"A pretty site doesn't sell — social proof above the fold and speed do"

Direct answer first: a site with a gorgeous animated hero that loads in 3.5s converts worse than a plain site that opens in 1.2s with three client logos above the fold. Conversion doesn't come from design. It comes from two things — social proof in the first viewport and load speed. The numbers are brutal: 100ms of extra delay drops conversion by 7% (Akamai, 2017), and shaving 0.1s lifts retail conversion by 8.4% (Deloitte / Think with Google, 2020). A client logo above the fold lifts conversion by 12% on average (2,000-page landing study, 2026).

I'm Ulisses, I run Hens, a Brazilian software studio. I built OverAir (WhatsApp digital memory, 0 paying customers today — I'll be upfront about that) and Studio Kallos, and I ship landing pages and brand sites for paying clients. This post is an opinion with numbers behind it: most "premium" sites I see spend the budget on the wrong thing. They animate the hero, forget the social proof, and the LCP blows past 3s.

I'll walk you through the speed math, the social-proof math, the experiment that convinced me, and why a brand site and a conversion landing page are different products nobody should compare on price.

The speed math isn't an opinion — it's measured money

Site speed turned into marketing folklore, so let me anchor on primary data, not blog posts.

Akamai analyzed ~10 billion visits to top retailers in 2017. The finding: 100 milliseconds of load delay drops conversion by 7%. A two-second delay raises the bounce rate by 103% (Akamai, 2017). One hundred milliseconds. That's faster than a blink, and it costs 7% of your sales.

Deloitte went the other direction — instead of measuring the damage of slowness, it measured the gain from speed. The "Milliseconds Make Millions" study tracked mobile retail, travel, and luxury sites across Europe and the US for 4 weeks. Result: a 0.1s improvement in load time lifted retail conversion by 8.4%, average order value by 9.2%, and travel conversion by 10.1% (Deloitte / Think with Google, 2020). One tenth of a second moving the needle nearly 10%.

And Google measured bounce probability: when load time goes from 1s to 3s, bounce probability rises 32%; from 1s to 10s, it rises 123% (Think with Google). And 53% of mobile visitors abandon a page that takes longer than 3s to open.

The number I use as the cutoff is LCP — Largest Contentful Paint, the time until the biggest visible element paints. Google defines 2.5s at the 75th percentile as "good" (web.dev). The 75th percentile matters: it's no use that your site opens fast on your MacBook over fiber. It has to open fast for 75% of your real visitors, including the guy on 4G on the metro in Dubai.

Metric	"Good" threshold	What happens if you blow it
LCP (75th percentile)	≤ 2.5s	Each extra second ≈ −7% conversion per 100ms
Mobile abandonment	< 3s total	53% leave past 3s
Speed gain	−0.1s	+8.4% retail conversion

This isn't a technical detail for a developer to show off. It's the difference between the lead tapping your WhatsApp button or closing the tab before your hero finishes animating.

Social proof above the fold beats the generic hero

This is where almost every brand site gets it wrong. The first viewport — what shows up without scrolling — is the most expensive real estate on your site. And most sites waste it on a generic line like "We turn ideas into digital solutions" over a background video.

Wrong. That space wants social proof.

A 2026 study of 2,000 landing pages measured what each element adds in conversion. Putting client logos, a count, or a rating inside the first viewport lifts conversion by 12% on average. The highest-lift pattern tested was a named, quantified claim — something like "Used by 8 of the Fortune 50" — which delivered +22% versus no proof (Digital Applied, 2026). And the detail that closes the argument: 57% of desktop visitors and 64% of mobile visitors never scroll past the first viewport. If your social proof is below the fold, it doesn't exist for two-thirds of your traffic.

Social proof that works, in order of strength:

A real, specific metric. "327 studios use it" beats "thousands of happy clients." A concrete number beats an adjective.
A recognizable client logo. If you have a strong name, its logo above the fold is worth more than any headline.
An anchored testimonial with a name and a face. No marquee logo? A real testimonial, with a photo and a title, carries the proof.
A result screenshot. A dashboard, a 5-star rating, a revenue number.

What's not social proof: a "secure site" badge, a tech-stack badge, "since 2019." That's noise. Nobody converts because you use React.

My hard opinion here: if you can only optimize one thing on your site, move the social proof above the fold and stop animating the hero. The animated hero costs you LCP and moves no conversion. Social proof costs 20 minutes of editing and moves 12 to 22%. It's not even a close call.

Why a "pretty site" fools the client (and the developer)

The 800 KB Lottie hero, the parallax scroll, Framer Motion on every section — it impresses in the approval meeting and kills in production. Heavy animation pushes LCP past 3s easily. Every animation library is JavaScript the browser has to download, parse, and execute before it paints the content that matters.

