CRO

Landing Page Navigation: Should You Remove the Menu?

Removing landing page navigation often wins on the form and loses elsewhere. Measure signups per visitor across the whole site, not on one page.

Flat illustration of a hand lifting away a top bar of rounded menu tabs, above one wide path that leads to a single large rounded button, in deep green and teal on a soft mint green background

Removing landing page navigation is a test of exits, not of persuasion. The visitors who used to click the menu do not disappear when you take it away: some of them were signing up on another page of your site. If your test only counts the form on the landing page, the metric is tilted toward the no-menu variant by construction. The right metric is signups per randomized visitor, on any page, within a fixed window. This guide is part of the conversion rate optimization guide and applies the method from landing page optimization to one classic decision: should a landing page have a menu? It shows one illustrative test read three ways, what to test instead of a yes or no, which traffic needs a landing page menu and which does not, and where Google Ads fits.

The advice to remove navigation from landing pages is not wrong, it is incomplete: it assumes every menu click is a lost conversion.

What a menu does to a visitor

A menu is a set of exits. Each link offers a reason to leave: check pricing, read about features, go back to the home page. On a dedicated campaign page, every exit competes with the single action the page exists for.

The idea has a name: the attention ratio, defined in Unbounce’s conversion glossary as “the ratio of links on a landing page to the number of campaign conversion goals”. The glossary’s recommendation is blunt: “In an optimized campaign, your attention ratio should be 1:1”, because “every campaign has one goal”. A full site menu, a footer with twenty links and social icons push that ratio far from one.

There is published evidence that cutting links can raise the conversion rate of the page. HubSpot reported a test across five of its highest traffic landing pages in which version B “Removed all external links, including the top navigation, footer navigation, and social media share links”. According to HubSpot, top of funnel pages saw “a 0-4% bump in conversion rate”, while middle of funnel pages “had a 16% and 28% lift”. The article’s text does not report sample sizes, p-values or intervals, and the metric it describes is the conversion rate of the landing pages themselves.

That last detail is the whole story of this guide. A landing page conversion rate counts what happens on the landing page. It does not count what happened to the visitors who left through the menu.

Why removing landing page navigation is a test of exits

Who clicks the menu on a campaign page? Some are distracted and would have converted without it. Some browse and leave for good. And some do what a careful buyer does: open pricing, read the features page, and then sign up somewhere else on the site, since pricing and home pages usually have their own signup button.

A test that measures only the form on the landing page treats all three groups as losses for the control. When the menu disappears, the first group converts on the page, which is a real gain. The third group also converts on the page, because the menu is gone and the form is the only door left. That is not a gain at all. It is the same signup, moved from one page to another. The local metric records it as a win for the variant.

Where the 20,000 control visitors with a menu end upA flow from one box on the left, 20,000 visitors randomized to the control landing page with a menu, to three outcomes on the right. 560 sign up on the landing page form, 2.80 percent. 160 click the menu and sign up on another page such as pricing, features or home. 19,280 do not sign up within 7 days. A bracket shows that the local metric counts only the 560, while the per visitor metric counts 720, or 3.60 percent.Control with menu: where 20,000 randomized visitors end up in 7 days20,000 visitorsrandomized to thecontrol, with menu560 sign up on the page form2.80% of visitors160 click the menu, then sign upon pricing, features or home19,280 do not sign upwithin the 7 day windowlocal metriccounts 560per visitorcounts 720Local metric: 560 / 20,000 = 2.80%. Per visitor metric: (560 + 160) / 20,000 = 3.60%.Illustrative data, not customer data.
The 160 visitors in the middle box are invisible to a metric that only watches the landing page form. They are the reason a no-menu variant can “win” without creating a single extra signup.

Kohavi, Deng, Longbotham and Xu describe this pattern from Bing’s experiments in “Seven Rules of Thumb for Web Site Experimenters”. Their rule 5 is titled “Reducing Abandonment is Hard, Shifting Clicks is Easy”. They warn that “Many times all the feature is doing is shifting clicks around and cannibalizing other areas of the page”, that “the overall experience is what really matters”, and they put it in one line: “local improvements are easy; global improvements are much harder.” A landing page form is a local metric. Your signups are the global one.

