A homepage serves everyone and sells to almost no one. Below: the structure of a real landing page, how to match it to the ad, what to cut, and the test order that finds the version worth funding.

Q1. Why is the homepage such an expensive place to send a paid click?

Because you just paid for one person with one question, and the homepage answers with a table of contents. WordStream’s 2025 benchmarks, built on 16,446 US search campaigns, put the median click at $5.26 (up 12.88% in a year) and the median lead at $70.11. That is a pricey visitor to hand a menu with nine options.

1. It talks to too many people. Tas Bober, who has worked on 400+ paid landing pages for B2B SaaS, calls the homepage send the biggest mistake in paid: the message gets diluted and campaign data gets blended with organic and direct visits (Exit Five).
2. Focus is structurally hard there. Fletch PMM, with 350 venture-backed homepage projects behind them, flags the same problem: serving several segments makes one page tough to keep focused (PeakSpan master class).
3. The math of attention works against it. Julian Shapiro’s formula is “Purchase Rate = Desire – (Labor + Confusion)” (Julian.com). Each menu item adds labor. Each extra audience adds confusion.

Q2. What does that cost in real dollars?

Run the same $5,000 through two doors. Unbounce’s Conversion Benchmark Report covers 41,000+ landing pages, 464 million visitors and 57 million conversions, which makes it a decent yardstick.

Where the click lands	Clicks at $5.26	Conversion rate	Leads	Cost per lead
Homepage (illustrative rate, plug in yours)	950	2.5%	24	$208
Dedicated page at Unbounce all-industry median	950	6.6%	63	$79
Dedicated page at Unbounce paid search average	950	10.9%	104	$48

Table 1. Same budget, same clicks. The 2.5% homepage figure is my working assumption for illustration; the other rates and the click price are published benchmarks.

Q3. What is structurally different about a real landing page?

Element	Homepage	Paid landing page
Job	Orient every type of visitor	Get one action from one audience
Navigation	Full menu, footer, social links	Logo only, or a few anchor links down the page
Headline	Company-level positioning	Repeats the promise made in the ad
Calls to action	Several, for several intents	One primary ask, repeated
Proof	General logo wall	Results tied to this specific offer
Reporting	Mixed with organic and direct visits	Clean data per campaign

Table 2. Structural differences, drawn from Bober, Dry, Shapiro and Demand Curve.

1. Harry Dry’s ten-part anatomy (Marketing Examples): title, subtitle, visual, social proof and CTA up top; features and objections, more proof, FAQ, a second CTA and a founder’s note below. His summary: “Above the fold you earn the customer’s attention. Below the fold you earn the sale.”
2. Bober’s 90/10 rule (Dreamdata): 90% of the page delivers value and 10% makes the ask. She puts the form at the bottom, with anchor links from the top buttons.
3. Even a good homepage keeps the menu short. Demand Curve suggests 2 to 4 links plus a CTA. A paid page goes further and drops the menu.

Q4. What is message match, and how much is it worth?

Message match means the page finishes the sentence the ad started. Same words, same offer, same look. Picture an ad reading “Payroll for restaurants, set up in 10 minutes” that lands on “The all-in-one HR platform for modern teams.” The visitor wonders if they clicked the wrong thing, and the back button is right there.

1. 66% lift. KlientBoost added “Fully Free for Life” to both the ad and the page for ChatSupport and reported a 66% jump in conversion rate.
2. 10 to 30%. Demand Curve, after writing several hundred above-the-fold sections, says header and hook changes move conversion that much.
3. +54% signups. GetUplift built Teamwork a dedicated comparison page for comparison shoppers instead of sending them to the general site.
4. Match the audience too. Jon Loomer describes a page converting at 10% that fell to under 1% once it was fed cheap click-optimized traffic. “Not all clicks are created equal.”

Bober’s test is blunt and useful: if the page introduces something the ad didn’t promise, remove it.

Q5. What should come off the page, and what should stay?

1. Navigation and stray links. HubSpot stripped them from five pages: top-of-funnel offers moved 0 to 4%, middle-of-funnel offers rose 16% and 28%. Yuppiechef removed one nav bar and signups went from 3% to 6% (VWO).
2. Big words. Unbounce found pages written at a 5th to 7th grade level convert at 11.1% versus 5.3% for professional-level prose. Bober says about 95% of the pages she reviewed were “buzzword stuffed.”
3. Competing buttons. One primary ask. “Schedule a 15-Minute Demo” beats “Learn More” in her playbook.
4. Seconds. See Table 3. Unbounce reports 83% of landing page visits are mobile, and Google’s research found 53% of mobile visitors leave a page that takes over three seconds.

Load time	B2B lead gen conversion rate	B2C ecommerce conversion rate
1 second	Almost 40%	3.05%
2 seconds	34%	1.68%
3 seconds	29%	1.12%
4 seconds	Not published	0.67%

Table 3. Goal conversion rate by load time. Source: Portent, 20 sites, 100 million+ page views, 27,000+ landing pages.

Two things to keep. FAQs: Bober’s heat maps showed 17% of all page clicks landing there. And form fields that qualify. Michael Aagaard of ContentVerve cut a form from 9 fields to 6 and leads dropped 14.23%; rewriting the labels on all 9 raised leads 19.21% (CTAConf slides). Shorter is a hypothesis to test.

