Most Published Terms Would Not Bind Anyone
Online terms are contracts, judged under ordinary state contract law rather than any special internet doctrine. Two elements are examined: reasonably conspicuous notice that the terms exist, and an unambiguous manifestation of assent to them. Both have to be present. Notice without assent fails; assent without notice fails.
Most corporate marketing sites publish terms of use as a footer link and nothing more. Nobody was shown them, nobody agreed to them, and no record exists of either. On the usual analysis that is not a contract, however carefully the document was drafted.
Two federal statutes are often cited here and settle less than people expect. E-SIGN, 15 U.S.C. 7001 and following, and UETA in most states, provide that a record is not denied legal effect solely because it is electronic. Neither supplies a formation rule, and neither says whether a particular layout gave notice. Because this is state contract law, the analysis can differ between states on the same facts, and outcomes are fact-specific down to font size, contrast, position and device.
A Spectrum, Not Three Boxes
The familiar labels are judicial shorthand rather than statutory categories, and courts increasingly treat them as points on a spectrum. The shorthand is still useful, ordered here from strongest to weakest.
Clickwrap requires an affirmative action whose specific meaning is agreement, most commonly an unchecked box next to "I agree to the Terms of Service" that the user has to check before proceeding. This is the most reliably enforced form, because it produces both notice and an unambiguous act.
Sign-in-wrap is the dominant pattern on modern business sites. The user takes an action that is not itself an act of agreement — Sign Up, Create Account, Continue, Download, Get the Guide — next to text stating that by doing so they agree to the linked terms. Outcomes here are genuinely mixed and depend on execution.
Browsewrap makes the terms available by hyperlink, usually in the footer, and asserts that use of the site constitutes agreement. There is no affirmative act of assent at all. It is the weakest form, and the one most corporate marketing sites actually use.
Why a Footer Link Generally Fails Both Tests
The problems with browsewrap are structural, not editorial, and they cannot be fixed by improving the document.
There is no affirmative act. Continuing to read a page is not conduct that unambiguously signals agreement to terms the reader has not seen. There is, in the ordinary case, no actual notice either: a link in a footer, below the fold, among a dozen others, is not something a court will presume a user saw. That leaves constructive notice as the only route, and it requires a presentation conspicuous enough that a reasonably prudent user would be on inquiry notice that terms exist and govern the visit. A grey link in small type at the bottom of a long page generally does not clear that bar.
One factor cuts modestly in favour of business sites. The sophistication of the party can matter, and courts have been more willing to find inquiry notice where the counterparty is a repeat commercial user, a business with prior dealings, or someone with demonstrated actual knowledge of the terms. That is a point worth having, not a foundation to build on.
The Design Decisions That Decide It
Where sign-in-wrap is unavoidable, and on most gated content and trial signups it is, the factors that improve the position are interface decisions rather than legal ones.
- Proximity. The notice sits immediately adjacent to the button, not at the foot of the page or in another column.
- Legibility and contrast. Readable size, sufficient contrast against the background, not grey on grey at the smallest type size in the design system.
- Position. Visible without scrolling on the relevant device, which means testing on mobile rather than assuming the desktop layout represents what people see.
- Link affordance. The hyperlink to the terms visually obvious as a link, through colour, underline or weight, rather than styled like body text.
- Explicit wording. Language that names the act and its effect: by clicking Create Account, you agree to the Terms of Service and Privacy Policy. Not a passive statement that terms apply.
The stronger option, wherever the friction is tolerable, is to make the user do something. That is why the usual recommendation for anything that matters — account creation, trials, purchases, API keys, gated downloads, community registration — is clickwrap rather than a better-worded notice.
What Rides on Formation
Formation is antecedent to everything else. If no contract formed, none of the provisions companies care about exists.
Arbitration clauses and class-action waivers. The Federal Arbitration Act, 9 U.S.C. 2, makes written arbitration agreements valid, irrevocable and enforceable save upon such grounds as exist at law or in equity for the revocation of any contract. The federal policy favouring arbitration does not engage until a court concludes that an agreement to arbitrate was formed, and formation is decided under state contract law.
Limitation of liability. Damages caps are generally enforceable in commercial contracts, subject to unconscionability and public-policy limits. A negotiated master services agreement governs paying customers. The website terms govern everyone else — anonymous visitors, prospects, downloaders, free-tool users, trial users, forum participants. A company with a strong MSA and a footer-link browsewrap has covered the small group and left the large one uncovered.