Choosing a primary metric in general is covered in the overall evaluation criterion; here the obvious local metric lies in a predictable direction.

The illustrative test: one SaaS landing page, three readings

Everything below uses one illustrative scenario, not customer data. A SaaS landing page receives paid traffic, about 10,000 eligible visitors a week. The test runs for four weeks, so each arm gets 20,000 visitors. The goal is a free trial signup. Variant A is the current page with the full site menu; variant B removes the menu and keeps everything else the same.

How big a test this page can afford

Sizing comes before launch, and it depends on which metric you size for. The per visitor signup rate across the site is 3.6 percent in this scenario; the local form rate is 2.8 percent. At 95 percent confidence, 80 percent power and a two-sided test:

relative MDE per visitor signups (3.6%), per variant days at 10,000 a week local form signups (2.8%), per variant
10% 44,054 62 57,135
15% 20,027 29 25,979
20% 11,516 17 14,942

At this traffic a four-week test can only see large effects, and sizing on the local rate would buy a bigger sample for the wrong question.

Reproduce the middle row. With current conversion rate 3.6, minimum detectable effect 15 relative, confidence 95, power 80 and 10,000 visitors per week on a two-sided test, the calculator shows 20,027 visitors per variation, 40,054 in total for two variations and about 29 days.

Sample size calculator
-Visitors per variation
-Total (2 variations)
-Estimated duration

Two-proportion normal approximation, 2 variations (50/50). Tweak the inputs and watch it update live.

Run it backward: with 20,000 visitors per variant and a 3.6 percent baseline, the smallest lift this test can reliably detect is about 14.5 percent relative, or 0.52 percentage points, from 3.60 percent to about 4.12 percent. The MDE calculator does that backward calculation for any traffic level, and how many visitors an A/B test needs explains the trade-offs.

Reading 1: the local metric says “winner”

The first reading counts only signups made through the form on the landing page itself.

arm visitors signups on the page form rate
A, with menu 20,000 560 2.80%
B, no menu 20,000 660 3.30%

The relative lift is +17.9 percent, or +0.50 percentage points, with a z of 2.91 from the stats engine, a p-value of 0.0036 and a 95 percent confidence interval of the difference from +0.16 to +0.84 points. By the book, a significant winner. This is the result most teams ship.

Reading 2: the per visitor metric says “nothing happened”

The second reading counts every signup made by each randomized visitor, on any page of the site, within 7 days of their first visit to the test page.

arm signups on the page signups on other pages total signups rate per visitor
A, with menu 560 160 720 3.60%
B, no menu 660 40 700 3.50%

In the control, 160 visitors clicked the menu and signed up on pricing, features or the home page. In B the logo links nowhere, so its 40 signups elsewhere come from visitors who came back later on their own.

The result: minus 2.8 percent relative, minus 0.10 percentage points, a z of minus 0.54 from the engine, a p-value of 0.5889 and a 95 percent interval from minus 0.46 to +0.26 points. Inconclusive.

Removing the menu changed where people signed up, not how many did. The 100 extra form signups in B are almost exactly the signups A was already getting through its menu, and the interval still allows a real loss of up to 0.46 points. Shipping B on Reading 1 would be a decision made on a metric that could not see the cost.

Reading 3: when removing the menu really wins

Removing navigation can work. Consider a different page: cold ad traffic with a single offer, where the menu mostly leaked attention. Measured on the right metric:

arm visitors signups per visitor, any page rate
A, with menu 20,000 720 3.60%
B, no menu 20,000 828 4.14%

That is +15.0 percent, a z of 2.80 from the engine, a p-value of 0.0051 and an interval from +0.16 to +0.92 points. Significant on the metric that counts every signup. Here the menu really was costing conversions, and removing it is a gain you can bank.

