A homepage passes the three-second test when a stranger can say what you sell, who it is for and what to do next from the first screen, on a phone. Rate it on 10 observable questions covering the H1, audience, one primary action, visible proof, concrete offer, navigation and Largest Contentful Paint at 2.5 seconds or less. Fix the lowest-rated question first.
The rule
The Stranger Standard
Judge the first screen by what a stranger takes from it in a few seconds on a phone, not by what you know is on the page. If a fact requires scrolling, clicking or prior knowledge to find, it does not count.
Self-check · about 4 minutes
Homepage three-second test
Answer all 10 questions about the first screen of your homepage, on a phone. Rate what a stranger would see, not what you know. Each question is worth 0 to 3 points, for a maximum of 30.
0OF 30
0 of 10 answered
Your result
Fails the three-second test
A stranger probably cannot tell what you sell or who it is for, so most visitors leave before they see your proof. Do not redesign yet. Fix the lowest-rated question first, starting with the H1 and audience.
The first screen has some of the right parts but leaves the visitor guessing on at least two of: service, audience, action or proof. Work through the priorities in the order of your weakest answers.
The message mostly lands, but one or two weak spots, often proof, phone layout or speed, are costing you visitors. Fix every question that got 1 or lower, then retest with five strangers.
Your first screen communicates clearly and loads fast on a phone. A pass does not predict conversions, so move to measurement: track the primary action and compare against your prior period.
Fails the three-second testReadable but vagueStrong, with leaksPasses the test
Your result
10 questions to go.
Your result, a short reading of it and your weakest areas appear here when you finish.
0 of 10 answered
1
2
3
4
5
6
7
8
9
10
Where you stand, weakest first
1. Does the H1 name the service you sell?
2. Does the first screen name who the service is for?
3. Can a stranger tell you what you do after five seconds on the first screen?
4. Is there one clearly primary action on the first screen?
5. Is proof visible without scrolling?
6. Is the offer concrete: what the buyer gets, in what time frame, at what price or scope?
7. Is the first screen written in plain, objective language, without unproven superlatives?
8. On a 390 by 844 pixel phone screen, are the H1, one proof element and the primary button all visible without scrolling?
9. Is Largest Contentful Paint on mobile 2.5 seconds or less at the 75th percentile?
10. Is the navigation focused, with specific labels and no more than seven top-level items?
Fix in this order
Weakest answer first, then up.
1
Move 1 of 6
Fix the H1 so it names the service and the audience
You are here when: Questions 1 and 2 got below 3, or a stranger guessed your business wrong in the five-second test.
Why now
Nothing else on the page matters until a visitor knows they are in the right place. NN/G's homepage guidance says the homepage should work as an elevator pitch, and that a concise tagline should explicitly convey what the site does.[1] Users also decide fast: a large dwell-time analysis found users often leave within 10 to 20 seconds, and that communicating value within 10 seconds is the way to earn minutes.[3] So this fix comes first, because every other fix is wasted on a visitor who left at second five.
Do this
Write the H1 as: [service] for [named audience], with an outcome if you can prove it.
Remove slogans, greetings and category labels from the H1 and move any slogan below it.
Run the stranger test: show the first screen for five seconds, hide it, and ask what you sell and who it is for.
Repeat with five people. The H1 stays in revision until four can answer both questions.
2
Move 2 of 6
Reduce the first screen to one primary action
You are here when: Question 4 got below 3, or your first screen has more than one button competing for attention.
Why now
One action comes second because a clear H1 with no obvious next step still loses visitors, and because multiple equal buttons split attention you have only seconds to hold. NN/G's homepage guidance advises making high-priority tasks stand out through visual hierarchy.[1] Our rule: if two buttons look equally important, neither is important. A single primary action also makes your measurement clean, because you can count one thing.
Do this
Pick the one action that most often leads to revenue, such as booking a call or requesting a quote.
Give it one button style, one label that says what happens next, and repeat it in the header.
Demote other actions to text links or move them below the fold.
Check on a phone: the primary button must be visible without scrolling.
