id. at 1157. So that user might miss the proposal because it fell below the Facebook button. See id. The court distinguished Oberstein because Ticketmaster displayed its proposal “directly above or below” the action button. Id.

Question Three: Did the website operator use a font size or color that would draw attention to the proposal? Courts are more likely to find a proposal conspicuous if the operator uses a larger font or at least colored hyperlinks that draw attention to the proposal despite a “small font” for its terms. Meyer, 868 F.3d at 78; see Cullors v. Cerebral, Inc., 2024 WL 3385530, at *1 (9th Cir. July 12, 2024) (memorandum); Keebaugh, 100 F.4th at 1020; Oberstein, 60 F.4th at 515–16; Selden, 4 F.4th at 156–57. And they are more likely to find the proposal inconspicuous if operators use a small font with normal colors for the terms. See Berman, 30 F.4th at 856.

A comparison of Selden with Berman confirms this third point. Selden addressed Airbnb's sign-up page. 4 F.4th at 152–53. The page gave users three options to sign up: through Facebook, Google, or email. Id. Below the button for email sign-ups, it told users: “By signing up, I agree to Airbnb's Terms of Service” and other policies:

Id. at 152. The D.C. Circuit found this notice sufficiently conspicuous. Id. at 156–57. Apart from the simple design, the court emphasized that the hyperlinked terms of service stood out because they “appeared in red text against a white background[.]” Id. at 156.

The opposite was true in Berman. Apart from the clutter on the webpages in that case, the Ninth Circuit found that Fluent had used “a tiny gray font” that was “so small that it [was] barely legible to the naked eye.” Berman, 30 F.4th at 856–57. Fluent had also merely underlined the hyperlinked “Terms & Conditions” without changing the color. Id. at 854, 857. This fact made the decision to use tiny font even worse. Id. at 857. As the court explained, “[s]imply underscoring words” will typically not “alert” reasonable users that a “clickable link exists” at those words. Id.

Question Four: Did the website operator and users engage in the kind of interaction that one would expect to include contractual terms? See Keebaugh, 100 F.4th at 1020; Blizzard Ent., 292 Cal. Rptr. 3d at 61–62. If users would reasonably anticipate “some sort of continuing relationship” with an operator (such as when they sign up for an account), courts are more likely to find a proposal conspicuous because consumers would expect terms to govern the relationship. Meyer, 868 F.3d at 80; see Blizzard Ent., 292 Cal. Rptr. 3d at 61. But if users would not anticipate any relationship with the operator once they leave the site, courts are less likely to find the terms conspicuous because users would not expect contracts to govern that one-off visit. See Sellers, 289 Cal. Rptr. 3d at 26.

A comparison of the Ninth Circuit's decision in Keebaugh with the California Court of Appeals’ decision in Sellers shows this fourth (and final) point. Keebaugh addressed the free gaming app “Game of Thrones Conquest.” 100 F.4th at 1009–11. When users opened the app, they encountered one of two similar starting screens (depending on the year). Id. The screens contained background art, the game's title, and a blue “Play” button. See id. at 1010–11. Below the “Play” button, the app informed users that they would agree to the terms of use if they tapped it. See id. And below that offer, the operator included hyperlinked terms (including an arbitration provision) in an outlined box. Id. One screen looked like the following image:

Despite the proposal's small font size and light color, the Ninth Circuit held that it was sufficiently conspicuous. Id. at 1020–21. The court reasoned that users would expect a continuing relationship with an app operator—especially one that allows for in-app purchases— when they download the app even if they do not create any type of account and got the app for free. See id. at 1020.