Three readings of a menu removal test, 95 percent intervals of the differenceThree horizontal confidence intervals on a percentage point scale from minus 0.6 to plus 1.0, with a vertical line at zero. Reading 1, the local form metric, runs from plus 0.16 to plus 0.84 with a point estimate of plus 0.50, entirely above zero. Reading 2, signups per visitor on the same test, runs from minus 0.46 to plus 0.26 with a point estimate of minus 0.10, crossing zero. Reading 3, a cold ad page measured per visitor, runs from plus 0.16 to plus 0.92 with a point estimate of plus 0.54, entirely above zero.95% CI of the difference, no menu minus menu, in percentage points-0.50+0.5+1.0Reading 1, page formlocal, blind to other pages+0.16 to +0.84, p = 0.0036Reading 2, per visitorsame test, inconclusive-0.46 to +0.26, p = 0.5889Reading 3, per visitorcold ad page, real gain+0.16 to +0.92, p = 0.0051Dots are point estimates: +0.50, -0.10 and +0.54 pp. Illustrative data, two-sided z-test.
Readings 1 and 2 are the same test. The only difference is which signups each metric is allowed to see. Reading 3 is what a real win looks like on the honest metric.

Check the numbers in the calculator

Statistical significance calculator
Control (A)
Variation (B)
Control (A) · Rate-
Variation (B) · Rate-
Relative lift-
p-value-
95% CI of the difference-

Two-sided two-proportion z-test. "Not significant" almost always means not enough sample, not that the versions are equal.

Enter 20000 and 720 for Control (A) and 20000 and 700 for Variation (B), the per visitor data from Reading 2. The calculator shows rates of 3.60% and 3.50%, a relative change of -2.8%, a p-value of 0.5889 and a 95% CI of the difference of -0.5% … +0.3% (pp). Now enter the local numbers from Reading 1, 560 and 660 on the same visitor counts: the screen shows 2.80% and 3.30%, +17.9%, a p-value of 0.0036 and a CI of +0.2% … +0.8% (pp). Finally Reading 3, 720 against 828: 3.60% and 4.14%, +15.0%, 0.0051 and +0.2% … +0.9% (pp). The screen rounds the interval to one decimal and does not display z; the z values and two-decimal intervals quoted above come from the same engine. More on reading these outputs in statistical significance for A/B tests.

Defining the right metric: unit, numerator, window

A metric that can judge a menu test has three parts, and each one closes a specific leak.

part local metric (misleading here) per visitor metric (use this)
unit sessions or pageviews of the landing page the visitor randomized to an arm
numerator form submissions on the landing page signups by those visitors on any page of the site
window whatever happens during the session fixed, for example 7 days from first exposure
what it misses every signup that happened after a menu click nothing the visitor did on your site in the window

The unit is the visitor who was randomized. Assignment must be sticky: a visitor who returns on day 3 sees the same arm and is counted once.

The numerator is every signup by those visitors, wherever it happens. Track the signup at its confirmation step (the “account created” event or the welcome page), not at the landing page form, so it fires the same way from any page. If you use GA4 for the event, how to track A/B test events in GA4 shows how to attach the arm to it.

The window is fixed in advance, such as 7 days from first exposure, so late arrivals get the same time to convert. If your tool cannot cut windows, run whole weeks and count conversions for the whole test period; randomization keeps the comparison fair between arms.

Before trusting any of it, confirm the split is close to 50/50 with a sample ratio mismatch check, and that the signup event fires on every page with a signup button.

What to test instead of menu versus no menu

“Menu or no menu” is a crude question. The menu does several jobs at once, and some of them serve the visitors you want. A better test cuts the exits that leak attention and keeps the ones that serve real needs.

variant what it removes what it keeps when it fits
Reduced menu most links two or three links that answer real objections, such as pricing and security visitors who compare before buying
Logo without a link the escape to the home page brand recognition at the top cold traffic that may bounce to the home page
In-page anchor links every link that leaves the page a menu that jumps to sections of the same page: how it works, pricing, FAQ long pages that answer many objections
Minimal footer the full site footer privacy policy, terms and the other legal links you are expected to show almost every campaign page
Visible sign-in link nothing that matters a “Sign in” link for existing customers any page that customers also land on

Who needs a landing page menu, and who does not

Whether a landing page menu helps or hurts depends on who arrives and what they need before they can decide. The same page can need one for brand search traffic and not need one for a cold social ad.