3
Move 3 of 6
Put proof in view next to the headline
You are here when: Question 5 got below 3, or your proof, such as client logos, reviews or numbers, sits below the first screen.
Why now
Proof comes third because it only works once the visitor knows what you do and what to click. NN/G's 2018 eyetracking study of 120 participants found 57 percent of page-viewing time fell above the fold, and 74 percent within the first two screenfuls.[2] The first screen therefore gets the most attention, but not all of it, so proof placed below the first screenful still gets read by those who stay. Put at least one proof element where the first look lands.
Do this
Choose one proof element you can verify: a client count, a review rating with the source, a certification or a dated result.
Place it within one screen of the H1, not in a footer strip.
Name the source and the date of any rating or number.
Do not use logos you lack permission to show, and do not use invented counts.
4
Move 4 of 6
Make the offer concrete
You are here when: Questions 6 and 7 got below 3, or the first screen uses promotional adjectives and no price, scope or time frame.
Why now
A concrete offer comes fourth because it needs a clear audience and action to land on. The Morkes and Nielsen study found objective text scored 27 percent better on usability than promotional text in a 51-person sample from 1997. That is old, but it fits NN/G's later advice to show specific examples instead of generic category labels.[4][1] Specifics also help readers decide whether the offer is for them, which reduces unqualified enquiries.
Do this
State what the buyer gets, in what time frame, and starting at what price or scope where you can.
Replace adjectives with numbers: delivery days, project size, response time.
Show a specific example of the offer, such as a sample deliverable, instead of a category name.
Get Largest Contentful Paint to 2.5 seconds on mobile
You are here when: Question 9 got below 3, or PageSpeed Insights field data shows LCP above 2.5 seconds on mobile.
Why now
Speed sits fifth because a faster slow page is still a slow bad page, so fix what the visitor reads first. But once the content is right, load speed decides whether it gets seen. web.dev defines a good LCP as 2.5 seconds or less, measured at the 75th percentile of page loads and segmented by mobile and desktop.[5] Google research cited by NN/G found 53 percent of mobile visits ended when a page took longer than 3 seconds to load.[7] That study dates from 2016, so use it as a direction, not a benchmark.
Do this
Check field data in PageSpeed Insights for your homepage on mobile and note the LCP element.
Add fetchpriority="high" to the hero image if it is the LCP element, and never lazy-load it.[6]
Make the LCP image discoverable in the initial HTML, and serve it at the display size in a modern format.
Aim for the web.dev guideline split: about 40 percent time to first byte, under 10 percent load delay, about 40 percent load duration and under 10 percent render delay.[6]
6
Move 6 of 6
Focus the navigation
You are here when: Question 10 got below 3, or your menu has many equal-weight links and generic labels.
Why now
Navigation comes last because it is rarely what makes a stranger leave in the first seconds. It shapes what happens once they stay. NN/G advises putting primary navigation in a highly noticeable place and using specific link labels rather than generic ones.[1] Our rule: keep the top level to seven items or fewer, which is our judgment, not a studied limit. Each item should answer a question a buyer actually has.
Do this
List your top-level items and cut those no buyer clicks, using analytics click data.
Rename generic labels such as Solutions to what they contain, such as Website Redesign.
Make the primary action a distinct button in the header.
Test on a phone: can a first-time visitor find pricing and contact in two taps?
Decision finder
Your result, your situation, your first fix
Lowest-rated area
H1 and audience
What you see
Visitors guess your business wrong
First fix
Rewrite the H1 as service plus audience
Skip for now
Hero photography, animation
Primary action
What you see
Several equal buttons on the first screen
First fix
Choose one action and demote the rest
Skip for now
A/B testing button colors
Proof
What you see
Reviews and logos sit below the fold
First fix
Move one verifiable proof element beside the H1
Skip for now
Building a full case study library
Concrete offer
What you see
Adjectives, no price, scope or time
First fix
Add time frame and starting price or scope
Skip for now
Long explainer text
Phone speed
What you see
LCP above 2.5 seconds in field data
First fix
Fix the LCP element, hero image and server time
Skip for now
Micro-optimizing footer scripts
Navigation
What you see
Too many links with generic labels
First fix
Cut to seven or fewer, rename specifically
Skip for now
Mega-menu redesign
01
Evidence
Why the first screen decides so much
The first screen gets the largest share of attention on a page. NN/G's eyetracking study of 120 participants and more than 130,000 fixations found 57 percent of viewing time above the fold in 2018, down from 80 percent in a 2010 study. It also found 74 percent within the first two screenfuls.[2] The trend is that attention is spreading down the page. It is not that the first screen stopped mattering.
