Week 1: AI Marketing with Rezerv – Paid Ads

Singapore: 12.5% CR. Philippines: $20.53 CPL. Room for Improvement.

Kicking off something new.

Each week, I’ll document how I’m using AI as a marketing co-pilot at Rezerv.

Biggest mistake this week?
Step 3 – Drafting the Landing Page.

Started with cold traffic paid ads on Meta.

• SG hit 12.5% CR (solid).

• But 2 countries got zero leads (not great).

Here’s the actual system I built to launch and improve paid ads with AI at Rezerv.

---

Why Paid Ads for Rezerv?

• Churn < 5% → Once a studio is onboarded, they rarely leave

• Strong LTV → Most stay for over a year

• Demo-to-paid ~50% → Product delivers, SDRs convert. The funnel works when people enter it

But Here’s the Gap:

• Homepage branded/organic traffic converts at 0.5%

That’s super low. And to me that’s not a traffic issue. It’s a positioning and copy issue.

Why Paid Ads Make Sense:

They let us control the narrative, tailor messaging by ICP, and drive high-intent traffic into a dedicated funnel that actually converts.

---

System Snapshot: AI-Powered DR Ad Loop

Objective
Build, test, and iterate cold traffic funnels using AI tools and DR frameworks.

Inputs

• Psychographic ICP profile (via NotebookLM + Perplexity)

• VoC clusters from G2, Capterra, Trustpilot

• Direct response copy frameworks (PAS, AIDA, Enemy Hook, etc.)

Process

1. Research and extract ICP fears/desires

2. Cluster VoC to identify dominant pain

3. Feed pain into ad copy + visual prompt generator

4. Route high-performing hooks to a pain-specific LP

5. Measure CTR/CR + annotate feedback

6. Update prompt framework or LP structure based on weak point

Results (7 Days)

• 3%+ CTR on cold ads (signal: hooks/visuals resonated)

• <2% cold traffic → LP → Booked Demo CR (signal: landing page mechanism weak)

---

Inside Step 1: Deep ICP Research

• Open NotebookLM

  → Prompt: “I want to research first-time fitness studio owners and build a psychographic profile.”

• Review the sources pulled
  → Prioritize articles, interviews and YouTube transcripts from studio founders

• Extract patterns
  → Fear: tech overwhelm, data loss, churn
  → Wants: peace of mind, bookings on autopilot, human support
  → Drivers: full classes, clean UX, control

🧠 Why: Your ICP’s pain will shape the landing page, ad copy, and ad visuals. Paid ads only work if the message hits where it hurts.

Inside Step 2: Pulled VoC Data

• Go to G2, Capterra, or Trustpilot → search competitors

• Copy reviews manually or scrape with Apify

• Drop all reviews into a Google Sheet

• Upload the sheet into ChatGPT
  → Prompt: “Cluster the top recurring pain points. Prioritize by intensity and frequency.”

🧠 Why: VoC (Voice of Customer) data strengthens your ICP research and reveals the one pain your audience actually cares most about 

Inside Step 3: Drafted Landing Page (Biggest Mistake)

I made a classic mistake here.

I asked ChatGPT to write the headline using “DR best practices” but didn’t give it actual structure or examples.

What went wrong:
❌ No copywriting framework
❌ No guidance beyond “use the main pain”

The result? Decent copy, but it missed critical elements:

• Big transformation

• Specific timeframe without common pain

• Clear mechanism

What I should’ve done:

• Given this structure: Get [Transformation] in [Timeframe] Without [Pain] — Powered by [Mechanism]

• Fed it proven headline examples

This is what happens when you ask AI for “DR copy” without giving it an actual framework.

Inside Step 4: Generated Ad Copy

• Used ChatGPT to generate 15 cold traffic hooks across Pain, Enemy, Curiosity, and Identity types

• Paired each hook with body copy using PAS, PASTOR, or AIDA frameworks

• Shortlisted top 3 based on clarity, emotional intensity, and angle-message match

For the copy, I edited key elements like the opening hook and audience callout to sharpen the message.

🧠 Why: These hook types and copy frameworks are generally proven to convert cold traffic. 

Inside Step 5: Generated Visuals

• Used ChatGPT to brainstorm 1–2 visual ideas per top hook (e.g. WhatsApp bookings, admin chaos)

• Generated the images directly in ChatGPT

• Made minor edits in Canva for brand consistency (top = AI, bottom = edited)

🧠 See the actual copy + visual used:here & here

What the Data Told Me?

• Copy and visuals hit → CTRs across most markets were 3%+, which means the pain-based messaging resonated

• LP mistake hurt conversions → missed key DR elements (no big transformation, no clear mechanism), which likely caused the drop to <2% Booked Demo CR in most markets.

Perplexity’s take on the average cold traffic to booked demos conversion rate.

Next Steps

• Fix the landing page
  (Already done. Details coming in Week 2 breakdown)

• Broaden the ICP to increase spend
  Meta campaign is underspending

  → Instead of targeting Pilates studio owners only, expand to fitness studio owners
  → Update LP and ad copy to match the broader audience

• Test different ad angles and copies

  → Launch five AI-generated angles based on VoC (Zero-Admin, Price-Slash, Migration, Human Support, Revenue Boost), run 3-day split tests, kill any ad below 3 % CTR, and scale the winners.

• Plan an AI evaluation workflow to catch mistakes

  → Build a prompt-based checklist agent to review LPs and ad copy

Wrap Up: My Journey Using AI for Marketing

Just an honest and transparent sharing of how I actually used AI to solve real marketing problems.

Because I think we need more open, practical sharing in this space.

I test AI workflows on real marketing problems.

And turn the useful ones into 10x Playbooks.

So business owners and solo marketers doing it all themselves can skip the guesswork… and copy what actually works.

→ I drop one of these 10xPlaybooks 🚀 every week, don’t miss the next.

John