Podcast One-Pager · Paid Media

The Right Way to Test Landing Pages on Meta Ads in 2026

The Andrew Faris Podcast · guest Brad Plock (Scalability School / AJF Growth) · ~54 min · Jun 22, 2026

The one idea

Stop testing landers as "which page wins?" (either/or). Test them as "can this page unlock more total spend at the same efficiency target?" (AND). A landing page is just another asset to find new customers — like new creative. You're looking for more spend, not better spend.

◆ The mental-model shift

  • Split tests off Meta traffic are noisy — rarely reach significance, and "winners" almost never hold when implemented.
  • Don't pick lander vs PDP. Different buyers need different paths — the "marketing tornado" (credit: Shreen): right person × right time × right funnel path.
  • Valid paths coexist: ad → PDP, ad → collection, ad → lander, ad → quiz. Keep them all available.
  • A win shows up 3 ways: more total spendbetter efficiencyreaches a new sub-segment

◆ The SOP — how to actually test

  1. Work inside the existing product-specific CBO, one ad set, ads jammed in for scale.
  2. Pick the top-performing ad. Note who responds and why.
  3. Duplicate it, swap the URL to the new lander. Publish. Leave both live. Don't replace or edit the original.
  4. Let Meta allocate. No forced spend, no min budgets.
  5. Read it like new creative: if the lander earns incremental spend, it found buyers you couldn't reach.

? "Won't Meta just ignore the new URL?"

  • The fear: Meta IDs the duplicate as the same ad and delivers on upper-funnel signals, ignoring the lower-funnel lander.
  • Reality (per a Meta rep): Meta already crawls the landing page (for health/policy). That page difference is enough to differentiate delivery.
  • Bonus: an ad with zero spend on the PDP became the #2 spender once pointed at a matched lander.

↗ Proof — women 45+ appliance case

+$1k/day
Total campaign spend ($5k → $6k), target held
~20%
Share the lander took — incremental, not stolen
90% → 70%
Bounce PDP → lander (confirming signal)
  • Delivery got more efficient for the exact sub-segment (45–54) spoken to, then scaled; other ages unchanged.
  • CPMR held/improved while spend doubled over 2 months (lagging, not predictive).

⚖ Duplicate-ad method vs. split test — pick by what changes

▶ DUPLICATE THE AD when the change reaches a different person

  • New persona, new offer, or new funnel path
  • Listicle → advertorial; PDP → pre-sell lander
  • Anything that changes who you reach → it's an AND

▶ SPLIT TEST (IntelliGyms) when the same person stays

  • On-page optimization only: 5 → 7 reasons why
  • Buy-box tweaks, adding a shoppable section
  • Marginal lift, same audience → an OR a split test answers well

⌗ Media-buying placement

  • Same ad set vs. new ad set in the same CBO: doesn't matter much. A same-targeting ad set is "just a folder" — use one only to label the test.
  • Ensure enough volume for Meta to optimize within the ad set.
  • Be wary of separate forced-spend testing campaigns: only if you're OK spending dollars that may create no value. Prefer letting the machine decide.

◎ Where to test first

  • Start with highest-spending ads — clearest signal to extend with a matched lander.
  • Use Motion AI tags to find the biggest persona/message angle in aggregate (sometimes ≠ the top ad).
  • Build the lander around one ad / one message — never average three top ads into one page.
  • A seasonal angle is worth a lander even for ~3 months if it lifts efficiency.

✎ Landing-page design principles

  • Match the scent. Ad hook = lander headline = images = reasons = persona, end to end. Aesthetics ad → comfort page = you've lost.
  • Start with listicles (easiest): images + copy + a comparison chart vs. alternatives. Copy matters most.
  • Don't let AI write all the copy — weak without thoughtful instructions. AI to enable, human to nail the message.
  • Advertorials are harder, need a defined audience. Best trick: founder-video transcript → advertorial in the founder's voice.
  • Lander → PDP is a fine start (pre-sell / "squeeze" page). Add a shoppable section later; one extra click won't kill you.
  • One SKU can carry many personas/pages (Mars: ~30 pages, 1 SKU). Make PDP images persona-specific.
  • Install Microsoft Clarity (free) and watch recordings — e.g., nobody started a quiz → added a "Shop the product" button → +20% revenue-per-session.

" Lines worth stealing

"You're trying to find more spend, not better spend."
An ad set with the same targeting? "It's a folder."
The message grabs attention and interest; the offer closes the deal — own both across the funnel.