The state court in Sellers reached a different conclusion. That court considered a website operated by “JustAnswer” that allowed consumers to ask questions of “experts” on several topics (ranging from medical to legal to scientific) for a fee. Sellers, 289 Cal. Rptr. 3d at 5–6. New consumers who typed in a question were taken to the payment page:

Id. at 6. The webpage suggested that consumers could get a 7-day trial for $5. Id. at 7. In much smaller font, it also included the following proposal: “By clicking ‘Start my trial’ you indicate that you agree to the terms of service and are 13+ years old.” Id. The hyperlinked terms included an arbitration clause, and JustAnswer automatically renewed all consumers who hit “Start my trial” beyond the 7-day trial. Id. The plaintiffs claimed that this automatic renewal violated California's Automatic Renewal Law, which required businesses to use large type (or similarly sized type in different colors) for the terms of use and to obtain affirmative consent for automatic renewals. See id. at 10–11. The court found that the proposal was not sufficiently conspicuous. See id. at 26–30. Critically, though, it reasoned that the Automatic Renewal Law's specific requirements for automatic renewals should apply to these contracts. See id. at 26–29. And when applying those standards, it held (among other things) that a party who paid $5 for a 7-day trial would not expect any lasting relationship beyond the trial period. See id. at 29–30.

B. LowerMyBills’ Offer and Dahdah's Acceptance

Applying these legal principles, we conclude that LowerMyBills and Dahdah formed a binding arbitration agreement. LowerMyBills’ website told users that if they click specific buttons, they would “consent” to its Terms of Use (including the arbitration provision). Viner Decl. Ex. A-1, R.11-3, PageID 96–97. But the website never explicitly showed the terms to users; instead, they had to click a hyperlink to view them. LowerMyBills thus proposed a hybrid offer (not a clickwrap or browsewrap offer). See Keebaugh, 100 F.4th at 1014; Oberstein, 60 F.4th at 516–17. So Rocket must establish that the LowerMyBills website gave “reasonably conspicuous” notice of its proposal to Dahdah under the “fact-intensive” test that applies to hybrid offers. Oberstein, 60 F.4th at 515. And it must then show that Dahdah engaged in conduct that objectively showed his assent. Berman, 30 F.4th at 856. We will address each element in turn.

1. LowerMyBills Gave Reasonably Conspicuous Notice

Whether the LowerMyBills website made a reasonably conspicuous offer is a close question. Indeed, this question split Michigan's own district judges. Compare Dahdah, 2023 WL 11944898, at *2–9, with Shirley v. Rocket Mortg., 2022 WL 2541123, at *5–8 (E.D. Mich. July 7, 2022). So one might think that a jury should decide whether a “reasonably prudent Internet user” would recognize that a click of the relevant buttons showed acceptance of the offer. Oberstein, 60 F.4th at 515 (quoting Berman, 30 F.4th at 856); see Domer v. Menard, Inc., 116 F.4th 686, 711–12 (7th Cir. 2024) (Hamilton, J., concurring in the judgment). After all, courts generally call on juries to provide the “fact-intensive answer” when a legal test turns on “how an ordinary person” would view an issue. Hana Fin., Inc. v. Hana Bank, 574 U.S. 418, 422, 135 S.Ct. 907, 190 L.Ed.2d 800 (2015).

But California decisions leave no doubt that this contract-formation question belongs to the court where (as here) the parties agree on the historical facts. See Blizzard Ent., 292 Cal. Rptr. 3d at 61; Long, 200 Cal. Rptr. 3d at 123–24. Unless and until the California Supreme Court says otherwise, then, we will treat it as a question of law (not a question of fact) just as other circuits have done. See Oberstein, 60 F.4th at 517. For what it is worth, both parties also assume that this fact-versus-law question raises a substantive issue dependent on state law under Erie. One could debate that assumption. Like the parties, at least one circuit court has suggested that “[w]hether an issue is a question of law or a question of fact is a substantive question, to which state law applies.” Encompass Ins. Co. v. Coast Nat'l Ins. Co., 764 F.3d 981, 984 (9th Cir. 2014). But another circuit has suggested the opposite—namely, that “the ‘fixing of the boundary between questions of law and questions of fact’ is a matter of federal procedural law, not state substantive law.” Beckner v. Maxim Crane Works, L.P., 109 F.4th 968, 972 (7th Cir. 2024) (per curiam) (citation omitted). Given the lack of briefing on this Erie question, we also decide the case based on the parties’ assumption that state law controls.

