Landing page examples for startups: six pages, section by section

The six startup landing pages we read in October 2026 (Linear, Raycast, Cal.com, Plausible, Loom and Superhuman) share four first-screen jobs: a headline that says what the product does, one primary call to action (CTA), proof that others use it, and the product itself. Unbounce put the median landing page conversion rate across all industries at 6.6% and treats 10% or above as good.

What makes a good startup landing page

A good startup landing page does four things in the first screen: it says what the product does in one clear headline, it asks for one action through a single primary call to action (CTA), it shows proof that other people use it, and it shows the product itself. Everything else on the page supports one of those four jobs or gets cut.

Start with the numbers that set the bar. Unbounce reported that in Q4 2024, it analyzed 41,000 landing pages with 464 million visitors and 57 million conversions, and that the median conversion rate across all industries is 6.6%. Medians by industry range from 3.8% to 12.3%, so check your own category before judging a page. The same source treats 10% or above as "good" in general.

Here is a worked example, not a benchmark. If 1,000 visitors reach your page, the 6.6% median would produce 66 sign-ups, and the 10% bar would produce 100. In our reading, that gap of 34 sign-ups from the same traffic is what one cleaner page can be worth.

In our reading, the second number explains why short pages tend to win. Nielsen Norman Group found that users have time to read at most 28% of the words during an average visit; 20% is more likely. Visitors spent only 4.4 seconds more for each additional 100 words, and they read half the text only on pages with 111 words or less. That study used browsing data collected in 2005, so treat it as a direction for page design.

Element	Job on the page	What to check this week (our reading)
Headline	Says what the product does and for whom	A stranger can repeat it back after one read
Primary CTA	Asks for one next step	Every button leads to the same action
Proof	Shows others already trust you	Logos, a named quote or a usage figure sit near the CTA
Product visual	Shows the thing, not a stock photo	The screenshot matches what users see after sign-up

On CTAs, Unbounce's guidance is direct: if you're using multiple CTAs, align them towards the same goal. In our reading, this is where many early pages lose people: a demo button, a newsletter box and a free trial all compete for one visitor.

To act on this, measure your current rate (conversions divided by visitors), cut the first screen to the four elements above, and test one change at a time using a growth experiment process. If the CTA is a trial or freemium sign-up, track what happens next against free-to-paid conversion benchmarks.

What sections should a SaaS landing page have

We read every page in this piece against the same seven parts. Each part has one job, and a page can be strong in one and weak in another, so scoring them separately tells you what to fix first.

Part	Its job (our reading)	What we check on each page
Headline	Say what the product does for whom	Could a stranger repeat it after one glance
Subhead	Add the how, or the main outcome	Does it answer the first doubt the headline raises
Primary call to action (CTA)	Give one next step	Is the button visible without scrolling, with a verb on it
Proof	Show that others already use it	Are logos, numbers or quotes specific and attributable
Product visual	Show the thing itself	Is it a real screen, not stock art
Objection handling	Answer price, setup and switching worries	Are the top three objections named and answered
Closing CTA	Catch readers who scrolled to the end	Does it repeat the primary CTA's offer

On the CTA itself, one format detail recurs in the examples. Unbounce described a page where, "Whether you consider this a SaaS page or an ecommerce page, the marketers at Amazon made the right call to use a click-through CTA instead of embedding a form on the page." SaaS here means software as a service.

Why the top band and left edge matter most

Users first read across the upper part of the content area, then across a shorter second line, and then scan the content's left side in a vertical movement. The firm said a study 11 years later confirmed these findings.

The same study found readers read the third word on a line much less often than the first two words. Unbounce made a related point about depth: with the average scroll depth only being about 50%, the most important material belongs in the top half.

In our reading, that puts the headline, subhead and primary CTA in the top bar of the F. Proof and objection handling sit in the stem, where only the opening words of each line get seen.

A five-minute scan test

Run this on your own page this week:

1. Cover everything below the first screen. Check that the headline, subhead and CTA alone explain the offer.
2. Read only the first two words of every subhead and bullet. Rewrite any that start with "Our," "We" or "The."
3. Confirm the closing CTA uses the same words as the primary one.
4. Mark each of the seven parts present or missing. Fix the missing ones before tuning the rest.