I'm not talking from the sidelines. I've got five A/B variants on hens.com.br that weigh 1.6 to 1.9 MB each because they bundle React + ReactDOM + Babel + 19 inline WOFF2 fonts, all decoded at runtime in the browser. I know exactly what a heavy hero costs because I shipped one. They work for the experiment I'm running, but if they were a client's main front door, the LCP would embarrass me. Speed is debt you pay in lost conversion, not in an invoice.

The root problem is incentive. The agency selling a "premium site" gets judged on how beautiful the delivery looks — the client views it on the agency's desktop, over the agency's fiber, and approves it. Nobody opens PageSpeed Insights in the meeting. Six months later the client complains the site "doesn't bring leads" and nobody connects it to the 3.5s hero that turns away half the mobile traffic before it paints.

I'd avoid heavy scroll animation on any page whose job is to convert. For a design studio's portfolio, where beauty is the product, fine — the site is the sample. For a SaaS or service landing page, animation is a luxury that costs leads. It's the wrong call in most cases.

The experiment that convinced me

In March 2026 I rebuilt a client's landing page after they complained about weak lead flow. The old version was textbook: hero with a background video, generic headline, social proof — where it existed — buried in the footer, and a mobile PageSpeed score around 40.

I didn't touch the offer copy. I changed three things:

Pushed three client logos and one real metric ("X appointments booked") into the first viewport.
Killed the background video, swapped it for a raw static photo optimized as WebP.
Stripped the animation libraries and the dead JavaScript that was left. Mobile PageSpeed went from ~40 to 95+, LCP under 1.5s.

The click-through to WhatsApp — taps on the "Get in touch" button over visitors — went from 1.8% to 4.3%. Same offer, same traffic, same price. CTR more than doubled just from cutting weight and raising social proof. That's when I stopped treating speed and social proof as "best practice" and started treating them as the main thing.

The raw photo, by the way, outperformed the "produced" version. I don't have an isolated A/B number on that one point to claim causation — but it matches what you see in Y Combinator startups that rank well: simple by design, no-frills photo, CTA above the fold, zero animated hero. People who need to convert learn this early.

A brand site and a conversion landing page are different products

Here's the budget confusion I see most. A client asks for a quote on "a website" and compares three proposals on price, as if they were the same thing. They're not.

	Brand / institutional site	Conversion landing page
Goal	Credibility, brand, presence	Generate leads / direct sale
Success metric	Perception, brand SEO	CTR, conversion rate
Content	About, services, portfolio, contact	One offer, one action
Price band (US)	$3,000–8,000	$1,200–3,000
Where to invest	Visual identity, SEO, structure	Copy, social proof, speed, A/B

A brand site in the $3,000–8,000 range is a brand investment — it's worth the polished visual identity, the SEO structure, the multiple pages. A landing page in the $1,200–3,000 range is a single-function machine: turn a visit into a lead. Comparing the two on price is comparing a storefront window to a cash register. They do different jobs, cost different money, get measured differently.

My recommendation: if you're pre-launch and need to validate, don't pay $6k for a pretty brand site. Pay $1,200–3,000 for a fast landing page with social proof and put traffic on it. You build brand after you know you have a market. I've watched founders burn the whole budget on a gorgeous window before knowing anyone would buy.

What I'd do today, in order

If you're building or rebuilding the site that has to bring in clients, this is the order I'd follow — highest impact per dollar to lowest:

Social proof above the fold. Logo, real metric, named testimonial. Before anything visual. Costs almost nothing, lifts 12–22%.
LCP under 2.5s at the 75th percentile. Measure in PageSpeed Insights, on simulated 4G, not on your desktop. Kill the background video, lazy-load images, set font-display: swap.
One offer, one CTA, above the fold. If your main CTA needs a scroll, it doesn't exist for 60% of your traffic.
Raw photo before expensive production. Test simple first. You add beauty if the number justifies it.
Animation last, and only where it costs no LCP. A micro-interaction on tap, fine. A hero that animates for 2s while the lead waits, no.

The line I'd carve in stone: pretty is what the client praises in the meeting; fast-with-social-proof is what the lead taps at 11pm on their phone. They're not the same thing, and almost every budget gets spent on the first.

If you're deciding between spending on an animated hero or on speed + social proof, Hens has built landing pages, brand sites, and the A/B experiments running right now on hens.com.br itself. I can show you the math on your specific page — send me the link on WhatsApp and I'll run PageSpeed and tell you where you're leaking conversion.