Why PKR ROAS doesn't behave like USD ROAS
The instinct is to apply Western playbooks because that's where the public case studies come from. The math doesn't translate. The Pakistani buyer's funnel is different in three specific ways.
Trust starts much lower at the top of the funnel. Cash-on-delivery is still 60-75% of online transactions in most categories. Returns culture (largely thanks to Daraz) trains buyers to expect the option of opting out. Skepticism toward online-only brands is the default, not the exception. You can't drive a cold visitor to a checkout button on the first session and expect it to work.
WhatsApp is where most conversions actually close. Especially in F&B, real estate, salons, fashion, and SaaS, the "ad click" is the start of a DM conversation, not a checkout. Your funnel architecture has to accommodate this. If you only measure on-site purchases, you're systematically under-attributing your ad spend.
Mobile-first means data-conscious. Most Pakistani users browse on 4G with monthly data caps. Heavy autoplay videos, unoptimised hero images, and 8 MB JavaScript bundles cost you the click before your landing page renders. The brands winning on Meta in Pakistan have landing pages under 1.5 seconds on Slow 3G — not as an aspiration, as a constraint.
Where to spend your first PKR (and where not to)
Don't run ads to a cold audience until your retention layer is built. The sequence matters and skipping any step compounds the waste downstream.
- Pixel + event tracking, properly configured. Meta Pixel + Conversions API. GA4 with proper events. Don't trust the auto-install — verify each event fires with real values in Events Manager. Skip this and every dollar after is flying blind.
- A landing page that loads under 2 seconds on Pakistani 3G. Test with throttling enabled in DevTools, not just on your office WiFi. Lazy-load videos, compress images to WebP, defer non-critical JS.
- WhatsApp / CRM integration. If your funnel ends in a conversation, automate the handoff. WhatsApp Business API + a CRM that logs the conversation source.
- Retargeting your existing engagement first. Page likes, IG followers, video viewers from the last 12 months. Tight budget (PKR 500-2000/day). Get a real ROAS baseline before you scale anything.
Only after these four steps should you point spend at cold lookalikes. Brands that skip this typically waste 30-50% of their first three months optimising into noise.
The funnel: TOF, MOF, BOF for the Pakistani buyer
Three layers, each with its own job and its own measurement bar.
Top-of-funnel (TOF). Educational plus entertainment. The objective is reach + ThruPlay, not conversions. CPM in Pakistan is cheap relative to global benchmarks — use that to over-feed the next layer. Audiences: 1% lookalikes from purchaser data, plus broad interest-based with proper exclusions. Healthy benchmark: ThruPlay rate >25%, video view cost under PKR 0.5.
Middle-of-funnel (MOF). Social proof + product-led. Retarget video viewers (75%+) and page engagers. Concept: "you've seen us, now here's the case for taking us seriously." Click-through to product page should land under PKR 8 CPC for most categories. If yours is higher, the creative is the problem — not the targeting.
Bottom-of-funnel (BOF). Direct response + urgency. Retarget add-to-cart and initiate-checkout audiences with explicit offers. Goal: cost-per-purchase under AOV × (1 / target ROAS). For most Pakistani D2C brands targeting 3x ROAS on a PKR 2,500 AOV, that means cost-per-purchase under PKR 833. If you're above, the bottleneck is usually checkout friction, not ad cost.
Creative principles that win in this market
Specific things we see win, repeatedly, across categories:
- Urdu or Punjabi hooks in the first second. Even one phrase. English-default scrolling is broken by the cognitive shift — the viewer pays attention because something feels addressed to them.
- Real local context. Karachi traffic, a Lahore street, a Multan kitchen. Stock library imagery underperforms native-shot content by 2-3x on ThruPlay, in our experience.
- Price call-out within the first 2 seconds. "Sirf PKR 1,499" consistently outperforms "Limited time offer" for D2C product ads. Pakistani buyers price-shop hard; surfacing it early earns you the attention.
- WhatsApp number on-creative. Even when your funnel ends on-site. Doubles the inbound DM volume and your highest-intent conversations start there.
- Faces beat product shots. A real person holding the product, talking to camera, in their environment — outperforms studio product shots almost universally.
Measurement: what's different in PKR
ROAS in isolation is misleading. Meta and Google attribute conversions differently and overlap heavily in Pakistan because most users are on the same handful of platforms. Track blended ROAS: (Meta revenue + Google revenue + organic revenue) / total ad spend. The number that matters is the one your accountant sees, not the one Meta's dashboard shows.
WhatsApp is dark traffic. Use UTM-tagged links in CTAs (?utm_source=meta&utm_campaign=...) and have your WhatsApp Business account log conversation sources. For the first 90 days, track manually: ad spend → conversation count → close rate → revenue. Once you have that ratio, you can extrapolate.
Optimise against high-intent actions, not vanity metrics. Define what "high intent" means for your funnel — add-to-cart, sticker-tap, initiate-checkout, DM-click — and optimise campaign objectives against it. Impressions and CTR are flattering and useless.
Common mistakes (and how much they cost)
- Running Performance Max before having pixel data. Google's algorithm will dump 30%+ of budget into non-converting placements (display network apps you've never heard of). Wait until you have 50+ conversion events to feed it.
- "Boosting" Instagram posts as a strategy. No funnel, no retargeting layer, no learning. You're paying Meta to show your worst content to the wrong people.
- Copying a US agency's creative format. Talking head, English-only, no local context. Burns through TOF budget without engaging anyone.
- Setting ROAS targets that ignore COD return rates. If 20% of your COD orders are refused at the doorstep, your reported 4x ROAS is actually 3.2x. Calculate against net revenue, not gross.
- Pausing campaigns based on 3-day data. Meta needs at least 7 days to optimise; killing a campaign in 72 hours just throws away the learning phase. Set a 14-day evaluation window and stick to it.
Pulling it together
The structure isn't complicated. What's hard is the discipline to set up the foundation before you turn the spend on, and the patience to measure correctly when your dashboard isn't telling you the full story. The brands we see win in Pakistan are the ones treating their funnel like an engineering problem — specific inputs, specific outputs, specific tests — not a creative-led campaign series.
If you want a second pair of eyes on your account, we'll do an audit of your current setup at no cost. We've run these structures for D2C brands, F&B, SaaS, fintech, real estate, and salons across Pakistan. We'll tell you what we'd change, in order, with the reasoning — whether or not you ever work with us.