How to Design a Hero Section That Does Not Slow the Page
A joiner I helped last year had a homepage that looked excellent in the designer's mockup. A full width photograph of a finished kitchen, a headline over the top in white, and a button. On my laptop it loaded in under a second.

A joiner I helped last year had a homepage that looked excellent in the designer's mockup. A full width photograph of a finished kitchen, a headline over the top in white, and a button. On my laptop it loaded in under a second. On his own phone, standing in a customer's driveway with two bars of signal, the headline arrived after four and a half seconds and the photo a second later. He had been sending that link to every enquiry for eight months.
The photo was 2.4 MB. That single file was heavier than every other asset on the site put together. Nothing about the layout was wrong, and nobody had done anything stupid. The hero section is simply the one part of a page where design choices and loading time collide, because the biggest visual element is also the first thing a visitor waits for.
What a hero section is actually for
The hero is the band at the top of a page, above the point where most people start scrolling. Its job is narrow: tell a stranger what this site is, give them one thing to do, and provide one reason to believe it. That is three pieces of information, not six.
Most people decide whether a page is relevant in two or three seconds of looking. In that window they will read a heading of about eight words, glance at one image, and notice one button. Everything else in the hero is decoration competing with those three items. I have watched session recordings where a visitor scrolled straight past a beautifully set subheading of forty words without their eye pausing once.
So the content list is short. A heading that names the thing you do and where you do it. One line of supporting text, fifteen words at most. One button with a verb in it. One proof element, which can be a star rating, a client logo row, or a single sentence of plain fact such as the year you started. If you cannot fit those four things comfortably, the problem is the heading length, not the available space.
The full viewport height habit is worth dropping
Design tutorials have repeated the same instruction for a decade: make the hero one hundred percent of the viewport height, so it fills the screen on arrival. I used to build them that way. I now think it is wrong for almost every small business site, and the reason is specific rather than aesthetic.
On a phone, the usable viewport is already reduced by the browser address bar and the home gesture area. A hero set to 100vh therefore fills the screen exactly and leaves not a single pixel of the next section visible. Visitors who are unsure whether the page continues sometimes do not scroll at all. On a short iPhone in portrait that gives you roughly 650 logical pixels for four content items, so the heading shrinks or the button slides under the fold.
What works better is a hero around 55 to 70 percent of viewport height on mobile, with the top edge of the following section deliberately visible. That sliver of the next block does more for scroll rate than any animated arrow. On desktop I cap hero height at about 600 pixels regardless of screen size, because a 1440 pixel tall hero on a large monitor is mostly empty background.
The background choice is the expensive one
Every hero style has a weight and a loading consequence. This is the table I use with clients before anyone opens a design tool, because the decision is easier to make in advance than to undo later.
| Hero style | Typical weight | Effect on load | Reasonable when |
|---|---|---|---|
| Flat background colour, text only | 0 KB | Largest paint is text, so it renders almost instantly | Services, consultancies, anything where words sell |
| CSS gradient or pattern | Under 2 KB | Same as above, no extra request | You want visual interest without a photo |
| One product or portrait photo beside the text | 40 to 120 KB as WebP | Small and predictable, easy to size properly | Most small sites. This is my default |
| Full bleed photograph behind the text | 150 to 400 KB done well, 2 MB done badly | Becomes the largest contentful paint, needs preloading | Venues, hotels, anything where the place is the product |
| Background video loop | 1 to 6 MB | Competes with everything for bandwidth, often blocked on mobile data | Rarely. Use a still frame and link the video |
| Rotating carousel of three or more slides | Three times a single image | Loads slides nobody sees and shifts layout as it initialises | Almost never. Pick the best slide |
If you do choose a full bleed photograph, three things make the difference between 200 KB and 2 MB. Export it at the real display width rather than the camera's output, which usually means 1600 pixels wide for desktop and a separate 800 pixel version for phones. Serve it as WebP or AVIF. And preload it in the page head so the browser starts fetching before it has finished reading your stylesheet. Our guide on compressing images for a faster website covers the export settings in detail.
Do not lazy load the hero image. Lazy loading is excellent for everything below the fold and actively harmful at the top of the page, because it tells the browser to deprioritise the one image your visitor is waiting for.
Text over a photograph, and the contrast trap
White text on a photograph reads beautifully on the two or three images the designer tested and badly on the fourth. Skies are the usual culprit. A heading that sits over dark trees in the mockup ends up over bright cloud when the client swaps the picture in spring.
The fix is to stop relying on the photograph. Put the text on a solid or near solid panel, or lay a consistent overlay across the whole image, somewhere around 40 to 55 percent black for white text. Then check the result against a contrast ratio of at least 4.5 to 1 for body text and 3 to 1 for large headings. A fixed overlay means the client can change the photo later without quietly breaking readability, which they will do, usually on a Friday.
Give the image container an explicit width and height, or an aspect ratio in CSS, so the browser reserves the space before the file arrives. Heroes are the single most common source of jumping content on small sites, and the mechanism is explained in our notes on fixing cumulative layout shift.
How to tell whether yours is good enough
Open your own homepage on a phone, on mobile data rather than home wifi, from a cold cache. Count the seconds until you can read the heading. Under two and a half seconds is fine, three to four is borderline, and anything past five is costing you enquiries. Then do the harder test: show the top of the page to somebody who has never seen the site and ask what the business does. If they hesitate, the heading is the problem, not the picture.
The quickest improvement available to most sites this week is to delete the carousel and keep one slide. It removes two image requests, the slider script, and the layout shift as the first slide initialises, and in my experience nobody ever notices the other slides are gone. More layout decisions of this kind sit in our web design section.
FAQ
How tall should a hero section be?
Around 55 to 70 percent of the viewport on mobile, and capped at roughly 500 to 600 pixels on desktop. Leave the top of the next section visible so visitors can see the page continues.
Does a hero section help SEO?
Indirectly. Search engines read the heading in it, which is usually your H1, and page speed is a ranking signal, so a heavy hero can hurt you. The hero itself is not a ranking factor.
Should the hero image be the same on mobile and desktop?
Use the same picture but a different crop and a smaller file. A wide landscape shot loses its subject entirely when squeezed into a portrait phone screen, so a square or vertical crop usually reads better.
Is a video background ever a good idea?
Only where the motion is the product, such as a dance studio or a machining workshop, and only with a static poster image that displays while the video loads. Even then I would keep the file under 1.5 MB and mute it.
More in Web Design
Web Design
How to Design a Blog Post Layout People Actually Read
A reader wrote to me in September with a problem that sounded like an SEO problem and was not one.
Web Design
How to Design a Call to Action Button People Actually Click
Your page gets visitors, they scroll, and then they leave without booking, buying or signing up.
Web Design
Mobile Friendly Website Design That Works on Any Phone
Your website looks clean on your laptop, but on a phone the text is tiny, the menu overlaps the logo and visitors pinch and...