Time is the other constraint. A 2011 analysis of more than 2 billion dwell times found users often leave in 10 to 20 seconds, with the highest risk in the first 10.[3] The same analysis found that users who pass about 30 seconds become more likely to stay for minutes. So the first screen has to earn the next 20 seconds, not close the sale.
People also scan before they read. NN/G's original reading research found 79 percent of test users always scanned new pages, and only 16 percent read word by word.[8] A first screen written for scanning puts the service, the audience and the action in the words that get seen first.
02
Evidence
How the points work and why they are defensible
Each of the 10 questions is worth 0 to 3 points, for a maximum of 30. Every question is observable: you check it with a stranger, a phone or a free tool, not with your own opinion. The help text under each question says how to check honestly, and we recommend you check with someone who does not work for you.
The points are equal-weighted on purpose. It would be easy to invent weights, but we have no study that proves one is worth more than another, so we do not pretend to. The fix order comes from dependency instead: a visitor who cannot tell what you sell never reaches your proof, and a slow page may never show your headline. Our rule is to fix the lowest-rated question first, and break ties in the order the steps appear on this page.
Two questions rest on thresholds you can verify. LCP at 2.5 seconds or less is web.dev's published good threshold, at the 75th percentile and segmented by device.[5] The seven-item navigation limit and the stranger test pass mark of four in five are our judgment, which is why we label them as rules and not as findings.
03
Evidence
Test on a phone, and measure with field data
Most buyers meet your homepage on a phone, so the phone is the test. Open the page on a real device on mobile data, not on your office wifi. Check whether the headline, the proof and the button fit on the first screen without scrolling, because a desktop layout that passes can fail on a 390 pixel wide viewport.
For speed, look at field data first and lab data second. web.dev says to measure LCP at the 75th percentile of real page loads, segmented by mobile and desktop.[5] A lab score on your fast laptop can be misleading, and it will not tell you what your actual visitors experienced. If you have little traffic, field data may be missing, and then a lab test on throttled mobile is the best available estimate.
When LCP is the problem, find the element first. It is often the hero image, a video poster, or a large block of text. The web.dev optimization guide advises setting fetchpriority="high" on the likely LCP image, never lazy-loading it, and making it discoverable in the initial HTML.[6] Slow server response is the other common cause, and it is a hosting or caching fix, not a design one.
04
Evidence
What this test does not tell you
A high result does not guarantee conversions. The test measures whether the first screen communicates clearly and loads quickly, which are necessary conditions, not sufficient ones. Pricing, offer, trust and the rest of the page still decide whether a clear first screen converts. We do not claim a result predicts a specific lift.
Some of the evidence is old, and we say so. The 1997 writing experiment used 51 participants on one travel site.[4] The 2011 dwell-time analysis predates mobile-first browsing.[3] The 2016 mobile speed figure is cited second-hand through NN/G.[7] Each supports a direction that newer sources, such as the 2018 eyetracking study and the current web.dev guidance, are consistent with.
Finally, a test of your own site is a single measurement. If your traffic is mostly branded search, your visitors already know what you sell and the stranger test may overstate the problem. Run it anyway, because new visitors, referrals and AI search referrals arrive without that context.
Scroll through
5 mistakes that cost the most
01
Rating your own page
You know what the page means, so you will pass questions a stranger would fail. It costs you a result that is too high and a fix list that is too short. Fix: run questions 1 to 3 with five outsiders, and use a real phone.
02
Treating a lab speed score as the answer
A lab score on a fast connection hides what your visitors experience, so you can pass the test and still fail mobile users. It costs you the visitors on slow connections. Fix: use field data at the 75th percentile, segmented by mobile.
