Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Conversion Rate Optimisation › Landing Pages

Speed and Mobile for Landing Pages

How speed and mobile design affect landing pages: Core Web Vitals in plain terms, page weight and third-party tags, and mobile layout and thumb reach.
Guideintermediate11 min readUpdated

Why speed is a conversion factor

Speed is the first thing a visitor experiences and the only thing they experience until the page appears. A paid click that lands on a blank screen is a click the visitor may abandon before the headline loads; a page whose button moves as a banner drops in above it produces mis-taps; a page that takes a noticeable moment to respond when the button is tapped produces a second tap, a doubt, or a second submission. None of these appear in the design file. All of them appear in the conversion rate (CVR), and they appear most on the slow phones and weak connections that lab tests on a developer's laptop never show.

Speed also compounds. Every section of the page, every proof element and every form field is only useful to the people who are still there when it renders, so a slow page lowers the effective quality of everything on it. The order of work, then, is usually speed first and persuasion second, because a persuasive page nobody waits for converts no one. The landing pages hub puts the speed checks in its review checklist alongside the structural ones for that reason.

Google documents page experience as a set of signals its ranking systems use, with Core Web Vitals among them, and also says that good content with weaker page experience can still rank. The Core Web Vitals guide covers that ranking side, the Search Console report and the template-level fixes; this guide is about what the same measures mean for a visitor who has already clicked.

Core Web Vitals in plain terms

Web.dev documents Core Web Vitals as three metrics, each measured from real users and each with a documented threshold for good, needs improvement and poor, assessed at a documented percentile of page loads so that the slower experiences count rather than the average. The table says what each one means for a landing page.

MetricWhat it measuresWhat the visitor feelsUsual causes on landing pages
Largest Contentful Paint (LCP)How long until the largest visible element, typically the hero image or headline, has rendered"Is anything happening?" — the wait before the page is usable at allOversized hero images; images lazy-loaded that should not be; slow server response; render-blocking scripts and styles; web fonts that delay text
Interaction to Next Paint (INP)How long the page takes to visibly respond to taps, clicks and key presses, observed across the whole visit"Did that register?" — a tap on the button or a field that seems to do nothingHeavy JavaScript on the main thread; third-party tags running during interaction; form validation or analytics work on every keystroke
Cumulative Layout Shift (CLS)How much visible content moves unexpectedly while the page loads"Where did the button go?" — a tap that lands on the wrong thing because the page movedImages and embeds without reserved dimensions; cookie banners and announcement bars inserted above content; fonts swapping late; content injected by tags

Two properties of the measurement are worth repeating from the documentation. The metrics are assessed from field data, meaning real visitors on real devices, with lab tools used to find causes rather than to judge status. And the assessment uses a percentile that represents the slower loads, so a page that is fast on average and slow for a sizeable minority is rated by the minority. For a landing page that minority is often the mobile paid traffic the campaign is paying most for.

The simulator below shows how changes to each metric move a page between the documented bands, using the thresholds from the cited sources rather than any figure of this page's own.

Core Web Vitals simulator

Page weight and third-party tags

Most landing page slowness has two sources: what the page itself ships, and what other people's scripts add to it.

Page weight is the bytes the browser has to download before the page is usable. The usual offenders on landing pages are hero images and videos exported at desktop size and served to every device, fonts loaded in several weights that the page uses one of, and page-builder templates that include the styles and scripts of every component whether the page uses it or not. The fixes are well documented: images sized to the viewport that will show them, in a modern format, with dimensions declared so the layout does not shift and the one above the fold loaded eagerly; video that does not autoplay on mobile; one or two font weights with a loading strategy that shows text immediately; and templates stripped to the components the page uses.

Third-party tags are the scripts added by marketing: analytics, ad platform pixels, chat, heatmaps, session recording, personalisation, consent management, testing tools. Each one is small; together they are often the largest part of the page, and because they run on the main thread they are the usual cause of poor interaction responsiveness. A tag manager (GTM) makes them easy to add, which is both its value and its risk. The GTM vs hard-coded tags comparison covers who should own which tags and why; for a landing page the operational rules are: an inventory of every tag with an owner and a purpose, removal of the ones nobody can justify, a loading order that puts conversion measurement first and everything optional after the page is interactive, and a review whenever a conversion rate drops without a visible change, because a new tag is a frequent invisible one.

