How We Achieved a R68 CPA Across 5,000 Clients Using Lead Ads
This isn't a theory piece. This is an accounting of how we spent R205,000 of our own money on Meta lead ads in the insurance vertical — one of the most restricted categories on the platform — and generated 5,000 subscription clients at an average CPA of R68.
We're sharing this because most marketing content is either too vague to be useful or written by people who've never spent their own money. When it's your own money, you optimise differently. You're brutally honest about what's working and what isn't because waste comes out of your pocket.
The Starting Point
Industry: Insurance (life and funeral cover) Category: Restricted — Special Ad Category for Credit Starting budget: R5,000/month Final monthly run rate: R25,000+/month Total spent: R205,000+ Total subscription clients acquired: 5,000+ Average CPA across the full period: R68
Month 1–2: The Expensive Education
What We Did
- Launched 3 campaigns with interest-based targeting
- Created 5 ad variations (static images with different headlines)
- Used Meta's native Lead Forms
- Targeted all of South Africa, ages 25–55
What Happened
- CPA: R115–R140
- Lead quality was mixed — roughly 40% of leads didn't answer the phone
- Interest targeting was severely limited due to Special Ad Category restrictions
- We burned through about R15,000 learning what doesn't work
What We Learned
Interest targeting in restricted categories is nearly useless. Meta removes most meaningful interest options, leaving you with broad demographics. The CPA was too high because the algorithm had no clear signal of who to target.
Month 3–4: The Creative Breakthrough
What We Changed
Instead of trying to fix targeting (which was restricted), we focused entirely on creative. We wrote ad copy that functioned as a targeting mechanism:
Before (generic):
"Get affordable life cover. Apply today."
After (self-selecting):
"If you're the main breadwinner and your family depends on your income, you need to read this. Most South Africans are one bad day away from financial disaster. R350/month can change that."
The second version doesn't need demographic targeting. It self-selects: breadwinners, people with dependents, people anxious about financial security.
What Happened
- CPA dropped to R75–R85
- Lead quality improved dramatically — contact rates went from 40% to 65%
- We could tell from the leads' responses that the ad was attracting the right people
The Insight
In restricted categories, creative IS targeting. The algorithm can't find your ideal customer through data restrictions, so your ad copy and imagery must do the filtering. This single shift reduced our CPA by 35%.
Month 5–8: The Lookalike Engine
What We Changed
By month 5, we had accumulated ~800 converted subscription clients. We uploaded this customer list and created a 1% Special Ad Audience (the restricted-category version of a Lookalike).
We also implemented:
- Conversions API for server-side tracking
- Value-based custom audiences (weighted by customer lifetime value)
- A retargeting sequence for lead form openers who didn't submit
What Happened
- CPA dropped to R55–R65
- The Special Ad Audience outperformed all interest-based targeting by 40%
- Retargeting converted at 3x the rate of cold campaigns
- Monthly volume scaled from ~200 leads to 600+ leads
The Insight
First-party data is the weapon in restricted categories. Every customer you acquire makes your next campaign cheaper. By month 5, our data advantage was compounding and competitors running generic interest targeting couldn't match our costs.
Month 9–12: Optimisation and Scale
What We Changed
- Increased creative testing to 8+ new variations per month
- Segmented campaigns by province to identify geographic performance differences
- Introduced video testimonials (using customer stories with consent)
- A/B tested lead form length: short (name + phone) vs detailed (name + phone + age + income bracket)
What We Found
- Short forms: More leads, lower quality, R50 CPL
- Detailed forms: Fewer leads, much higher quality, R85 CPL, but higher conversion to subscription
- Net CPA was lower with detailed forms because fewer leads were wasted in the sales process
- Video testimonials outperformed static creative by 45% on conversion rate
- Gauteng and KZN were the highest-performing provinces; Western Cape had the highest CPAs
Final Numbers
- Average CPA across all 12+ months: R68
- Total clients acquired: 5,000+
- Best month CPA: R42
- Worst month CPA: R95
- Average CTR: 4.31%
- Lead-to-client conversion rate: ~35%
The Full System
Achieving R68 CPA wasn't just about the ads. It was a system:
1. Ad Creative (Updated Monthly)
8 new variations per month. Mix of static, carousel, and video. Every ad designed to self-select the right audience.
2. Lead Form (Optimised for Quality)
Qualification questions included in the form to filter out casual clickers. This increased CPL slightly but dramatically improved conversion rates.
3. Instant Contact
Automated WhatsApp message sent within 60 seconds of lead submission. Human follow-up within 15 minutes during business hours. This single automation improved lead-to-sale conversion by 40%.
4. Retargeting
Anyone who opened the lead form but didn't submit was retargeted within 24 hours with a different creative and a "still thinking about it?" angle. This recovered 15–20% of lost leads.
5. Conversions API
Server-side tracking ensured Meta had accurate conversion data. This improved campaign optimisation and reduced CPA by an estimated 20–30% compared to Pixel-only tracking.
6. Data Feedback Loop
Converted customer data was fed back to Meta monthly to refresh Lookalike/Special Ad Audiences. Each refresh improved targeting quality.
What We'd Do Differently
Start With the Conversions API From Day One
We installed it in month 4. The first 3 months of data loss from browser-only tracking cost us in optimisation quality. If we'd had server-side tracking from the start, the learning phase would have been shorter and cheaper.
Test Video Earlier
We didn't introduce video until month 9. Video testimonials were our best-performing creative. Starting video testing in month 3 could have accelerated our CPA reduction.
Segment by Province From the Start
We ran national campaigns for the first 8 months. When we segmented by province, we discovered that Gauteng's CPA was 30% lower than Western Cape. We could have been more efficient by allocating budget geographically from the start.
Use Longer Lead Forms From the Start
The short forms generated volume but wasted sales time. The net cost per acquired customer was actually lower with longer forms that pre-qualified leads. We should have tested form length in month 1, not month 9.
Key Takeaways
- R205K spent, R68 CPA, 5,000 clients — in a restricted category, with our own money
- Creative is targeting in restricted categories — write ads that self-select your audience
- First-party data compounds — every customer makes the next one cheaper
- Speed to contact is a conversion multiplier — 60-second automated response, 15-minute human follow-up
- The system behind the ads matters more than the ads — tracking, follow-up, retargeting, and data feedback loops are what turn good ads into great results
- Test constantly — our best and worst months were separated by creative quality, not budget
This is how we run campaigns for our clients too. From R4,999/month, no contracts, you own everything. Get your free game plan.