Ecommerce & D2C in Pakistan: Performance Marketing When COD Dominates

In most markets a conversion is revenue. In Pakistan it is a request for revenue that a customer can refuse at their front door. With cash on delivery still accounting for somewhere between 55% and 80% of ecommerce orders, and return rates running 25–35%, the single most expensive mistake a Pakistani D2C brand makes is optimising performance marketing campaigns on orders rather than on delivered, collected orders. Every downstream decision inherits that error.

A spoke of Digital Marketing in Pakistan. This page covers ecommerce and D2C; the hub covers the market across 25 industries.

1. The conversion that is not revenue

Performance marketing rests on a assumption so basic that nobody states it: that a recorded conversion corresponds to money received. Break that assumption and every mechanism built on top of it inverts. Bid strategies chase the wrong audiences. Lookalike models train on people who never paid. Return on ad spend reports a number that your bank account contradicts.

Cash on delivery breaks that assumption comprehensively. The order is recorded, the pixel fires, the campaign is credited, the algorithm learns — and then a courier knocks, nobody answers, or somebody answers and declines the parcel. In Pakistan that happens to roughly a quarter to a third of orders.

Your ad platform is optimising toward whoever is most willing to click “order”. In a COD market that is not the same population as whoever is most willing to pay.

Why this compounds rather than averages out

If refusals were random, you could simply discount revenue by the return rate and carry on. They are not random. The audiences most likely to place a frictionless COD order — impulse-led, low-intent, price-driven — are also the audiences most likely to refuse it. So the algorithm does not just include bad orders in its training data; it actively concentrates spend toward the segment that produces them.

That is why brands report the pattern of a campaign that “worked brilliantly for a month and then stopped”. It did not stop. It got better at finding the customer it was told to find.

2. The real unit economics of a COD order

Before any channel decision, the arithmetic needs to be honest. A returned COD order does not cost you nothing. It costs you forward shipping, return shipping, handling, the COD collection fee, packaging, and the working capital tied up in the interim — against zero revenue.

Structural cost comparison based on standard Pakistani courier COD arrangements. Exact fees vary by courier and volume — substitute your own contracted rates before modelling.

The number you actually need

Contribution margin per delivered and collected order, not per order. Once you have it, the maximum acceptable cost per acquisition follows directly, and the entire campaign structure can be judged against a real ceiling instead of a reported ROAS that includes parcels sitting in a returns warehouse.

Illustrative calculation using published Pakistan return-rate estimates of 25–35%. The PKR 700 base is an example, not a benchmark — substitute your own reported CPA.

At a 30% return rate, a reported acquisition cost of PKR 700 is really PKR 1,000. If your margin was built on the first number, you are losing money on every order and your dashboard says you are winning.

3. What to optimise campaigns on instead

The fix is not clever bidding. It is changing what you tell the platform to count as success.

Operational guidance based on how conversion-signal quality affects algorithmic learning. Feasibility depends on your ability to upload offline conversions back to the ad platform.

The delay problem, and how to live with it

Delivered-and-collected is the honest signal, but it arrives days after the click, and platforms learn poorly on long delays with low volume. The practical compromise most Pakistani D2C brands land on is a two-tier setup: use confirmed orders as the primary optimisation event because it is fast and correlates well with delivery, and upload delivered-and-collected data on a regular cycle to correct the model and to audit which audiences are genuinely profitable.

4. Moving buyers from COD to prepaid

Because prepaid orders return 40–50% less often, converting COD buyers to prepaid is the highest-leverage optimisation available to a Pakistani ecommerce brand. It improves margin twice: collection becomes certain, and the return that would have happened does not.

With JazzCash and EasyPaisa holding more than 50 million accounts between them and Raast expanding instant payments, the infrastructure is no longer the obstacle. Trust and habit are.

Levers observed across Pakistani ecommerce operations. Impact varies by category and price point; test before rolling out, particularly the surcharge.

How to price the incentive

The discount you can afford is bounded by what a return costs you. If a refused order costs you forward shipping, return shipping, handling and packaging, then a prepaid discount worth less than that expected loss is profitable even if it converts nobody who would have paid anyway. Most brands set this by instinct and set it too low.

A prepaid discount is not a discount. It is an insurance premium against a return, and it should be priced like one.

5. The channel mix that works

Pakistan is a discovery-led market before it is a search-led one. With a median age of about 22 and roughly 60% of the population under 30, demand is frequently created in a feed rather than captured in a search box — which inverts the usual channel sequencing.

Channel roles based on Pakistan platform audience data (Facebook the largest ad audience; Instagram and TikTok skewing younger) and observed operating patterns. Prepaid-propensity rankings are directional and should be validated against your own order data.

WhatsApp is a performance channel here, not a support channel

The single most underrated intervention in Pakistani ecommerce is a fast WhatsApp confirmation on every COD order. It converts an anonymous click into a human commitment, and a customer who has replied to a message is materially less likely to refuse the parcel. Treat it as part of the funnel with a staffing model and a response-time target, not as an inbox someone checks.