Testing tools deserve a specific mention. A client-side testing script that hides the page until it decides which variant to show adds a delay to every visit, including the visits in the control group; the delay is a cost of the test that the test's own results do not show. Server-side or edge-based variation avoids it, and where a client-side tool is used, the delay before the page is revealed should be as short as the tool allows. The guide to choosing between client-side and server-side testing sorts planned tests by the change each makes, so that only light changes that pass a flicker check stay in the browser.

Mobile layout and thumb reach

Most landing page visitors from paid social and a large share from paid search arrive on a phone, held in one hand, often outdoors or in motion, with a thumb that reaches the bottom of the screen comfortably and the top with difficulty. The layout has to work in those conditions, not in a desktop browser narrowed to a phone's width.

The practical rules follow from the thumb. The primary call to action is reachable: either in the lower part of the first screen or in a bar that stays at the bottom as the visitor scrolls. Tap targets are large and separated, so a thumb lands on the intended one; the Web Content Accessibility Guidelines document a minimum target size and spacing that is a reasonable floor. Nothing important depends on hover, because there is none. Text is readable without zooming, with line length and contrast set for a small bright screen in daylight. The first screen contains the headline, the offer and the request, since that is all many visitors will see.

Layout stability is part of mobile design, not only a metric. Reserve space for every image, embed, banner and notice before it loads; never insert content above what the visitor is reading except in response to their own action; and keep the cookie notice from pushing the page down after it appears. A shifting layout on a phone is the direct cause of accidental taps on the wrong element, which on a landing page means a wrong link, a lost scroll position or an abandoned form.

Finally, the mobile page is usually a different page from the desktop one in all but URL: different hero crop, shorter copy, proof placed in the screen with the button, navigation removed. Designing the mobile layout first and expanding it for desktop produces a better result than the reverse. The anatomy guide describes the sections; on a phone, their order and density change.

A business weighing an app against a better mobile site can use the guide to deciding between an app and the mobile website, which explains why the mobile page still serves every first-time visitor once an app exists.

Forms on phones

A form is where mobile speed and layout problems concentrate, because it is where the visitor has to interact rather than read. The rules in the forms guide apply in full; the mobile-specific ones are these.

Input types drive the keyboard: an email field that opens the email keyboard, a phone field that opens the number pad, a numeric field that does not open the full keyboard. Autofill works when the fields are named in the standard way, and on a phone it is the difference between typing an address and tapping once. Fields are full width and stacked, one column, with labels above, so that the keyboard, which covers the lower half of the screen, does not hide the label of the field being filled. The button sits where the thumb is and is not covered by the keyboard when the last field is focused.

Validation runs when the person leaves a field, not on each key press, because validating on every keystroke on a phone adds script work during typing, which is exactly what the responsiveness metric measures, and produces error messages while the person is still mid-entry. Error messages appear next to the field and the page scrolls to the first error on submission, since on a phone the error may be off screen.

Multi-step forms earn more on mobile than on desktop: one or two questions per screen, the button always visible, progress shown, contact details captured early so that an abandoned later step still leaves a lead. A long single-page form on a phone is a wall the thumb has to climb.

Illustrative example — not a real result

A travel company finds the tag that cost a quarter of its leads

A tour operator's enquiry page converts 4.8% of mobile visitors. One Monday the rate drops to 3.6% with no deploy on the page. Real-user data shows mobile interaction responsiveness worsening from a median of 180 milliseconds to around 420, and layout shift rising as a new announcement bar inserts itself above the hero after load. The tag inventory shows two additions the previous Friday: a session-recording tool loading at page start, and a personalisation script that injects the bar. Moving the recording tool to load after the page is interactive and reserving the bar's height in the template bring responsiveness back to about 190 milliseconds and remove the shift; the conversion rate returns to 4.7% within the week. The team adds a rule that any new tag goes through a performance check on a mid-range Android phone over a throttled connection before publication.

