Landing Pages for Paid Traffic: Message Match, Page Speed, Form UX, and A/B Testing That Holds Up
You pay for every click, so the landing page decides how much of that spend turns into leads or sales. John
Williams builds landing pages from the ad backward: the page says what the ad promised, loads fast on a phone,
asks only for what sales needs, works for people using assistive technology, and gets tested properly.
8.4% more retail conversions per 0.1 s faster mobile speed (Deloitte)
24×24 px WCAG 2.2 minimum target size
51% of top home pages had unlabeled form inputs (WebAIM 2026)
TL;DR
Message match first: the headline repeats the promise in the ad and the query that triggered it. A mismatch
here costs more than any button color.
Speed is measured on real phones: LCP within 2.5 seconds, INP at or under 200 ms, CLS at or under 0.1 (web.dev).
Forms ask for the minimum. Baymard found most checkouts need about 8 fields but averaged 11.3 in 2024 (Baymard). Lead forms follow the same logic.
Pages meet WCAG 2.2 AA, which also tends to make forms easier for everyone.
Tests get a sample size before launch and run for full weeks. If traffic can't support a valid test, we say
so and improve the page another way.
Who this is for
Advertisers sending Google, Microsoft, or Meta traffic to a homepage or a generic service page.
Lead-gen businesses whose cost per lead is rising while click costs hold steady.
Teams with a "below average" landing page experience status on important keywords.
Marketers who've run A/B tests that "won" and then didn't hold up.
What's included
Message-match map. Every ad group or audience matched to a page and a headline, so each click
lands on the promise it responded to.
Page design and build. On your CMS (WordPress, Webflow, Shopify) or a dedicated landing page
tool, with reusable sections your team can edit.
Form design. Field-by-field review with sales, labels and errors done properly, autocomplete,
and spam protection that doesn't block real people.
Performance and accessibility. Core Web Vitals checked on field data where available, and a
WCAG 2.2 AA review of each template.
Tracking. Conversion tags, consent mode, and lead IDs passed to your CRM so offline outcomes
can flow back to the ad platforms.
Test plan. A prioritized list of hypotheses, each with a primary metric, sample size, and run
time.
How John builds and tests landing pages
Start from the ad and the query
We read the search terms report and the ads, not just the keyword list. People follow cues that promise
they're getting closer to what they want, which NN/g calls information scent (NN/g). The headline, first paragraph, and hero image should confirm that the click went to the right place.
Write the page before designing it
Offer, proof, objections, and the next step, in that order. Proof means things a visitor can check:
licenses, real reviews, guarantees, pricing ranges where you publish them. We don't invent testimonials or
stats, and we won't put them on your page either.
Design the form with sales in the room
Every field has to earn its place. NN/g recommends keeping forms short, putting labels outside fields, and
showing clear errors next to the field (NN/g form usability). Placeholder text used as a label disappears as soon as someone types (NN/g on placeholders). Baymard saw 42% of test participants type their full name into a "First name" field at least once, which
is one reason we often use a single name field (Baymard).
Make it fast on a phone
Deloitte's study, commissioned by Google, found that a 0.1-second mobile speed improvement went along with
8.4% more retail conversions and an 8.3% better bounce rate on lead-gen information pages (Deloitte). It's a correlation across 37 brands, not a promise, but it matches what we see. We compress and size the
hero image, cut unused scripts, and load chat and heatmap tools after the page is usable.
Build to WCAG 2.2 AA
WCAG 2.2 added criteria that matter on landing pages, including Target Size (Minimum), Focus Not Obscured,
Redundant Entry, and Accessible Authentication (W3C). Buttons and links need a 24×24 CSS pixel target or enough spacing (W3C Understanding 2.5.8). WebAIM's 2026 scan found detectable WCAG failures on 95.9% of the top million home pages, and missing
form labels on 51% (WebAIM Million).
Track the outcome, not just the form
The conversion tag fires on a confirmed submission, consent is respected, and the lead ID travels to the CRM
so qualified leads can be sent back to Google, Microsoft, and Meta. A page that wins on form fills and loses
on qualified leads isn't a winner.
Test with a plan
Each test gets one hypothesis, one primary metric, guardrail metrics, a sample size, and a fixed run time
before it starts. The method is below.
Common landing page mistakes John fixes
Every ad pointed at the homepage. Visitors land on navigation instead of the answer to their
search.
Headlines that ignore the ad. The ad says "same-day water heater repair" and the page says
"Welcome to Smith Plumbing."
Ten-field forms for a callback. Fields sales never uses, required phone extensions, and
CAPTCHAs that block real people.
Hero images loaded lazily. The largest element on the page is the last to arrive, which hurts
LCP.
