Marketing

Small Meta Ad Sets With 1-2 Creatives and Claude-Built Landing Pages: A Shopify and Amazon Seller's Way Through the Meta Ads Turbulence

  • Level Beginner
  • Setup~30 min (estimated)
  • Evidence Anecdotal
  • CostNot stated in source
  • No-codeYes
ClaudeSource-linkedBeginner-friendly

What they did

A Shopify and Amazon FBA seller switched to small, tightly targeted Meta campaigns with interest targeting instead of broad, and small-budget ad sets with 1-2 creatives each. He also replaced his PageFly landing pages with 2-3 Claude-built ones tailored to different audiences, with faster-loading images and video for desktop and mobile. He gives no performance numbers.

How it went

X post by @sandiegocausa (Oct 30, 2026 UTC): small interest-targeted Meta ad sets plus 2-3 Claude-built landing pages replacing PageFly; no results numbers given. No results shared. Small ad sets can take longer to exit Meta's learning phase, so judge them over a week or more

Worth knowing

  • X post by @sandiegocausa (Oct 30, 2026 UTC): small interest-targeted Meta ad sets plus 2-3 Claude-built landing pages replacing PageFly; no results numbers given.
  • No results shared. Small ad sets can take longer to exit Meta's learning phase, so judge them over a week or more.

Prompt

Cleaned version rewritten from the source — not a direct quote.

I sell [product] and run Meta ads. Help me plan [2-3] small ad sets, each with 1-2 creatives, using interest targeting for these audiences: [audiences], with a daily budget of [$] each. Then build a fast-loading landing page for each audience, matched to its ad, and optimized for mobile. Show me everything before anything goes live.

Recommended

Try it in Grok Bot

Grok Bot can do this too: ask it to draft the ad sets and build the landing pages as Pages for you to review.

  1. Copy the prompt (button below).
  2. Open Grok Bot chat and paste it. Replace any [bracketed] placeholders.
  3. Connect the listed tools (as needed) as connectors when Grok Bot asks.
  4. If the task repeats, ask: "Make this a daily routine."

Alternatives: the same prompt works in Claude, ChatGPT Agent, OpenClaw.

Set it up

Stack: Claude

  1. Read the original source for full context.
  2. Connect the tools your agent needs.
  3. Paste the prompt and replace [bracketed] placeholders.
  4. Run once manually and check the output before scheduling it.

Generic suggested steps, not from the source.

Evidence

X post by @sandiegocausa (Oct 30, 2026 UTC): small interest-targeted Meta ad sets plus 2-3 Claude-built landing pages replacing PageFly; no results numbers given. No results shared. Small ad sets can take longer to exit Meta's learning phase, so judge them over a week or more

Source: X · 2026-10-30 · Verified by us: not yet · Report broken or outdated

Related use cases

MoneyWeb · 2026-02

Check a hospital bill against what Medicare would pay

A writer gave Claude the coded hospital claim (UB-04) for his late brother-in-law's $195,628 ER bill and asked it to build a spreadsheet of Medicare rates per code. Claude flagged bundled items billed separately, a bypass charge for a procedure that never happened, and a ventilation charge Medicare does not allow.

  • Level Intermediate
  • Setup1–3 hrs*
  • Evidence Strong evidence
ClaudeSource-linked
ResearchReddit · 2026-10

"Eternal Star / We Hope So": Claude Decoded Super Mario 64's Unreadable 32x32 Courtyard Plaque After 30 Years

The plaque in Super Mario 64's castle courtyard is a 32x32-pixel texture fans have argued about since the 1990s. No upscaler can read it. Claude built a model of how a 1996 artist likely made it (font, canvas, resize, blur, stone), searched fonts, sizes and words on a GPU, and tested the method on fake plaques with known text. Its best reading: "Eternal Star / We hope so" in Palatino Bold Italic. Code and a 40-page report are public.

  • Level Advanced
  • Setup1–3 hrs*
  • Evidence Strong evidence
ClaudeSource-linked
ResearchHN · 2026-10

PI Was Set to 3.1415: Claude Reviewed His 2013 Math Internship Code and Cut a Simulation's Drift 750 Times

In 2013, as a math intern, he wrote C code for a shallow-water equations scheme. This week he pointed Claude at the repo for a review. It found PI defined as 3.1415, which made a lake that should stay still slowly move; fixing it cut the drift 750 times. It also found an error norm multiplied by dx that hid errors by a factor of 50, and a sign typo in one of the paper's integrals. A better quadrature rule fixed the remaining drift and made the run 20 times faster.

  • Level Advanced
  • Setup1–3 hrs*
  • Evidence Strong evidence
ClaudeSource-linked