Linear and Raycast, read section by section

When we read both pages in October 2026, each hero did two jobs in order: say what the product is, then show it running. Most of what followed was proof.

Linear

The headline read "The product development system for teams and agents" (Linear). The subhead set the scope in two short sentences: "Purpose-built for planning and building products. Designed for the AI era." (Linear).

The hero visual was a working issue, not an empty dashboard. Its activity feed showed "Triage Intelligence added the labels Performance and iOS" and ended with "Changed 2 files. Draft PR awaiting your review" (Linear). In our reading, the feed turns the claim about agents into a short sequence of events a visitor can follow without reading feature copy.

Proof came late and stacked. A changelog entry dated Oct 22, 2026 showed recent shipping (Linear). Then a count: over 40,000 product teams (Linear). The closing block offered four buttons: Get started, Contact sales, Open app and Download (Linear).

Raycast

The headline was "Your shortcut to everything." (Raycast). The main button, Download for Mac, carried one line under it: "macOS Tahoe and Apple Silicon required" (Raycast). We think that line stops a visitor on the wrong machine before a wasted download, so the clicks that do happen are from people who can install.

The visual was the launcher itself, under the line "One interface, everything you need." (Raycast). Community counts came later, a Slack of 37k members and 90k followers on X (Raycast). The page closed on the same button and "Download and use Raycast for free." (Raycast).

Element	Linear	Raycast	What it may do (our reading)
Headline	Category plus who it serves	One benefit	Tells a visitor in one line whether to keep reading
Hero visual	An issue moving through agent steps	The launcher with commands listed	Shows the product working, not a feature list
Main action	Four buttons at the close	One download button with a requirement under it	Matches the action to how each product is adopted
Proof	Changelog, named quotes, team count	Crash-free rate, community size	Backs each adjective with a fact

Copy this on your page this week

1. Write the headline as a category plus its user, or as one benefit. Pick one.
2. Under the main button, add any requirement that would stop someone after the click.
3. Replace a static hero screenshot with a short sequence that ends in finished work. It shows time to value before signup.
4. Pair every adjective in your copy with one measured number you can defend.

Cal.com and Plausible, read section by section

Both pages answer the first doubt right under the button. Cal.com's hero, below the headline "The better way to schedule your meetings", offers two sign-up buttons with "No credit card required" beneath them (Cal.com). Plausible pairs "Start free trial" with "View live demo" and sets "30-day free trial · No credit card required" under both (Plausible).

Cal.com: remove the setup worry

The hero buttons read "Sign up with Google" and "Sign up with email" (Cal.com). In our reading, the Google option answers a second objection without saying it: no new password to create.

Next comes a three-step "How it works": "Connect your calendar", "Set your availability" and "Choose how to meet" (Cal.com). Step one names the fear outright: "so you don't have to worry about double bookings."

The benefits block answers price: "Unlimited and free for individuals." Paid tiers appear in the FAQ, with Teams at $16/user/month (Cal.com). Testimonials then name competitors, as in "More elegant than Calendly, more open than SavvyCal."

Plausible: name the incumbent

Plausible puts the comparison in the headline: "Easy to use and privacy-friendly Google Analytics alternative" (Plausible). The first section states the problem as "Google Analytics is overkill for most site owners." Proof follows in the same terms: "Our script is 54 times smaller than Google Analytics."

The stats strip lists 22k paying subscribers and 99.99% uptime over the last 90 days (Plausible). A founders' letter adds that the company is a "team of 10, running since 2018." One testimonial ties back to the trial line: "Became a paying customer within 1hr of the 30-day trial."

Objection (our reading)	Cal.com's answer	Plausible's answer
Will this cost me now	No credit card required	30-day free trial, no card
Is setup work	Three numbered steps	Live demo button
Why switch from what I use	Testimonials naming Calendly	Headline naming Google Analytics

A template to run this week

1. List three doubts your sales calls or support tickets raise before sign-up.
2. Answer the top one in under ten words, directly beneath the primary button.
3. Answer the second in a numbered "How it works" block of three steps.
4. Answer the third by naming the tool your buyer uses today, then give one proof figure you can stand behind.

Decide the trial terms before you write the line under the button; our guide on freemium vs free trial covers the choice. Then test the new microcopy as a single change with the method in how to run a growth experiment.