As for our answer to the legal question, recall that Dahdah visited LowerMyBills.com three times and that the website had a different format on the first and third visits as compared to the second. We opt not to decide whether the format on the second visit sufficed because a “reasonably prudent Internet user” would have seen LowerMyBills’ offer during Dahdah's first and third visits. Berman, 30 F.4th at 856. On those visits, Dahdah encountered the offer on the fourth page after inputting his personal information. The offer followed a green “Calculate” button and indicated: “By clicking the button above, you agree to be contacted by LowerMyBills, at the address entered above for promotional emails and consent, electronically via E-sign, to the LMB Lending Terms of Use, Privacy Policy, and Consent to Doing Business Electronically.” Viner Decl. Ex. A-1, R.11-3, PageID 96. Clicking the “Calculate” button then took Dahdah to the fifth page. Below a green “Calculate your FREE results” button, this page displayed the offer a second time. It indicated that “[b]y clicking the button above,” consumers would “express [their] understanding and consent” to (among other things) the Terms of Use. Id., PageID 97.

Did this double notification convey a “reasonably conspicuous” offer? Berman, 30 F.4th at 856. In our view, we must answer “yes” to stay consistent with the existing circuit precedent on this topic. To start, the proposal on the fourth page followed a “simple design” that did not contain much clutter (other than a logo for Quicken Loans as the “Featured Provider”). Selden, 4 F.4th at 156. In this respect, then, this page resembles the simple sign-up pages for Uber or Airbnb. See id. at 152, 156; Meyer, 868 F.3d at 78, 82. And it differs from Fluent's webpages in Berman, which contained other eye-catching images and information. 30 F.4th at 857. Admittedly, the fifth page had far more terms than the fourth page. It also identified Dahdah's consent to the specific “Terms of Use” in the second of four paragraphs of details. But we view this page as serving a belt-and-suspenders role for the fourth page's proposal. And we resolve this case based on the notice that consumers would have received across the pages in combination.

Next, the “spatial and temporal coupling” of the proposal with the key buttons made it even more conspicuous. Meyer, 868 F.3d at 79. LowerMyBills placed the proposal “directly” “below the action button” on each of the pages. Oberstein, 60 F.4th at 516. And it used a “dynamic scrolling function” in which these pages automatically scrolled down as users inputted information in the boxes. Appellant's Br. 59. So users would always see the offer on the same screen as the action buttons. See id. The LowerMyBills pages thus resemble Ticketmaster's website in Oberstein more than ClassPass's website in Chabolla. In Chabolla, the Ninth Circuit criticized ClassPass's website because its offer allowed users to acknowledge their consent using one of two action buttons for those who used email or Facebook to sign up. 129 F.4th at 1157. Yet ClassPass had placed the offer below the second of these action buttons (for Facebook sign-ups). Id. The offer thus sat “on the periphery” for anyone who signed up via email because these users might click the first button without looking below the second. Id. In contrast, LowerMyBills (like Ticketmaster) put its proposal immediately below the single action button.

To be sure, LowerMyBills’ decision to use a very small font for its offer on the fourth and fifth pages provides Dahdah's strongest counterargument. But that font size seems comparable to the offer on Uber's sign-up screen, see Meyer, 868 F.3d at 78, and may well be larger than the offer on the Game of Thrones app, see Keebaugh 100 F.4th at 1020–21. Like Uber and Ticketmaster, LowerMyBills also highlighted its Terms of Use with a hyperlinked “bright blue font” that contrasted sharply with the white background. Oberstein, 60 F.4th at 516; see Meyer, 868 F.3d at 78. And unlike Fluent, LowerMyBills did not hide the critical hyperlink using the same font color as the other text. See Berman, 30 F.4th at 857.

Lastly, although a close call, the offer arose in a context in which one would expect an “ongoing relationship” with LowerMyBills and its affiliate lenders rather than a one-off transaction. Keebaugh, 100 F.4th at 1020. Dahdah may have visited LowerMyBills.com simply “to get the calculated results” of refinancing options, Appellee's Br. 18, but it is also true that LowerMyBills.com matches users with one or more lenders so that the lenders can contact users in the future about mortgage refinancing using the provided contact information. Reasonable mortgagors thus would expect a further relationship. And reasonable users also would expect that the free referral service comes with some contractual strings attached (most obviously, that they consent to receive the communications for which they are signing up). So LowerMyBills provides a service that is more forward-looking than the one-off purchase of “a pair of socks” that one would expect would come with no other terms. Sellers, 289 Cal. Rptr. 3d at 16.