Decision tree: does your traffic need the landing page menuThree questions in sequence. First, is this a home, category, pricing or content page? If yes, keep the full menu and do not test removing it. If no, second question: do brand, returning or existing customer visitors land here? If yes, keep a reduced menu with a sign-in link and test reduced against full. If no, third question: does the buyer need to compare or research before deciding, as in high ticket B2B? If yes, use in-page anchor links and test them against the menu. If no, the page serves cold ad traffic with a single offer: test removing the menu, keep the legal footer, and judge per visitor.Is this a home, category, pricingor content page?yesKeep the full menu.Navigation is the job of the page.noDo brand, returning or existingcustomer visitors land here?yesReduced menu plus sign-in.Test reduced against full.noMust the buyer compare or researchfirst, as in high ticket B2B?yesIn-page anchor links.Answer objections on the page.noCold ad traffic, single offer: test removing the menu.Keep the legal footer, logo without a link, and judge signups per visitor, any page.The questions go from cheapest to answer to most expensive. Any “yes” means the menu is serving someone.
Most of the “remove the menu” advice is really about the bottom box. The higher you exit this tree, the more the menu is doing useful work.

Visitors who need navigation:

Nielsen Norman Group’s research on hidden navigation shows how much people rely on visible menus when they have something to find. In their study of 179 participants doing tasks on real sites, Pernice and Budiu report that on desktop, navigation was used in 48 percent of cases when it was visible and in 27 percent when it was hidden behind a menu icon, and they summarize that “Discoverability is cut almost in half by hiding a website’s main navigation”. Removing a menu goes further than hiding it.

Visitors who do not need it: cold traffic from an ad with one offer, landing on a page built for that ad. They have one question, whether the offer is worth it, and the page either answers it or does not. For them, the menu is mostly an exit. That is the case in Reading 3, and it is also the case where message match between the ad and the page matters most.

Segmenting by traffic source: a second look

Because the answer depends on who arrives, it is tempting to slice the results by source after the test and pick the segment where the no-menu variant won. Resist it.

When the pattern is real, the fix is usually routing: brand search lands on a page with navigation, cold ads on a focused page.

Guardrails for a navigation test

A menu test can win on signups and still do damage. Declare these guardrail metrics before launch, with a threshold for each.

guardrail why it moves when the menu goes how to read it
Clicks on sign-in or the customer area existing customers lose their way to log in a drop means customers are stuck, not converted
Support contacts and chat openings from the page people who cannot find what they need ask for it a rise is a cost the signup metric does not see
Bounce rate visitors who need another page now leave instead of browsing diagnostic only, never the decision
Signup quality, such as trial activation a funnel with one door can push in less qualified signups read it once the cohorts mature

Bounce rate is a diagnostic, not a verdict: a bounce can be a visitor who signed up later through another page.

If the instinct is to keep the visitor on the page, an exit-intent popup is the other common lever, with its own traps; see exit-intent popup A/B testing.

Google’s help center article “Optimize your ads and landing pages” has a section titled “Make sure your landing page is easy to navigate”. It asks advertisers to “Make it quick and easy for customers to perform the action you want them to take”, says “Don’t make people hunt around for information they might need”, and advises against “cluttering your site with too many ads or pop-ups”. Among the ways to build an effective mobile site, it lists “Use simple navigation”. Landing page experience is one of the three components of Quality Score, described as “How relevant and useful your landing page is to people who click your ad”.

In February 2025, Shashi Thakur, Google’s VP and GM of Search Ads, wrote that “landing pages must be both relevant and easy to navigate” and that Google “developed a new prediction model” to “more precisely capture the quality of your navigation experience”. The case Google describes is an ad that “leads to an unexpected destination and doesn’t offer other helpful navigation options”, and Google says it now shows fewer ads like that.

None of the text we read says a landing page needs a top menu. A focused page that delivers the ad’s promise and makes the action “quick and easy” fits the guidance. The risk is the case Google describes: a page that does not match the ad and gives the visitor nowhere to go. Google does not publish how its model weighs a missing menu, so do not assume a penalty or its absence; watch the landing page experience status of the keywords that send traffic to the page. Anchor links and a reduced menu are the safest options here.

