Lookalike Audiences Guide (2026): Meta Ads Targeting & Scaling Strategy
Lookalike audiences let Meta find new prospects who resemble your best customers. This guide covers source selection, size strategy, testing, and scaling — with Nepal market specifics.
→ Lookalike Audience dictionary · Meta CAPI Setup · Facebook Ads Nepal
How lookalikes work
Source audience (seed) → Meta analyzes patterns → Lookalike % of country population
Meta matches demographics, interests, behaviors, and engagement patterns — not just demographics you could target manually.
Source audience quality (most important lever)
| Source | Quality | Min. size |
|---|---|---|
| Purchasers (180 days) | ★★★★★ | 100+ (1,000+ ideal) |
| High-value purchasers (top 25%) | ★★★★★ | 100+ |
| Add to Cart (not purchased) | ★★★★ | 500+ |
| Email list (uploaded, matched) | ★★★★ | 1,000+ |
| Website visitors (180 days) | ★★★ | 1,000+ |
| Page engagers | ★★ | 1,000+ |
| Video viewers (95%) | ★★★ | 500+ |
Rule: Best source = people who actually paid you money.
Lookalike size selection
| Size | Reach (Nepal ~30M pop.) | Use case |
|---|---|---|
| 1% | ~300,000 | Highest quality prospecting |
| 2–3% | ~600K–900K | Scale when 1% CPA is profitable |
| 5% | ~1.5M | Broad prospecting with strong creative |
| 10% | ~3M | Maximum reach; quality drops |
Start 1%, expand only when CPA and ROAS hold.
Campaign structure for lookalikes
Campaign: Prospecting (Sales/Leads)
├── Ad Set: LAL 1% Purchasers — Broad (no interest stacking)
├── Ad Set: LAL 1% Purchasers — Age 25-45 (if data supports)
├── Ad Set: LAL 3% Purchasers — Broad
└── (Exclude: Purchasers, AddToCart 14d)
Campaign: Retargeting (separate budget)
├── Ad Set: Website visitors 30d
└── Ad Set: Abandoned cart 14d
Do not stack lookalike + detailed interest targeting — over-narrows in small markets.
Testing framework
Week 1–2: Launch 1% Purchaser LAL vs. broad interest control (same creative, same budget)
Measure: CPA, ROAS, frequency, CTR
Week 3: Winner scales; loser pauses. Add 3% LAL if 1% profitable.
Week 4+: Refresh source audience monthly as new purchasers accumulate.
Creative requirements for lookalikes
Lookalikes are cold audiences — they don’t know your brand.
- Lead with problem/benefit, not brand name
- Social proof (reviews, customer count) in first 3 seconds of video
- NPR pricing visible for Nepal audiences
- UGC-style creative often outperforms polished ads for cold prospecting
CAPI + lookalike synergy
Better conversion tracking = better purchaser seed = better lookalikes.
- Implement Meta CAPI
- Build Purchaser custom audience from CAPI-verified conversions
- Create 1% LAL from that audience
- Exclude purchasers from prospecting ad sets
Nepal market considerations
| Factor | Implication |
|---|---|
| Small population | 1–3% LAL may exhaust quickly — refresh creative weekly |
| COD orders | Seed from confirmed deliveries, not all orders |
| Messenger sales | Track with CAPI offline events or manual upload |
| Diaspora | Create separate LAL in US/UK/AU if shipping internationally |
| Low ESP match rates | Phone-based custom audiences often match better than email |
Common mistakes
- Lookalike from all page visitors — includes non-buyers, weak signal
- 1% + 15 interest filters — audience too small to exit learning phase
- Not excluding purchasers — paying to re-acquire existing customers
- Same creative for LAL and retargeting — different funnel stages need different messages
- Never refreshing seed — stale lookalike degrades as market shifts
Metrics to monitor
| Metric | Healthy range (Nepal e-commerce) |
|---|---|
| CPA vs. retargeting | 1.5–3× higher (expected for cold) |
| Frequency | <3 per week on prospecting |
| CTR | >1% on feed placements |
| ROAS break-even | Know your margin-based target → ROAS Calculator |
Worked example: Kathmandu D2C brand
Setup: Skincare brand, 850 purchasers in last 180 days (CAPI-verified), NPR 45,000/month Meta budget.
| Week | Action | Result |
|---|---|---|
| 1 | 1% Purchaser LAL vs interest “skincare” control | LAL CPA NPR 420 vs NPR 680 control |
| 2 | Scale LAL 50%, add 3% LAL ad set | 3% CPA NPR 510 — acceptable for scale |
| 3 | Refresh creative (3 new UGC hooks) | Frequency dropped 4.2 → 2.1 |
| 4 | Upload phone list (2,400 matched) → new 1% LAL | CPA improved 8% vs pixel-only seed |
Takeaway: Phone-matched custom audiences often outperform email in Nepal due to lower ESP match rates.
LAL + Advantage+ shopping (2026)
For catalog advertisers, test Advantage+ shopping with purchaser exclusion before building manual LAL ad sets. If ASC outperforms, use LAL only for prospecting products ASC under-indexes on.