In short, the offer on LowerMyBills’ webpages more closely resembles the offers on the webpages that other circuit courts have found sufficient. As a result, we would “sow[ ] great uncertainty in this area” if we found its webpages insufficient to make a cognizable offer. Chabolla, 129 F.4th at 1172 (Bybee, J., dissenting).

Dahdah's contrary arguments lack merit. He starts by claiming that LowerMyBills’ proposal qualified as a typically unenforceable “browsewrap” offer rather than a “hybrid” offer enforceable on a case-by-case basis. Appellee's Br. 17. Not so. Browsewrap offers seek to create contracts when users simply browse a webpage (hence, their name). See Keebaugh, 100 F.4th at 1014; Blizzard Ent., 292 Cal. Rptr. 3d at 60. LowerMyBills did not propose that type of offer. It required users to take a specific step to accept its offer: click the relevant buttons. Dahdah responds that hybrid offers typically involve signing up for an account (such as Uber) and that he did not do so. But courts have treated offers as hybrids in other situations. Users, for example, could obtain the Game of Thrones app for free without creating an account. See Keebaugh, 100 F.4th at 1020. And the Ninth Circuit treated the offer as a “sign-in wrap” (i.e., a hybrid offer) because the app operator proposed that users would accept its terms of use by taking a specific action: clicking “Play.” See id. at 1014. This case resembles Keebaugh. Like the users in that case, Dahdah obtained a free item (referrals) without creating an account, but he did have to click buttons to obtain the free item. And LowerMyBills proposed that he would accept the terms of use by clicking those buttons.

Dahdah also argues that he had no “anticipation of a continuing relationship” because he only wanted “to get the calculated results” of refinancing options. Appellee's Br. 18. But given that the site matches users with potential lenders, we cannot say the objective user would fail to anticipate some sort of continuing relationship.

Switching gears, Dahdah criticizes visual aspects of the LowerMyBills website. He first asserts that the offers on the fourth and fifth screens of LowerMyBills’ referral process were not conspicuous because the three earlier screens had similar “Calculate” buttons without any offer requiring users to agree to the Terms of Use. Appellee's Br. 19–20. If anything, though, the visual contrast between the initial pages and the later ones would make the offer “more conspicuous,” not less. Reply Br. 8. Dahdah cites no case suggesting that operators must place the offer on every page in a multi-page sign-up process. The Uber process, for example, included two screens. See Meyer, 868 F.3d at 70–71. Although only the last screen included the requirement to accept Uber's terms, the Second Circuit still found the offer valid. Id. at 77–79.

Dahdah next raises concerns with the “small size of the font” that LowerMyBills chose. Appellee's Br. 20. He raises a legitimate concern. And as Judge Boggs noted in Oberstein, those who make borderline hybrid offers create “risks” for themselves by “invit[ing] second-guessing” by judges. Oberstein, 60 F.4th at 517. As we have said, though, other courts have enforced offers made in a similarly “small font” when the operator used blue hyperlinks to draw attention to the terms. Meyer, 868 F.3d at 78. Given the uncluttered fourth page and the use of hyperlinks, we likewise conclude that LowerMyBills made a sufficiently conspicuous offer despite the small font.

Dahdah also criticizes LowerMyBills for placing the offer below—not above—the critical action button during his first and third trips to the website (in contrast to his second visit). We do not find this difference critical. Other courts have enforced these offers when placed below rather than above the button that signaled the user's assent. See id.; Keebaugh, 100 F.4th at 1020–21; Selden, 4 F.4th at 156. As long as the website operator places the offer “directly above or below” the critical button such that it would visibly appear when the user clicked that button, this location factor supports a conspicuous finding. Oberstein, 60 F.4th at 516 (emphasis added).

Dahdah similarly argues that the critical “Calculate” button (along with the Quicken Loans ad at the bottom) qualify as clutter that “distracts” from the small-font offer. Appellee's Br. 22. Yet LowerMyBills’ fourth screen looks much more like the “simple, streamlined” pages from Uber and Airbnb than the cluttered pages from Fluent. Selden, 4 F.4th at 152, 157; see Chabolla, 129 F.4th at 1174; Berman, 30 F.4th at 859; Meyer, 868 F.3d at 81.