03
Fixing the easy question instead of the lowest one
Changing a button color feels productive while the H1 stays vague. It costs weeks of effort on the wrong lever. Fix: start with the lowest-rated question, and break ties in the order of the steps on this page.
04
Adding proof you cannot verify
Invented client counts or unsourced ratings may lift a result and then damage trust when checked. Fix: show one proof element with its source and date, or write nothing.
05
Retesting only on desktop
A first screen that fits on a laptop can push the button below the fold on a phone. It costs you the majority of your visitors. Fix: check every question on a 390 by 844 viewport before you call a fix done.
Worked example
The rule applied
A fictional plumbing company rates its homepage
1. The situation
Illustrative: "Meridian Plumbing" is an invented plumbing company. Its H1 reads "Quality service you can trust". The first screen has three buttons, no visible proof, and a mobile LCP of 4.1 seconds. The numbers are invented to show how to read the result.
2. The decision
The owner gives 1 on service, 0 on audience, 1 on primary action and 0 on proof. Offer gets 1, objective language 1, phone fold 2, LCP 1, navigation 3 and the stranger test 1. The total is 11 of 30. The weakest answers are the H1 and audience, so those are fixed first: "Emergency and repair plumbing for homes in one city, with a quote in 30 minutes". Then one primary button, then one verified review count, then the hero image.
3. What it shows
The logic is that the H1 fix stops strangers leaving at second five, the single button gives the next step, and proof gives a reason to believe. The total should climb out of the 10 to 17 band once those three fixes land, and the LCP work follows. Whether calls increase is measured by comparing the next 60 days with the previous 60, not assumed from the result.
Questions people ask
01
What is the three-second test for a homepage?
It is a check of whether a stranger can say what you sell, who it is for and what to do next after a few seconds on the first screen. NN/G found users often decide quickly, with many leaving within 10 to 20 seconds.[3] Our version rates the answer on 10 observable questions and ends with a fix order, so you can defend the result instead of arguing about taste.
02
What is a good result on this test?
A total of 26 to 30 out of 30 is a pass, but a perfect total is not required. Each question is worth 0 to 3 points and the points are equal-weighted by design, because we have no evidence that one criterion is worth more. Aim to have no question at 0 or 1 before you polish anything that got 2.
03
What counts as above the fold on a phone?
It is what a visitor sees without scrolling on their device. We test on a 390 by 844 pixel viewport as a common phone size, which is our choice, not a standard. NN/G's 2018 study found 57 percent of viewing time fell above the fold on desktop monitors, so what is visible first still gets most attention.[2] Check on a real phone.
04
How fast should my homepage load on mobile?
Aim for a Largest Contentful Paint of 2.5 seconds or less. web.dev defines good LCP as 2.5 seconds or less, needs improvement as 2.5 to 4.0 seconds, and poor as above 4.0. It measures at the 75th percentile of page loads, segmented by mobile and desktop.[5] Use field data from real visitors where you have it, and test mobile separately from desktop, because the two results often differ and web.dev recommends segmenting them.
05
Does one primary action really beat several?
In our experience, which is not a study, yes, because a first screen with equal buttons gives a visitor no obvious next step. NN/G advises using visual hierarchy so high-priority tasks stand out.[1] You can still offer secondary paths as text links or lower on the page. Test it: after the change, count how many visitors click the single button against the old total across buttons.
06
Do I need to rewrite the whole homepage?
No. Fix the first screen first. The first screen carries 57 percent of viewing time in NN/G's 2018 study, and the first two screenfuls carry 74 percent.[2] If you rewrite the rest before the top works, most visitors will not see it. Once the result is in the top band, the messaging order applies to the rest of the page.
The evidence behind this page
Sources
Evidence freshness
8 sources, 3 dated in the last 3 years
Each bar counts sources by the date the source itself shows.