Loom and Superhuman, read section by section

Loom: the button first, then the claim

Loom's page opens with two buttons, "Get Loom for free" and "Install Chrome Extension" (Loom). The headline sits under them: "Unstuck your team with Loom" (Loom). It is five words, and it names a problem.

The first section heading is a scale claim: "Millions of people across 400,000 companies choose Loom" (Loom). The free button repeats right below it, so a reader who is convinced by the number has a place to click.

The testimonials carry full job titles. One is credited to a "Chief People Officer, HubSpot" and another to "Customer Success, Pearson" (Loom). The Pearson quote gives a concrete gain: "It has saved us hundreds of hours by creating informative video tutorials" (Loom). In our reading, the titles do part of the work, because a buyer looks for someone in their own role.

Superhuman: one product, then a menu of jobs

Superhuman's headline is "Bring AI inside every app and tab with Superhuman Go", followed by two buttons, "Try Go" and "Learn more" (Superhuman). The same pair repeats under the next section.

The use cases sit in tabs: "Featured; Prep for meetings; Write and edit; Automate your emails; Build agents" (Superhuman). Each card names one job and the feature that does it. A visitor picks their own task instead of reading every card.

The mail section leads with a time claim: "Save 4+ hours every week on email and scheduling" (Superhuman).

Element	Loom	Superhuman	What to copy (our reading)
First thing seen	Two buttons	Headline, then two buttons	Put the CTA in the first screen
Proof	400,000 companies, titled quotes	Tabbed jobs, a time claim	Match proof to the buyer's role or task
CTA repeats	Under the scale claim	Under the next section	Repeat the button after each proof block

Test it this week

Move your primary button above your headline on one page, and run it as a split test (how to run a growth experiment). As an example, not a benchmark: if 1000 visitors land and 100 click, the click rate is 10 percent. Compare that against your current layout over the same traffic.

For a tabbed block, use this template per tab: the job in the user's words, the feature name, one line on what happens. Keep it to five tabs, as Superhuman does, and order them by how often support hears the request. A short path from click to first result shortens time to value.

What the six pages have in common, in one table

The table sets Cal.com, Plausible and Loom side by side. Linear, Raycast and Superhuman are read in their own sections above, so their rows are left out here rather than repeated.

Page	Headline length (our reading)	Primary CTA wording	Button pattern (our reading)	Proof type	Product visual	Objection line	Closing CTA
Cal.com	One line, names the category	Sign up with Google, with email as the second option	Get started paired with Book a demo under most sections	Logo band "Trusted by fast-growing companies around the world", a user count (over a million people), named testimonials	A booking page in the hero	"Most scheduling apps limit free users heavily"	"Smarter, simpler scheduling" with Get started and Talk to sales
Plausible	One line, names the tool it replaces	Trial button, quoted above	Trial button paired with a demo button, repeated on pricing	Named customers: Stripe, Mozilla, MongoDB, Harvard University and Sentry, named testimonials	Live demo link	"No need for cookie banners or GDPR consent"	"Ready to ditch Google Analytics? Start your free trial today"
Loom	Short, names an outcome	Free download button, quoted above	Action verbs such as Record now further down	Company count, named testimonials	Hero video	None on its own line, the quotes carry it	An enterprise Learn more, then testimonials

Plausible also answers lock-in directly: "Our code is open source too, so you're never locked in."

What the CTA should say

In our reading, all three primary buttons name either the action or the price, never the company. Cal.com puts the sign-up method in the button. Plausible puts the trial in it. The closing CTA on Plausible repeats the headline's argument as a question and then the button. We think that pattern is the one to copy: the last line restates the switch the visitor is making.

A template you can use this week: verb + what they get + cost. "Start free trial", "Sign up with Google", "Record now" all fit the first two parts. Put the cost line directly under the button.

Whether to show customer logos

Plausible names customers in a sentence and backs them with quotes such as "We switched over everything from GA." In our reading, logos alone signal size, and a named quote with a job title answers fit. Ship both if you have them.

Testing it

As an example, not a benchmark: send 1000 visitors to each of two button wordings. If version A gets 20 sign-ups, that is 2% of 1000. If version B gets 50, that is 5% of 1000. Run it as a growth experiment with one change at a time, then watch whether the extra sign-ups reach activation.