Pages that should keep their menu

The no-menu question only applies to dedicated campaign pages built around one offer. On the home page, category and listing pages, pricing and comparison pages, and content pages such as blog posts and documentation, navigation is the job of the page. If a home page also receives ad traffic, give the ad its own page instead of stripping the home page.

A test plan you can copy

  1. Page type confirmed: a dedicated campaign page with one offer.
  2. Variant chosen from the middle options unless the page is pure cold traffic.
  3. Primary metric declared: signups per randomized visitor, any page, fixed window such as 7 days.
  4. Signup event checked at the confirmation step, on every page with a signup button.
  5. Sample sized on that metric with the sample size calculator.
  6. Guardrails and segments declared in advance, plus the Google Ads status of keywords that send traffic.
  7. Decision rule written: ship only if the per visitor metric wins and no guardrail crosses its threshold.

Make this automatic with Donnu

The trap in a navigation test is the metric: a test that only watches the landing page form is tilted toward the no-menu variant, and can report it as a winner even when signups only moved from the pricing page to the form.

Donnu runs A/B tests and Split URL tests on web pages. You can build the no-menu or reduced menu variant in the visual editor directly on the live page, or send each arm to its own page with Split URL. Each visitor is randomized once and keeps the same variation for the whole test. The conversion goal can be a click, a form submission or a page visit, so you can set it to the signup confirmation page instead of the landing page form. When a visitor in the test reaches that page from anywhere on your site where the snippet is installed, including the pricing or home page, the conversion is credited to that visitor’s variation. That is the per visitor reading of this guide, without exporting anything. Reports are Bayesian, and the product is built PT-BR and LGPD first.

What stays with you is the judgment about which exits are leaks and which are service. The sample size calculator and the significance calculator reproduce every number above, and the landing page grader gives you a list of fixes to test next. When you are ready, create your Donnu account and run your navigation test on the metric that counts every signup.

References

Read next: Landing page optimization · Conversion rate optimization guide · Overall evaluation criterion · Guardrail metrics · CTA A/B testing · Exit-intent popup A/B testing · Ad to landing page message match · Leia em português

Frequently asked questions

Should I remove the navigation menu from my landing page?
Only if a test measured on the right metric says so. Removing the menu often raises the conversion rate of the form on that page, because people who would have clicked the menu now stay. But some of those people used to sign up on another page of your site. Judge the test on signups per randomized visitor anywhere on the site, within a fixed window, and remove the menu only if that number goes up.
Why does a no-menu variant tend to win on the landing page itself?
Because the metric counts the menu as a loss by construction. Every signup that the control produced on the pricing page, the features page or the home page is invisible to a metric that only counts the form on the landing page. In the illustrative test in this guide, the no-menu variant wins on the local form by 17.9 percent with a p-value of 0.0036, and on signups per visitor across the site it shows minus 2.8 percent with a p-value of 0.5889: inconclusive.
What is the right metric for a landing page navigation test?
Conversion per randomized visitor, anywhere on the site, within a fixed window. The unit is the visitor who was assigned to an arm. The numerator is every signup those visitors complete on any page, not just on the landing page form. The window is fixed in advance, for example 7 days from the first exposure, so both arms get the same time to convert.
Which pages should never lose their menu?
The home page, category and listing pages, pricing and comparison pages, and content pages such as blog posts and documentation. Their job is to help people find their way, so the menu is part of the product, not a distraction. The no-menu question only makes sense on dedicated campaign pages built around a single offer.
Does Google Ads penalize landing pages without navigation?
Google does not say that. Its help center asks advertisers to make the landing page easy to navigate and to make it quick and easy to take the action they want, and a February 2025 Google post says ads that lead to an unexpected destination without other helpful navigation options are now shown less. A focused page that delivers exactly what the ad promised is not that case, but Google does not publish how it weighs a missing menu.
What should I test instead of menu versus no menu?
Test the middle options: a reduced menu with two or three links, a logo that does not link anywhere, anchor links that jump to sections of the same page, a minimal footer with the legal and privacy links, and a visible sign-in link for existing customers. These keep the exits that serve real visitors and cut the ones that only leak attention.