Measuring real-user speed

Speed for conversion is measured on the visitors who actually arrive, which means field data rather than lab tests. Web.dev documents the sources: the Chrome User Experience Report, which aggregates real Chrome users' measurements and feeds the field sections of Google's tools and the Search Console report; and the option of collecting the same metrics from the site's own visitors with the documented web-vitals JavaScript library and sending them to an analytics tool. The second option is the useful one for landing pages, because it can be segmented by the dimensions that matter: campaign, device, connection, country, and whether the visit converted.

Lab tools — Lighthouse, the lab section of PageSpeed Insights, the browser's performance panel — simulate a load on one device and one connection and are for finding causes: which image is the largest element, which script blocks rendering, which element shifts. They are not a status report. A page that scores well in the lab and poorly in the field is usually a page whose real visitors have slower phones and connections than the lab assumes.

The practical loop: read field data by landing page and device, pick the page with the most paid traffic and the worst metric, reproduce the problem in the lab on a throttled mobile profile, fix the cause in the template or the tag configuration, and confirm in the field data over the following weeks. Alongside the metrics, watch the behaviour they predict: bounce from paid traffic by device, form starts over page views, and the conversion rate by device, because those are the numbers the work was for.

Paid social turns page speed into a media cost, and the Meta Ads setup for website traffic treats the in-app mobile load as part of the campaign itself.

When a landing page template fails layout stability in Search Console, the diagnosis on layout shift that is too high names the banner, embed, font or script block that moves the page.

Where phones convert far below desktops month after month, the diagnosis for a mobile conversion rate far below desktop separates speed from layout, checkout, cross-device journeys and traffic mix.

To put a value on speed work in conversions, the page speed impact test splits visitors on the server between the current build and an optimised one, with field speed recorded in each arm.

On the search side, the page experience guide reviews a template against the rest of Google's questions, from secure connections and mobile display to overlays and ad placement.

Common mistakes

  • Testing on the office network and a new laptop. The paid visitor is on a phone in a car park.
  • A hero image at desktop size for every device. The largest element is slow for everyone.
  • Lazy-loading the hero. The element the loading metric waits for is told to wait.
  • Tags added without an inventory. Each is small; together they are the page.
  • Testing tools that hide the page. The reveal delay is a cost in the control group too.
  • A cookie banner that pushes the page down. Layout shift and a mis-tap on the first screen.
  • Hover states that carry meaning. There is no hover on a phone.
  • Validation on every keystroke. Script during typing is what the responsiveness metric measures.
  • Reading lab scores as status. Field data from real visitors is what the visitor experienced.
  • Fixing speed without checking the conversion rate moved. The metric is the proxy; the behaviour is the goal.

Frequently asked questions

Which Core Web Vital matters most for a landing page?

The one that is failing for the visitors the campaign pays for, read in field data by device. In practice loading, measured by Largest Contentful Paint, is the most common failure on landing pages because of oversized hero media and heavy templates, and it is the one that costs visitors before they see anything. Layout shift matters most on pages with banners and notices inserted above the first screen, and responsiveness matters most where tags and validation run during interaction.

Why does the page pass in a lab test but fail in field data?

A lab test simulates one device on one connection, usually a better one than the campaign's real visitors have. Field data, as web.dev documents it, comes from real users and is assessed at a percentile that represents the slower loads, so a page that is fast for most and slow for a sizeable minority is rated by the minority. Paid mobile traffic is often that minority. The lab is for finding causes; the field is for knowing the status.

Should landing pages use a lighter template than the rest of the site?

Often, yes. A landing page needs a logo, a headline, media, proof, a form and a footer, and it does not need the site's menu system, search, chat widget or every tag in the container. A stripped template that shares the site's styles and policies but loads only what the page uses is faster and more stable, and it can be built once and reused for every campaign page. The trade-off is maintenance, which is why it should be a template rather than a one-off.

Terms used on this page

Where this page is used

Part of these learning paths

Also related

Sources

  1. Web Vitals — web.dev
  2. Understanding page experience in Google Search results — Google Search Central

Related