Calling tests early. Checking results daily and stopping at the first significant result
inflates false positives.
Policy problems. Pages that don't work on mobile, redirect unexpectedly, or don't match the
ad's claims can break Google's destination requirements (Google Ads policy).
A/B testing method we follow
Most failed tests fail on method, not on ideas. These are the rules we follow, from the experimentation
literature and the tools practitioners rely on.
Fix the sample size before starting. Evan Miller's well-known explanation shows how checking
results repeatedly and stopping when they look significant inflates the false positive rate (How not to run an A/B test). We calculate the sample size first (sample size calculator) and run to it.
Run whole weeks. Weekday and weekend visitors behave differently, so tests cover at least one
full week, usually two.
Check the traffic split. If a 50/50 test delivers 52/48, something is broken. Microsoft's
experimentation researchers call this sample ratio mismatch and treat it as a sign the results can't be
trusted (Fabijan et al., KDD 2019).
One primary metric, plus guardrails. For lead gen, the primary metric is the qualified lead
rate, with form fill rate and cost per lead as guardrails.
Split traffic the right way. For paid search, Google Ads experiments test one variable at a
time, such as the landing page, against an unchanged base campaign (Google Ads experiments). On-page testing tools suit tests that don't depend on the ad.
How much traffic a test needs
Visitors needed per variant to detect a relative lift with 95% confidence and 80% power, using the standard
two-proportion formula. Calculators such as Evan Miller's give similar figures.
Baseline conversion rate
Detect a 10% lift
Detect a 20% lift
Detect a 30% lift
2%
about 80,700
about 21,100
about 9,800
5%
about 31,200
about 8,200
about 3,800
10%
about 14,800
about 3,800
about 1,800
Not enough traffic? If a page gets a few hundred visits a month, a valid A/B test on small
changes could take years, as the table shows. We make the evidence-backed fixes (message match, speed, form
length, accessibility) directly, check the before-and-after with care, and save formal tests for big, bold
changes.
Pricing and engagement
Landing page work is quoted as a flat fee per page or template, with testing programs quoted
monthly. The number depends on how many ad groups or audiences need their own page, which platform the pages
live on, and how much copy and design work is involved, so we don't publish a single price. Per our
pricing page, fees are never a cut of your ad budget, and there's no long contract to
sign.
Platforms we work in
We build on the platform you already run where it can perform, and use these tools to measure and test.
Does landing page experience affect what I pay per click?
Landing page experience is one of the three Quality Score components, alongside expected click-through rate
and ad relevance. Google rates it above average, average, or below average for each keyword. Improving a
below-average page is one of the few levers that can help both conversion rate and ad rank.
How many form fields should a lead form have?
As few as sales can work with. Agree with sales which fields they actually use to qualify and follow up, then
cut the rest or move them to a later step. Every extra field needs a reason.
How long should an A/B test run?
Until it reaches the sample size calculated before launch, and for at least one full week, usually two.
Stopping early when the result looks good is the most common way tests go wrong.
Should we use a landing page builder or our main website?
Either can work. Builders are faster for marketing teams to change; your CMS keeps design and tracking
consistent. The deciding factors are speed, how easily you can edit, and whether tracking and consent work
properly.
Do landing pages need to meet WCAG?
We build to WCAG 2.2 AA because it's the current standard and it makes forms easier for everyone. The US
Department of Justice's web guidance points businesses to WCAG as a helpful standard. Whether a specific legal
requirement applies to you is a question for counsel.
What if we don't have enough traffic to A/B test?
Then we fix what research already supports (message match, speed, form length, accessibility), compare before
and after carefully, and save formal tests for big changes.
Talk to John about landing pages
Send the ads or campaigns you're running, the page they point to, and what a good lead looks like. You'll get a
written scope.
John, Kristy, or Sandeep will reply.One of the three of us will respond personally within 1 business day. No SDR queue.
Message received.
One of us will reply directly within one business day. While you wait, run a free
Buddy audit on your account.
Top 26 references
The primary sources, standards, research, and tools we rely on for this work. Every link was checked on
2026-10-11. We aren't affiliated with these publishers unless noted.
AI disclosure: This page was drafted with AI assistance and edited by a human. Research and
standards are cited to Deloitte, Baymard, NN/g, W3C, WebAIM, and Google as checked on 2026-10-11. Sample sizes are
calculated estimates. Nothing here is legal advice on accessibility obligations.
Check your landing page for free.
The Marketing Analytics Auditor checks PageSpeed, accessibility, SEO, and tracking across 13 categories. No
sign-up.
WordPress
Webflow
Shopify
Unbounce
Figma
Lighthouse / PageSpeed Insights
Google Ads
Google Tag Manager
WAVE accessibility checker