Choice of law and forum. Same dependency: an unformed contract selects nothing. Where formation succeeds these clauses are widely, though not universally, enforced in commercial settings, and a choice-of-law clause does not automatically displace every statutory claim a plaintiff might bring under their own state's law.
Proving What a Particular User Saw
This is the part companies discover last. Enforcing a clause means proving formation, and proving formation means showing what a specific user was shown on a specific date and what act they took.
The record has several parts. Versioned terms, each with an effective date, retained rather than overwritten when the document is updated. A record of which version was displayed on which pages during which period, because a company that has revised its terms four times cannot otherwise say which text a user encountered in a given month. And an assent log tying an individual event to a user: what was clicked or checked, when, from which address, with which user agent.
Most content management systems overwrite a page rather than version it, and most signup flows record that an account was created without recording what the screen said at the time. Dated screenshots of each version of each assent interface are a cheap partial substitute. These records are generally worth keeping for at least the longest limitations period that might apply.
Modification Clauses, and the Marketing Site Versus the Product
Almost every set of terms says the company may change them at any time and that continued use constitutes acceptance. It is weaker than companies assume, for a reason that follows from everything above: it depends on the same notice and assent the original terms depended on. Where the original was a footer link nobody saw, a unilateral change is a change to a document nobody agreed to — the same defect twice over. Even where the original was properly formed, the argument that a user assented to a revision they were never shown runs into the absence of any manifestation of assent to the new text. The stronger practice for material changes is to notify existing users and obtain renewed affirmative assent.
A public marketing site has almost no mechanism for forming a contract with a reader, and terms published there are best understood as a statement of position. A product, portal, trial signup or gated download has an interaction to attach assent to, a user identity to record it against. Companies routinely spend their drafting effort on the first and their design effort nowhere, when the second is where an enforceable agreement is actually available to them.
If It Becomes a Dispute
When enforceability is contested, the fight is evidential before it is legal. Counsel needs to establish what the signup page or download form actually displayed on the date the user in question interacted with it: the wording next to the button, the size and contrast of the notice, whether the link was visually distinguishable.
Reconstructing that means working from archived captures, deployment and release history, content management revision records, front-end code as it stood on the date, and whatever assent or consent logs survive. Authenticating the reconstruction, and explaining how it was produced, is the kind of examination an expert witness is engaged to carry out.
The recurring failure is that the record has already gone. Pages are overwritten rather than versioned, staging environments are rebuilt, and log retention on the systems holding assent events is often measured in weeks. Preserving the interface and the assent records as they change, rather than trying to recover them after a claim arrives, is what makes the question answerable at all.
Frequently Asked Questions
Does anyone actually agree to the terms of use on our website?
On a typical corporate marketing site, almost certainly not. Terms behind a footer link, with no point at which a visitor takes any action indicating agreement, generally fail both elements courts examine: reasonably conspicuous notice and an unambiguous manifestation of assent.
But it is a weak foundation for arbitration, a class waiver or a liability cap, and where those matter the usual advice is to attach them to a moment where the user does something.
Is a checkbox enough to make our terms enforceable?
An unchecked box that the user checks, next to a clear statement of what checking it means and a visible link to the terms, is the most reliably enforced pattern. Whether any particular implementation holds up is fact-specific, turning on details as granular as type size, contrast, position and device.
Two things separate a strong implementation from a weak one. The box starts unchecked, so the user performs the act. And the company keeps a record of the event.
Can we change our terms of use whenever we want?
Companies commonly publish a clause saying so, and it carries less weight than the drafting suggests. A change depends on the same notice and assent as the original, and continued use of a site by someone never shown the revision is a thin basis for saying they accepted it.
For material changes the stronger practice is to notify existing users and obtain fresh affirmative agreement, and to keep every prior version with its effective date.
What records should we keep to prove someone agreed to our terms?
Three things, kept together. Every version of the terms with its effective date, retained rather than overwritten. A record of which version was displayed on which pages during which period. And an assent log tying the individual event to a user: what was clicked or checked, when, from which address and user agent.
Dated screenshots of each version of the signup interface are a useful supplement.