Abandoned Cart Recovery Case Study in Pakistan

The Client

A Karachi-based women’s apparel brand selling ready-to-wear (pret) stitched lawn suits, unstitched fabric bundles, and a growing line of accessories through their Shopify storefront. The catalog sat at around 600 SKUs, average order value was roughly PKR 6,500, and monthly revenue had reached about PKR 16M — built largely on seasonal lawn drops that drove sharp traffic spikes each quarter. Acquisition was Meta-led: roughly PKR 2.5M a month across Facebook and Instagram, supplemented by influencer seeding around each collection launch.

The brand had a healthy top of funnel. Each lawn drop pulled 140,000-plus monthly sessions, overwhelmingly mobile at 78%, and add-to-cart rates on launch days were strong. The team had invested in creative, in product photography, and in a fast checkout built around cash on delivery — the dominant payment method for Pakistani fashion ecommerce, where most buyers prefer to pay only when the parcel arrives. By every acquisition metric, the engine worked. Acquisition cost per first-time buyer was acceptable, and the creative testing programme was mature.

What did not work was what happened next. Three out of every four carts were abandoned. And once a shopper left, the brand had almost no mechanism to bring them back.

When they approached WeProms Digital, the brief was narrow and specific: stop bleeding revenue at checkout. The team had read their own analytics carefully and understood that recovering even a modest fraction of abandoned carts would compound directly into the bottom line — without spending another rupee on acquisition. They did not need a bigger funnel. They needed the funnel they already had to leak less.

The Problem

Six issues were suppressing recovery and email revenue:

1. A 72% cart-abandonment rate with no recovery engine. Fashion carts in Pakistan abandon for predictable reasons — COD hesitation (will the parcel actually arrive, can I inspect it), fabric and sizing deliberation, price comparison across three or four competing stores in another tab, and the simple habit of adding now and deciding later. The brand was losing roughly PKR 11.5M of carted merchandise value every month and following up on none of it.

2. Email contributed only 7% of revenue. Klaviyo was installed but effectively dormant. A single generic cart email fired 24 hours after abandonment — unbranded, untimed, identical for a PKR 3,000 unstitched fabric buyer and a PKR 15,000 bridal pret buyer. No welcome flow, no browse flow, no post-purchase journey. The platform’s segmentation and conditional logic went unused. Email-attributed revenue sat at about PKR 1.1M a month against a PKR 16M top line.

3. Browse abandoners were completely ignored. The larger and more recoverable audience — shoppers who viewed products, sometimes multiple times over several days, but never added to cart — triggered nothing. In fashion, browse abandonment dwarfs cart abandonment in volume, and these are warm, high-intent visitors. The brand was paying Meta to acquire them and then silently letting them leave.

4. High drop-off on pending cash-on-delivery orders. A Pakistan-specific leakage: customers who completed the COD checkout but then did not confirm when a call-centre agent rang, or whose orders were cancelled at confirmation. These pending COD orders were not connected to any automated recovery — they sat in a manual callback queue that reached fewer than half of them within the window when the buyer was still warm.

5. No mobile-native recovery path. With 78% of traffic on mobile, recovery had to meet buyers where they actually are. Email alone was insufficient — open rates on mobile inboxes are noisy, and Pakistani fashion buyers live in WhatsApp. The brand had no consent-captured WhatsApp or SMS recovery touch, despite it being the highest-engagement channel in the market.

6. Tracking gaps undermined everything. The Klaviyo Shopify integration was firing inconsistently — the “viewed product” event was not capturing on mobile, add-to-cart was firing twice in places (causing deduplication issues), and the browse event the platform depended on was effectively dead. Attribution between recovery sends and placed orders was incomplete, so even the one cart email that ran could not be properly measured. You cannot recover a cart you cannot see.

The combined effect: PKR 11.5M of carted value abandoned monthly, 7% email revenue share, and a recovery programme that did not exist in any meaningful sense. The fix was not more subscribers or more campaigns — it was building a recovery engine across the channels Pakistani fashion buyers actually use, on top of tracking the brand could trust.

Phase 1 — Diagnosis and Tracking Repair (Weeks 1-2)

Recovery is a tracking problem before it is a creative problem. The first two weeks were about making the data trustworthy, so that every flow we built later fired on the right event and every result could be measured.

Event audit and repair. We audited every ecommerce event end to end — viewed product, added to cart, started checkout, placed order. The browse event was rebuilt so it captured reliably on mobile and fired once per session per product rather than duplicating. Add-to-cart deduplication was fixed. The started-checkout event — distinct from added-to-cart, because it marks a buyer who has actually entered the checkout form — was wired correctly for the first time. This event separation matters: cart abandonment and checkout abandonment are different failure points and warrant different recovery treatment.

Klaviyo integration rebuilt. We re-anchored the Shopify–Klaviyo integration, verified the catalog feed was syncing so product-specific browse and recommendation flows would work, and confirmed that historical order data had backfilled correctly. We added server-side event capture for the highest-value events so recovery triggers survived ad-blockers and iOS Safari Intelligent Tracking Prevention — both of which were silently suppressing client-side events for a meaningful slice of mobile traffic.