2
1
1
1
1
2
979901030507091113151719212325
Cite or share this page
Quote the rule or the order with credit and a link. Suggested citation:
Adnan Gourija (2026). The Homepage Three-Second Test: Rate Your Above-the-Fold in 10 Questions. The Marketing Field Manual, Webtec. https://www.trywebtec.com/resources/field-manual/homepage-three-second-test/
“Webtec is fantastic and you should hire them. Over the years, we have worked with many companies - and many of them good at what they do. The thing that really separates Webtec from others however is that Webtec really knows how to anticipate the needs of their clients, and then provides the necessary support needed - such that it always looks like magic.”
Daniel DroiletGoogle · client since 2021
“Been a webtec client for a while now and so far they're the best in what they do, they redesigned my website, improved my seo which has skyrocketed traffic to the website...I highly recommend their services.”
Nassim JokoGoogle · client since 2020
“Extremely professional team-- very consultative on how to develop a modern web interface. Excellent support for design and graphics. Met each timeline and was very responsive to edits and updates.”
Pat FlanaganDesignRush
“We are a small non-profit but Webtec design has treated us like a fortune 500 Company. Always responding promptly to any needs that may arise along with top notch customer service and sales.”
BJay SBark · client since 2023
“Great support and experience from these guys. I was apprehensive about using a web developer, but they proved me wrong. They continue to support me with any website requests and they resolve things within 24 hours, which is great for a business like mine.”
Moeed AminGoogle · client since 2019
“Wow, just wow! I am so impressed with WebTec's work on my website redesign. They took my vision and turned it into a stunning reality. The team was communicative, creative, and incredibly professional.”
BWilliam TBark · client since 2022
“We have used Webtec to build and maintain our website. We couldn't be more pleased with the speed and quality that they were able to deliver. We are in the process of refreshing our content on our website and the Webtec team has been very responsive and affordable! We would recommend them highly.”
Betsy RomeriGoogle · client since 2019
“I had a great experience working with Webtec. The team was knowledgeable, responsive, and professional throughout the entire process. They took the time to understand my business needs and created a website that exceeded my expectations. Good job guys!”
BEvelinaBark · client since 2023
“Webtec took the time to listen and understand what my business was truly about and came up with the creative solutions to help me showcase it.”
Rita GubalisGoogle · client since 2021
“Webtec is fantastic and you should hire them. Over the years, we have worked with many companies - and many of them good at what they do. The thing that really separates Webtec from others however is that Webtec really knows how to anticipate the needs of their clients, and then provides the necessary support needed - such that it always looks like magic.”
Daniel DroiletGoogle · client since 2021
“Been a webtec client for a while now and so far they're the best in what they do, they redesigned my website, improved my seo which has skyrocketed traffic to the website...I highly recommend their services.”
Nassim JokoGoogle · client since 2020
“Extremely professional team-- very consultative on how to develop a modern web interface. Excellent support for design and graphics. Met each timeline and was very responsive to edits and updates.”
Pat FlanaganDesignRush
“We are a small non-profit but Webtec design has treated us like a fortune 500 Company. Always responding promptly to any needs that may arise along with top notch customer service and sales.”
BJay SBark · client since 2023
“Great support and experience from these guys. I was apprehensive about using a web developer, but they proved me wrong. They continue to support me with any website requests and they resolve things within 24 hours, which is great for a business like mine.”
Moeed AminGoogle · client since 2019
“Wow, just wow! I am so impressed with WebTec's work on my website redesign. They took my vision and turned it into a stunning reality. The team was communicative, creative, and incredibly professional.”
BWilliam TBark · client since 2022
“We have used Webtec to build and maintain our website. We couldn't be more pleased with the speed and quality that they were able to deliver. We are in the process of refreshing our content on our website and the Webtec team has been very responsive and affordable! We would recommend them highly.”
Betsy RomeriGoogle · client since 2019
“I had a great experience working with Webtec. The team was knowledgeable, responsive, and professional throughout the entire process. They took the time to understand my business needs and created a website that exceeded my expectations. Good job guys!”
BEvelinaBark · client since 2023
“Webtec took the time to listen and understand what my business was truly about and came up with the creative solutions to help me showcase it.”
We use only the cookies this site needs to work. If you say yes, we also turn on Google Analytics to see what works, and our live chat, which records how you use the site so we can help you faster. Nothing optional runs until you choose. See our cookie policy.