Q6. What testing sequence finds the version worth spending behind?

A dose of humility first. Georgi Georgiev’s review of 1,001 A/B tests found a median lift of 0.08%, a mean of 2.08%, and a significant winner in 33.5% of tests; the median test ran 30 days on 60,342 users. Optimizely reports a 12% win rate across its customer experiments, and says four-variation tests deliver 3.5x the expected impact of a plain A/B. Translation for a normal ad account: skip button colors and test big swings, in order of leverage.

#	Test	What to do	Evidence
1	Traffic quality	Optimize ads for leads or purchases before judging any page	Loomer: 10% fell to under 1%
2	Homepage vs. dedicated page	Whole-page split, same ads, same budget	Unbounce median 6.6%
3	Headline match	3 or 4 headlines that echo the ad copy	KlientBoost 66%; Demand Curve 10 to 30%
4	Subtraction	Remove nav, footer links, extra CTAs	HubSpot up to 28%; Yuppiechef 3% to 6%
5	Clarity and proof	Simplify copy, add FAQ and specific results	Unbounce 11.1% vs. 5.3%; Bober 17%
6	Form and CTA wording	Rewrite labels before cutting fields	Aagaard: labels +19.21%, fewer fields 14.23% drop
7	Speed	Get under 2 seconds on mobile	Portent: 40% at 1s, 29% at 3s

Table 4. A seven-step sequence, biggest levers first.

1. Run full weeks. Aagaard’s tests ran four full weeks each with 500+ conversions. Stopping on a good Tuesday is how fake winners get budget.
2. Pick the winner on cost per qualified lead. Conversion rate alone can be gamed by a softer offer. Let sales tell you which version sends people who buy.
3. Borrow patterns, then verify. GoodUI logs tests such as Autodesk’s on 453,820 visitors where exposing menu options on a homepage was the idea under test. Context decides. A homepage serves explorers; a paid page serves one click.
4. You are ahead by starting. In Wynter’s 2026 survey of 50 senior B2B SaaS leaders, 58% had never tested their pricing page at all.

Q7. How many landing pages do I need, and does the channel change the target?

Fewer than you fear. Bober recommends one page per campaign, offer or audience segment, without building a small city of them. Short on time? Her quick fix is a stripped-down copy of the homepage with the navigation removed and one offer. Demand Curve suggests writing for your top 2 or 3 personas. The channel does change what “good” looks like, as Unbounce’s data shows.

Traffic source	Average conversion rate	What it means for your page
Email	19.3%	Warm readers; a short page can work
Instagram ads	17.9%	Visual, mobile-first, quick offer
Facebook ads	13.0%	Interruptive; lead with the hook
Paid search (Google 11.3%)	10.9%	High intent; echo the search term
LinkedIn ads	4 to 5x lower than Facebook	Expensive clicks; qualify hard, prove more

Table 5. Landing page conversion by traffic source. Source: Unbounce.

Q8. Can you show a before and after?

Take a made-up company, PlateRoll, selling payroll software to restaurants. The ad reads “Restaurant payroll with tip tracking, set up in 10 minutes.”

1. Before (homepage). Headline: “The all-in-one HR platform for modern teams.” Seven menu items, three buttons, a logo wall of tech companies, a blog feed. The restaurant owner has to hunt for the word “tips.”
2. After (landing page). Headline: “Restaurant payroll with tip tracking, live in 10 minutes.” A screenshot of the tip report. One button: “Book a 15-Minute Setup Call.” A quote from a named restaurant owner with a number in it. Six FAQs pulled from sales calls. Form at the bottom.
3. What changed. Desire went up because the page is plainly about restaurants. Labor and confusion went down because there is one path. That is Shapiro’s formula doing its job.

The Short Version

Keep the homepage for people who typed your name. For everyone you paid to meet, build one page with one promise and one button, cut what the ad never mentioned, then test in the order above. Your ad budget will stop funding a guided tour.

Sources

1. Tas Bober via Exit Five, “8 Reasons Your B2B Landing Pages Aren’t Converting”: exitfive.com; and via Dreamdata: dreamdata.io
2. Harry Dry, Marketing Examples, landing page guide: marketingexamples.com
3. Julian Shapiro, landing pages guide: julian.com
4. Demand Curve, above-the-fold playbook: demandcurve.com
5. Fletch PMM, homepage positioning master class: peakspancapital.com
6. Jon Loomer, conversion rate and Meta ads: jonloomer.com
7. Michael Aagaard (ContentVerve), CTAConf 2015 slides: unbounce.com
8. Georgi Georgiev, Analytics Toolkit, 1,001 A/B tests: blog.analytics-toolkit.com
9. GetUplift (Talia Wolf) case studies: getuplift.co; GoodUI pattern #14: goodui.org; Wynter pricing study: wynter.com
10. Unbounce Conversion Benchmark Report: unbounce.com
11. WordStream 2025 Google Ads benchmarks: wordstream.com
12. Portent site speed study: portent.com; Google “Need for Mobile Speed” via Marketing Dive
13. HubSpot navigation test: blog.hubspot.com; VWO Yuppiechef case: vwo.com
14. KlientBoost on message match: klientboost.com; Optimizely experimentation takeaways: optimizely.com