The night the SaaS site rewrote itself

The call came in at 11:47 p.m. on a Thursday. A founder I'd helped with one prior project — a smart, scattered woman named Priya who'd raised a small seed for a SaaS that does something useful with calendars — was launching on a Monday. Her co-founder, the one who'd been building the marketing site, had emailed her three hours earlier to say he was checking himself into a hospital for what turned out to be a serious-but-not-fatal thing. He'd be out for two weeks.

The SaaS landing page was 60% done. Priya had read it once that afternoon and her stomach had gone cold. She called me because we'd worked together before and because she knew I would not pretend it was fine.

"Can you look at it?" she said. "I think it's bad. I'm not sure if I think it's bad or if I'm panicking."

I opened the staging URL. It was bad. Not in the way that requires a full rebuild — in the much more common way, the way most SaaS landing pages are bad. It was trying to do nine things on one screen. The hero had three CTAs. The pricing table had six tiers. There was a section titled "Why teams choose us" that included the phrase "world-class" and a stock photograph of four people pointing at the same monitor.

I made coffee. I told Priya I'd call her back in an hour.

Eleven hours

What I did, between 12:30 a.m. and 11:30 a.m. on Friday, was not a redesign. I want to be precise about that. I did not open Figma. I did not redraw the page. I did not introduce a new color, a new typeface, or a new component.

What I did was take the onedb directory and treat it like a checklist. I worked through the hero, the social proof, the feature blocks, the pricing table, the FAQ, the footer. For each part, I opened the matching folder, looked at the references — Linear's pricing, Stripe's hero, Vercel's feature grid — and asked one question: what is this part trying to do, in one sentence?

If I couldn't answer in one sentence, I cut the section. Two sections went. The "world-class" one, and a testimonial carousel with three testimonials that all said roughly the same thing.

Then I rewrote the hero. The original hero said: Calendars that finally make sense for modern teams. Try free. Watch demo. Book a call. I replaced it with: The shared calendar your team won't fight about. Start free. One sentence. One CTA.

The pricing table came next. The six tiers — Free, Starter, Team, Pro, Business, Enterprise — were collapsed to three: Free, Pro, Enterprise. The middle one got a small recommended badge. I removed seven feature rows from the comparison and grouped the remaining ten into three categories. The page stopped feeling like a tax form.

By 4 a.m. the page was sixty percent shorter than it had started. By 8 a.m. it was forty percent shorter than that. By 11:30 a.m. I sent Priya a Loom that was four minutes long, walking through every cut and why.

What I cut, and why I cut it

The third CTA in the hero

Every SaaS hero wants three CTAs because every founding team has three internal stakeholders and each one wants to be represented. Sales wants "Book a call." PLG wants "Start free." Marketing wants "Watch demo." The compromise is a hero with three buttons of equal weight, none of which is the obvious next step.

The cure is to admit that the hero is for one buyer at one moment. Most SaaS sites — especially ones priced under fifty bucks a seat — are PLG. Start free. The other two CTAs go in the footer or, better, the second scroll.

Three of the six pricing tiers

Six tiers was a planning artifact. Someone on the sales side had said "we need a Business tier between Team and Pro," and someone else had said "we need a Starter for solo users," and the page had absorbed both. Real users don't read a six-tier table. They scan three, pick the middle one, and leave.

Three tiers, a recommended badge, ten feature rows, three category headers. That's the whole table. The Business tier, if it survives Monday's launch, lives inside the Pro card as a "per seat" toggle.

The testimonial carousel

Three testimonials that all say the product is great is one testimonial repeated. We pulled the strongest one out, placed it under the hero as a single quote with a real photo and a real title, and deleted the other two. Social proof works when it's specific. Generic praise reads as a placeholder, no matter how many small business logos you stack around it.

What I did not touch

I want to be honest about this part, because it matters.

I did not touch the type scale. Priya's co-founder had picked a clean editorial serif for the headline and a geometric sans for the body, and the choice was already correct. I did not change the color palette. The accent was a deep oxblood — unusual for a SaaS, but quietly distinctive. I did not change the spacing rhythm. The page was already breathing well in the parts that weren't trying to do too much.

What I cut was the additions. What I kept was the bones. This is the hardest part of the work, because the temptation is always to leave a mark of your own. The purple pen — and I think about this almost every night, working on someone else's homepage — does not leave a mark of its own. It marks the marks that are already there.

Monday

Priya launched on Monday at 9 a.m. Eastern. She posted the launch video — ninety seconds, no music, a quick demo, one founder talking — at 9:02. The site got eighteen thousand visits in the first six hours. Conversion to free trial was almost five percent. That is unusually high. I am not going to tell you that the rewrite did it, because the product was already good and the launch video was the real act, but the page didn't fail under load and the analytics looked, frankly, like a SaaS site that knew what it was for.

The Slack message I got from her at 6 p.m. that night said: "It's working. Thank you. I'm crying a little."

I cried a little too. I will not pretend I didn't.

What this taught me about SaaS landing pages

Most SaaS landing pages are not under-built. They are over-built. There are too many sections, too many CTAs, too many tiers, too many adjectives. The cure is almost never to add. The cure is almost always to subtract, in a specific order, with a specific reason for each cut.

The order, more or less, is this:

1. 01Cut every section whose one-sentence purpose you cannot state out loud. If you can't say it in one sentence, the section is hiding its purpose, and the visitor will not find it.
2. 02Cut every CTA after the first. A second CTA is fine if it's clearly secondary. A third is a confession.
3. 03Cut every pricing tier beyond three. If your business needs five tiers, build a comparison page. The pricing page is for the decision, not the configuration.
4. 04Cut every testimonial that is interchangeable. Keep the specific ones. The ones that name a number, a problem, a team. Throw the rest.
5. 05Cut every adjective that survived two rounds of edits. "World-class," "cutting-edge," "seamless," "intuitive." These do not describe your product. They describe the writer's panic.

After you cut, do not add. Sleep on it. Look at the page in the morning. The fewer-than-half you're left with is, almost always, the real product. The rest was the marketing brief, dressed up.

The directory, in case it helps

The directory's landing-page collection is fifteen real SaaS landing pages, each broken down by what they cut and what they kept. The pricing-section folder covers the four patterns SaaS pricing converges on, the three details that move conversion, and the four mistakes that quietly kill it. The hero folder has the variants — centered bold claim, split with product shot, animated loop, bento — and a tuned prompt for each.

If you read the SaaS-specific brief on the /for/saas page, it'll point you at the same six collections in the order that helps. Same directory. Different doorway. Depending on how you got here.

Priya emailed me last week. She's hiring her first marketing person. The job description, which she sent me as a draft, has a line near the bottom that says: "comfortable cutting more than they add." I asked her where she got that. She said: "from the Thursday night."

I am keeping that one.