6. Social commerce versus your own store

Social commerce is projected to account for up to 35% of Pakistan’s total online retail sales in 2026 — selling directly through Instagram and Facebook without a separate website. For a new brand it is the fastest route to first revenue. It is also a route with a ceiling.

Social commerce share projection: up to 35% of Pakistan online retail sales in 2026 (published 2026 Pakistan ecommerce reporting). Comparison rows are operational judgement.

The sensible answer is not either. Social commerce is where discovery and impulse purchase happen; the store is where repeat purchase, customer data and prepaid conversion live. Brands that refuse to build the second one stay permanently dependent on paid reach to generate every single order.

7. Daraz versus D2C: the marketplace decision

Daraz remains the anchor of Pakistani ecommerce — roughly $856 million in revenue in 2025 and over 9.4 million active users, with fashion and apparel around 28% of platform sales and electronics around 24%. The marketplace question is therefore unavoidable.

Daraz figures: approximately $856M revenue in 2025 and 9.4M+ active users; category mix approximately 28% fashion and apparel, 24% electronics (published 2026 Pakistan ecommerce reporting). Comparison rows are operational judgement.

8. Fixing the 71.5% abandonment problem

Pakistani cart abandonment runs around 71.5–72%, meaning roughly three in four shoppers who reach checkout leave without ordering. Because most brands treat abandonment as a checkout problem, most abandonment fixes fail — a good share of it originates earlier, in unclear delivery expectations and unclear total cost.

Cart abandonment rate of 71.5–72% per published 2026 Pakistan ecommerce reporting. Cause attribution is operational judgement based on common audit findings.

With over 85% of orders placed on mobile, every one of these is a mobile problem first. A checkout that works on desktop and is merely tolerable on a mid-range Android phone is, functionally, a broken checkout.

9. Delivery coverage as a targeting parameter

Logistics outside the major cities remains the binding constraint on Pakistani ecommerce, and the 3PL market — estimated at $200–300 million and growing above 35% annually — is still building out. This has a direct and usually ignored marketing implication: your delivery reliability map should be a targeting input.

Advertising into an area where your courier performs badly generates orders that convert into returns. That is worse than not advertising there, because you pay for the click, the forward shipping and the return shipping in exchange for nothing.

Marketing can create demand in a city you cannot reliably serve. That is not growth. It is a return rate with a media budget attached.

The practical version

Pull delivery success rate by city or postcode from your courier data, rank it, and treat the bottom band as either excluded or prepaid-only. Most brands have never looked at this data at the geographic level and are surprised by how concentrated the failures are.

10. The 90-day implementation plan

The order matters. Fixing measurement before spending more is the difference between scaling a business and scaling a loss.

Indicative sequencing based on operating experience. Durations vary with team size and technical resource; the ordering matters more than the timing.

11. Mistakes to avoid

Recurring errors observed in Pakistani D2C operations; illustrative rather than surveyed.

12. What changes in 2027

Prepaid share keeps rising. With Raast expanding and wallet accounts already above 50 million, the COD share will keep eroding — though industry commentary suggests a five-to-seven-year transition rather than a sudden shift. Brands that build prepaid conversion capability now compound the advantage.

Social commerce consolidates. At a projected 35% of online retail sales, social selling stops being an experiment and starts attracting the operational discipline — and eventually the platform fees — of a mature channel.

Logistics coverage improves unevenly. A 3PL market growing above 35% annually will widen reliable delivery, but unevenly. The geographic gap between where you can advertise and where you can deliver will narrow without closing.

Key Takeaways

• A COD conversion is not revenue. With returns at 25–35%, optimising on orders placed trains the algorithm toward buyers who do not pay.
• At a 30% return rate, a PKR 700 reported CPA is really PKR 1,000. Model against delivered and collected orders only.
• Prepaid orders return 40–50% less often, making COD-to-prepaid conversion the highest-value optimisation available.
• Price the prepaid discount as insurance against the cost of a return, not as a giveaway. Most brands set it far too low.
• WhatsApp confirmation is a performance channel in Pakistan, not a support function — it converts a click into a human commitment.
• Cart abandonment runs 71.5–72% and starts on the product page, not at checkout, on a market where 85%+ of orders are mobile.
• Your courier’s delivery success by city belongs in your targeting — advertising where you cannot deliver buys returns.

Frequently Asked Questions

Conclusion

Pakistani ecommerce is often described as difficult because of cash on delivery. It is more accurate to say it is difficult because most brands run it on measurement borrowed from markets where COD does not exist. The mechanics of performance marketing are the same here; the definition of a conversion is not.

Fix that definition and the rest follows in order: honest margin, a prepaid incentive priced against real return costs, WhatsApp confirmation as a funnel stage, targeting that respects where you can actually deliver. None of it is exotic. It is simply the version of performance marketing that survives contact with a market where the customer pays at the door.