Dahdah fares no better with his reliance on Seneca v. Homeaglow, Inc., 2025 WL 852896 (9th Cir. Mar. 19, 2025) (memorandum). There, users met two different offers on different pages with different terms. Id. at *1. One set of terms included an arbitration provision, but the other did not. Id. The Ninth Circuit found these offers inconspicuous because reasonable users would not expect the terms to differ from page-to-page. See id. at *2. Moreover, the website owner presented the second set of terms after users had paid for the service when they were not expecting to see them. See id. at *1. Seneca thus looks nothing like this case.

Dahdah lastly suggests that we must at least remand for a trial because the record does not “indisputabl[y]” show that a reasonably prudent internet user would see the offer. Appellee's Br. 25. We agree with him that the question is a close one. But his logic mistakes a question of law for a question of fact. Under the California law that we must apply here, a court (not a jury) decides whether the undisputed website design gave reasonably conspicuous notice of the operator's offer. See Blizzard Ent., 292 Cal. Rptr. 3d at 61. We decide that it does.

2. Dahdah Accepted LowerMyBills’ Offer

As in Oberstein, the second requirement (that Dahdah took actions showing his assent to LowerMyBills’ offer) becomes “straightforward” once we conclude that the offer was reasonably conspicuous. 60 F.4th at 517. LowerMyBills told users that they would accept its Terms of Use if they “perform[ed]” a “specified act”: “clicking” the “Calculate” and “Calculate your FREE results” buttons. Restatement (Second) of Contracts § 30(1); Viner Decl. Ex. A-1, R.11-3, PageID 96–97. And Dahdah does not dispute that he clicked these two buttons. This conduct qualified as a “manifestation of assent to the terms” by Dahdah “in a manner invited ․ by” LowerMyBills. Restatement (Second) of Contracts § 50(1); see Oberstein, 60 F.4th at 517. The parties thus entered a binding agreement to arbitrate because LowerMyBills made a valid offer and Dahdah gave a valid acceptance.

Dahdah's contrary claims again fail. He suggests that he could not validly assent because the website did not give him “reasonable notice” that clicking the buttons would qualify as an acceptance of LowerMyBills’ Terms of Use. Appellee's Br. 28. This claim merely repackages his argument on the first element that LowerMyBills did not make a valid offer. Because he mistakenly treats this offer as inconspicuous, this argument fares no better on the back end.

Dahdah next compares his assent to the inadequate assent in Berman. There, the court held that users did not unambiguously assent to Fluent's terms. Berman, 30 F.4th at 857–58. But Fluent's website did not “explicitly notify” users of the specific action they must take to accept Fluent's terms. Id. at 858. Rather, the website simply said abstractly that users “agree[d]” to them. Id. at 854. The court held that Fluent could have “easily” fixed this problem if it had changed its website to inform users that they would agree to the terms “[b]y clicking” the relevant button. Id. Unlike Fluent, LowerMyBills undertook this easy fix because its website did state that Dahdah would agree to the terms by taking certain actions. See Oberstein, 60 F.4th at 517.

Dahdah lastly argues that he did not enter a contract because he did not “knowingly assent” to it. Appellee's Br. 29 n.5; id. at 2 n.2, 7 n.3, 9. To show his subjective mindset, he points to his declaration asserting that he had no actual knowledge of LowerMyBills’ offer. Under California contract law, though, his subjective knowledge did not matter. If a party engages in conduct that would reasonably lead the other side to believe that the party has accepted the offer, courts will enforce a contract even if the offeree did not intend to contract. See Blizzard Ent., 292 Cal. Rptr. 3d at 58; Windsor Mills, 101 Cal. Rptr. at 351. That rule rebuts this final claim.