How long should a landing page be

In our reading, length is the wrong question. The eyetracking work behind the F-pattern says that position on the page decides what gets read, and length mostly decides how much gets skipped.

In the original Nielsen Norman Group study, the team recorded how 232 users looked at thousands of Web pages. Users read across the top, then across a shorter second band, then scanned down the left side. The same study found that users read the third word on a line much less often than the first two, and that the first two paragraphs must state the most important information.

The finding held up. The pattern was identified in 2006, and the team revisited it eleven years after discovering it. Its practical advice was blunt: assume that most users will be scanning.

Reading research puts a number on the scanning. On the average page, users have time to read at most 28% of the words during an average visit; 20% is more likely. Each extra block of copy buys little attention: users spend only 4.4 seconds more for each additional 100 words. Users read half the information only on pages with 111 words or less.

A worked example

Take a page of 1000 words, as an example rather than a benchmark. At the 20% rate, a visitor reads about 200 words of it. You do not get to choose which 200. The F-pattern does, and it picks the top two bands and the first words down the left edge.

So a long page is fine if those 200 words carry the argument. A short page fails if its headline is vague and its call to action (CTA) sits on the right of the second row.

The rule we apply

Put three things where scanning starts. We check every page against this before any copy edit.

Where the eye lands	What goes there (our reading)
Top band	The claim: what the product does, for whom, in one line
Second band	The CTA and the line under it that removes the first doubt
Left edge, first two words of each line	The objection line and section heads, starting with the noun that matters

Then run the squint test. Read only the first two words of each line down the left side. If those words alone tell a visitor what you sell and what to click, the page can be as long as it needs to be. If they do not, cutting words will not fix it; moving them will. Test the change with a single variant, using the method in how to run a growth experiment.

What is a good landing page conversion rate, and how to audit your page this week

Start with the benchmark from the first section: the all-industry median for landing pages. It is a median, so a few outlier pages do not drag it around, and it measures one thing. Conversion rate is conversions divided by visitors for one page and one action, such as a signup or a demo booking. Pick that single action before you measure anything, or the number will not mean much.

A worked example

These are example inputs, not a benchmark.

Your page had 1,000 visitors last month and 50 signed up, so it converts at 5% (50 divided by 1,000). Say your target is 10%, which on 1,000 visitors is 100 signups. The gap is 50 signups (100 minus 50).

You will not close that gap with one change. Spread the 50 extra signups across 5 fixes and each fix needs to add 10 signups. On 1,000 visitors, 10 extra signups is 1 percentage point. That is the size of lift to plan a test around. Check whether your traffic can detect it with the A/B test sample size calculator before you start.

The seven-row audit

Copy this table into a doc and score your own page: 1 if the row passes, 0 if it does not. The middle column points back to pages we read earlier in this piece.

Section	Passes when	Pages that pass (our reading)	Your score (0 or 1)
Hero headline	A stranger can say what the product does after reading it	Loom, Superhuman
Primary call to action	One clear action above the fold	Loom
Product visual	Shows the product doing its job, not a mood image	Cal.com
Logo band	Names customers the reader recognizes	Cal.com, Plausible
Testimonials	Named people with roles and a specific result	Loom
Objection handling	Answers the objection a buyer raises on a sales call	Cal.com, Plausible
Closing call to action	Repeats the main action at the bottom of the page	Cal.com, Plausible

A score under 7 gives you your list of fixes.

Test one change at a time

1. Write each 0 row as a hypothesis: the change, the page section and the lift you expect.
2. Score the hypotheses with ICE (impact, confidence, ease) and add them to the growth experiment backlog template.
3. Ship the top one alone. Change nothing else on the page until the test reaches the sample size you calculated.
4. Record the result, keep or revert, then move to the next row.
5. Watch what happens after signup. A page that converts more visitors who never reach activation has not gained you much.

What this page does not cover

This page does not cover where landing page traffic comes from, pricing page design or onboarding after signup beyond pointing to activation. The six pages were read in October 2026 and may have changed since. Nothing here is legal advice, including on privacy, cookie banners or consent requirements.

Related on Race to Scale: Pricing page examples from six SaaS companies and what each one does; Growth hacks that don't work anymore, and the platform rules that ended them; A/B test sample size and duration calculator, with the formula.