Superhuman's homepage
Audited: the marketing copy on superhuman.com’s homepage, captured as text on 2026-07-03. The page is script-heavy, so this works from the words alone — logo rows and visuals may not have survived the capture, and the verdicts below say so where it matters.
What’s working — and it’s a lot:
- The page shows the product instead of describing it. The hero is a working exchange — an assistant noticing a meeting needs booking, offering times, taking “book it for monday” — followed by a triaged inbox and a doc mid-edit. That’s the strongest claim a software page can make: here it is, doing the job. Protect this.
- It names things. “Connect to 800+ tools instantly.” “Connect Gmail, Drive, Jira, and all of your favorite apps.” A count and named systems are checkable, which is what makes them believable. Same instinct in “Schedule meetings without leaving the conversation” — a specific capability a reader can picture and later verify.
- The respect record is clean. No countdown, no manufactured scarcity, no “don’t get left behind.” The calls to action are plain — “Get Superhuman,” “Learn more” — and the closing vision invites rather than pushes. On the respect axis this page barely needs an auditor.
Line by line
"The most productive email app ever made"
A superlative sized past what any page can back — "ever made" is a ranking against every email app in history, and no comparison appears anywhere on the page. Possible spin, not proven: there may be benchmarks behind it. But the claims directly beneath it are doing real work (speed, follow-ups, hours saved), and an uncashable crown sitting on top reads as decoration and taxes their credibility. Fix-direction: let the specifics carry the section — scope the superlative to something the page can show, or cut it and lose nothing.
"Fly through your inbox twice as fast as before … and save 4 hours every single week"
These are the best-shaped claims on the page — specific, countable, the kind a reader could actually doubt or believe. They're one citation short of solid: no source appears next to them on the page, and "every single week" universalizes a number that presumably came from a measurement with a scope. The neighboring "never drop the ball again" has the same problem at higher magnitude — "never … again" is an absolute no product delivers literally. Fix is a question: where do the 2x and the 4 hours come from — an internal study, user telemetry, a survey? Put that next to the number and this section becomes the strongest thing on the page.
"Trusted by the most innovative companies in the world"
Two claims fused: that notable companies use this (very plausibly true — a logo row may accompany this line that our text capture didn't preserve), and that they are the most innovative in the world, which is a ranking no logo row can substantiate. Possible spin only on the second half — the superlative, not the trust. Fix-direction: name the customers and, better, one real thing one of them does with the product; drop the world ranking. "Trusted by [named companies]" is a smaller sentence that's fully believable.
"Everyone's favorite AI writing partner"
"Everyone's" is a universal nobody measures and "favorite" is a verdict handed to the reader instead of earned from them. The block continues in the same register — writing that's "impossible to ignore," "let your brilliance shine" — feeling asserted rather than produced by anything shown. The irony is that Grammarly is the product on this page most likely to have enormous real numbers sitting in a drawer. Fix is retrieval, not writing: if a real user count or a named-institution figure exists, it replaces "everyone's" at a straight profit; if not, say what the tool does, which the next three lines already do better.
"AI that actually works in every app you use" / "Work with any app, from any app"
The page argues against itself here, in a useful way: its own bullet scopes the claim precisely — "Connect Gmail, Drive, Jira, and all of your favorite apps" — while the headline widens it to every app and any app, a universal one unsupported app falsifies. "Actually" is a small tell, too: an intensifier standing where evidence would go. Possible spin — if the integration layer genuinely covers arbitrary apps and tabs, the honest version is showing how. Fix-direction: the page's own precise phrasing is the fix; use the named list, or state the mechanism that earns "every."
"Before you know it, you have the time to be more creative, strategic, and impactful—free to do what only you can do."
The closing vision. As pressure it's clean — it invites, doesn't shame, and a reader can walk away freely, which is worth saying out loud. As a claim it floats: "creative, strategic, and impactful" are three words a skeptical reader can neither believe nor doubt, resting on the sections above for all their weight. Low severity — vision copy is allowed to be vision — but it's carried, not carrying. Fix-direction: one concrete "before/after" of a real person's week would give this paragraph a floor; if that story exists, it belongs here.
The read
Six verdicts, and the pattern is a single habit, not a character flaw: this page has real specifics — a 2x, a 4-hours, an 800+, named systems, a product shown actually working — and then crowns nearly every one of them with a universal the specific doesn’t need and can’t cash. “Ever made.” “Everyone’s.” “Every app.” “The most innovative companies in the world.” “Never again.” Each crown is possible spin at worst and dead weight at best, because the sentence underneath it was already the believable one. The respect axis, meanwhile, is genuinely clean — no fake urgency, no manufactured belonging — which is rarer than it should be. The fix here is mostly subtraction and one errand: keep the numbers and go get their sources, keep the names, cut the crowns.
Want this run on your page?
Send me the URL — you get the read and keep it either way, and yours stays private unless you want it out.
You own this page? If anything here is factually wrong, tell me and I'll correct it. If you'd rather it come down, ask. Every negative verdict above is an evaluation of the published words, offered as possible — not an accusation, and never a claim about intent.