III

Dahdah makes three other arguments apart from his claim that the parties did not exchange a valid offer and acceptance. First, Dahdah argues that the parties did not enter an enforceable arbitration agreement because LowerMyBills’ Terms of Use say nothing on too many material factors. Which arbitration organization should conduct the arbitration? How many arbitrators should decide the dispute? How should the parties choose these arbitrators? LowerMyBills’ terms provide no answers to these questions (among others). But the terms do say that the “proceedings” should be “governed by the Federal Arbitration Act (‘FAA’).” Terms, R.11-7, PageID 114–15 (capital letters omitted). And the Act fills in many “details” when an agreement does not. Green v. U.S. Cash Advance Ill., LLC, 724 F.3d 787, 792 (7th Cir. 2013). It allows a district court to “designate and appoint” an arbitrator if the contract identifies “no method” for doing so, and it sets a default of a “single arbitrator” when the parties do not contract for more. 9 U.S.C. § 5. Many circuit courts thus have held that judicial orders under this section can facilitate the arbitration when the parties use a “detail-free clause” that leaves most procedural aspects about the arbitration open. Green, 724 F.3d at 792-93 (collecting cases); see Arabian Motors Grp. W.L.L. v. Ford Motor Co., 19 F.4th 938, 941 (6th Cir. 2021). This caselaw shows that we generally may not invalidate an otherwise enforceable agreement for a purported lack of arbitration details.

Dahdah points to no contrary caselaw. He instead cites decisions holding that courts may not allow an arbitration if the arbitration would not permit the plaintiff to effectively vindicate statutory rights. See Morrison v. Cir. City Stores, Inc., 317 F.3d 646, 653 (6th Cir. 2003); Heckman v. Live Nation Ent., Inc., 120 F.4th 670, 685 (9th Cir. 2024). But Dahdah does not explain why an arbitration supplemented by the FAA's gap-filling provision would bar him from vindicating his rights under the Telephone Consumer Protection Act. So this argument lacks merit.

Second, Dahdah argues that the arbitration agreement did not cover Rocket's calls to him because it made the calls over a year after he visited LowerMyBills. Yet parties can agree to arbitrate not just the merits of their dispute but also “threshold arbitrability questions” about whether certain claims fall within their arbitration clause. Henry Schein, Inc. v. Archer & White Sales, Inc., 586 U.S. 63, 65, 139 S.Ct. 524, 202 L.Ed.2d 480 (2019); see Rent-A-Center, West, Inc. v. Jackson, 561 U.S. 63, 68–69, 130 S.Ct. 2772, 177 L.Ed.2d 403 (2010); First Options of Chi., Inc. v. Kaplan, 514 U.S. 938, 943, 115 S.Ct. 1920, 131 L.Ed.2d 985 (1995). If, then, an agreement unmistakenly sends this threshold arbitrability question to an arbitrator, courts must respect that choice even when they believe that the arbitration clause should not cover the dispute. See Henry Schein, 586 U.S. at 68–69, 139 S.Ct. 524. And the agreement here leaves this question for the arbitrator. It states that an arbitrator must resolve “the issue of arbitrability” and later adds that “[a]ny controversy concerning whether a dispute is arbitrable shall be determined by the arbitrator and not by the court.” Terms, R.11-7, PageID 114 (capital letters omitted). Dahdah does not even try to distinguish this language. He thus must save this arbitrability argument for the arbitrator.

Third, Dahdah argues that the parties terminated the agreement before Rocket called him. To support this claim, he relies on an order by the Federal Communications Commission suggesting that a party's consent to receive phone calls automatically ceases after three months under the Telephone Consumer Protection Act. See In re Rules & Regulations Implementing the Telephone Consumer Protection Act of 1991, 18 F.C.C.R. 14014, 2003 WL 21517853, at *39–40 (July 3, 2003). And Rocket's calls occurred more than three months after Dahdah visited LowerMyBills.com. Rocket responds that this FCC decision would not apply to the parties’ relationship here. But we need not decide that point because Dahdah's claim fails on other grounds. For one thing, Dahdah does not explain why the FCC's order covers the parties’ agreement to arbitrate (rather than their agreement that Rocket could call him). For another thing, the parties’ agreement also leaves this threshold arbitrability question (about whether an alleged termination took the dispute outside the arbitration clause) for an arbitrator rather than a court. See Berkeley Cnty. Sch. Dist. v. Hub Int'l Ltd., 130 F.4th 396, 404 (4th Cir. 2025).

All told, we reverse and remand for proceedings consistent with this opinion.

MURPHY, Circuit Judge.