The landing page We rebuild every quarter.

• 01Framer — host + edit in productiondaily
• 02Figma — copy + designweekly
• 03Claude — copy passesweekly
• 04Plausible — analyticsdaily
• 05Tally — embedded formsmonthly

1. 0115 min

   Read the 90-day analytics.

   Open Plausible. Look at scroll depth, time on page, click-through to the CTA. Find the dead zones — the sections where most readers stop scrolling or leave the page.

   Dead zones are not bad sections. They're sections that aren't earning their place. Mark them. They're the cut list.

2. 0245 min

   Write the new copy in Figma — full doc, no design.

   Open a fresh Figma doc. Write the whole page as plain text — headline, subhead, sections, CTA, footer. No styling. No images. Just words on white.

   If the page reads well as text, it'll convert as a page. If it doesn't read well as text, no amount of design will save it.

3. 0320 min

   Cut 30%.

   The rule. Every rebuild. Read the doc top to bottom and remove 30% by word count. The cuts come from: section preambles, hedges, second adjectives, throat-clearing in the hero.

   If the cut hurts to make, it's probably the right cut.

4. 0430 min

   Rebuild in Framer.

   Open last quarter's page. Swap the copy section by section — don't move the structure. The structure works. Update one hero image if it feels dated. Update nothing else visually.

   Resist the urge to "modernize" the design. The page reading better is the modernization.

5. 0510 min

   Ship on a weekday morning. Watch five days.

   Ship on a weekday morning. Avoid the end of the week. Watch the Plausible numbers daily for five days. After five days, iterate the hook only — not the body. The body is fine; you just rewrote it.

• ×Rebuilding from scratch. The skeleton works. Replace the words.
• ×Multi-CTA pages. One primary, one secondary. Three is noise.
• ×Quotes without names. If you can't put a face on it, leave it out.
• ×Hero videos that take 3s to start. Static beats slow.
• ×"Comparison tables" with competitors. Punching down kills trust.
• ×Quarterly redesigns. Quarterly rewrites. Different thing.

The skeleton is forever. The words are quarterly. Stop rebuilding the wrong layer.

If you steal one thing — the "write the page in Figma as plain text first" rule. It exposes the words that weren't carrying their weight, every time.

After the first five days of clean data, run a headline-only A/B for two weeks. Same body, two openers. The winner becomes the published headline; the loser becomes the contender for next quarter.

A thin bar that appears after the user scrolls past the hero. Price + primary CTA. Conversion lift is small but consistent — ~5–8% on long pages.

When the page is linked from email, pre-fill the form with the visitor's name and source. Tally supports this natively. Form completion goes up because the friction drops to one field.

Once a year, ask if the page is in the right funnel position. Sometimes the right move isn't a better page — it's pointing traffic at a different page entirely. The rebuild is local; the position is strategic.