COD order mapping. This was the most Pakistan-specific repair. We mapped the order lifecycle so that pending COD orders — placed but unconfirmed — were exposed as a recoverable state in Klaviyo rather than buried in the call-centre queue. A pending COD order that ages past a defined window without confirmation now enters a recovery flow automatically, in parallel with the human callback. The handoff between automated recovery and agent contact was defined so the two never conflicted.

Deliverability and list hygiene. We authenticated the sending domain (SPF, DKIM, DMARC), ran engagement-tier segmentation across the 34,000-subscriber list, and sunset the long-inactive tail with a re-engagement sequence. Inbox placement on monitored seed lists moved from 85% to 95% by the end of the fortnight. A clean, authenticated list is the prerequisite for recovery — none of the flows we were about to build matter if the messages land in spam.

Segmentation foundation. Even at this early stage we enriched each profile with the properties recovery logic would need: AOV tier, primary collection browsed (lawn, pret, unstitched, accessories), new versus returning buyer, and COD versus prepaid preference. This let Phase 2’s copy and incentive logic vary by who the buyer was, not treat everyone identically.

Phase 1 results (by end of week 2):

Diagnostic	Before	After repair
Reliable ecommerce events	3 of 4 firing	4 of 4 firing, mobile-fixed
Browse event (mobile)	Not capturing	Capturing once per session per product
Pending COD orders in Klaviyo	Invisible	Mapped as recoverable state
Inbox placement (seed test)	85%	95%
Segmentable profile properties	2	6 (tier, collection, recency, etc.)
Server-side event capture	None	Active on 3 high-value events

Phase 2 — Build the Recovery Engine (Weeks 3-5)

With trustworthy tracking and a clean list, we built the recovery architecture across the three abandonment points that matter for fashion ecommerce — cart, browse, and checkout — plus the COD confirmation flow that is unique to this market. We cover the full scope of this work in our abandoned cart and browse recovery service; the build below is how it was applied here.

Three-email cart abandonment series. We replaced the single generic cart email with a timed, behaviour-aware sequence. Email one fired within 30 minutes — a simple, branded reminder with the exact items left behind and a one-tap return to checkout. Email two followed at roughly 12 hours, focused on reassurance rather than incentive: fabric and sizing guidance, real customer reviews for the items in the cart, COD and returns policy spelled out, and the questions Pakistani fashion buyers ask before paying PKR 6,500 online (is the colour accurate, will it arrive, can I return it). Email three, at roughly 48 hours, introduced a measured incentive only for carts above a value threshold and only for buyers who had not returned — the incentive ladder, not a blanket discount.

Browse abandonment flow. This addressed the largest ignored audience. Shoppers who viewed the same product at least twice across separate sessions, or who viewed three or more products in a collection without carting, entered a browse flow. The email showcased the viewed products with social proof, surfaced matching pieces from the same collection, and addressed the most common reasons a browse does not convert into a cart (price hesitation, sizing uncertainty, waiting for a drop). This flow alone became a meaningful revenue contributor because its addressable audience was larger than the cart audience.

Checkout abandonment (distinct from cart). Because the started-checkout event was now reliable, we built a separate, higher-intent flow for buyers who had entered the checkout form. These shoppers are closer to purchase, so the copy was lighter on persuasion and heavier on friction removal — a direct return-to-checkout link, a prepaid-discount nudge where the buyer had chosen COD, and a consented SMS nudge for high-value checkouts. Treating checkout abandonment separately from cart abandonment, rather than lumping them into one flow, is a meaningful recovery lift.

COD confirmation automation. We automated recovery for pending COD orders. The moment an order sat in pending-COD status past a defined window, the buyer received a confirmation touch — first via the channel they prefer (email or SMS, WhatsApp added in Phase 3), with a clear “confirm your order” path and the agent’s contact line. This sat alongside, not instead of, the human callback queue, and it reached buyers in the window when they were still warm rather than a day later.

Recovery copy framework. Every send was built around the four questions Pakistani fashion buyers actually need answered before completing a purchase: Is the product as described? Will it arrive, and how fast? Can I pay cash on delivery and inspect it? Can I return it if the fit is wrong? Trust-building content answering these four questions outperformed discount-led copy in every test we ran — a finding consistent across the fashion verticals we cover in our work for digital marketing for clothing brands.

Phase 2 results (by end of week 5):

Flow	Status	Early signal (first 2 weeks live)
Cart abandonment (3 emails)	Live	Recovery 4.1% to 8.6%
Browse abandonment	Live	First browse-driven orders attributed
Checkout abandonment	Live	Highest per-send revenue of any recovery flow
COD confirmation automation	Live	Pending COD confirmation rate climbing

Phase 3 — Optimize and Add WhatsApp Recovery (Weeks 4-8)

Coverage was in place; this phase was about performance and adding the channel that matters most in this market. As in the build phase, we began optimising each flow the moment it had enough volume to read.

WhatsApp Business API recovery. This was the single largest lever in the engagement. Pakistani fashion buyers live in WhatsApp — it is where they message the brand with sizing questions, where they confirm COD orders informally, and where they expect rapid response. We layered consent-captured WhatsApp recovery onto the cart and COD flows for buyers who had opted in. A concise WhatsApp nudge at the 90-minute mark, paired with the email sequence, lifted engagement materially versus email alone. Open and read rates on WhatsApp recovery were multiples of the equivalent email metrics, and the channel drove a disproportionate share of recovered COD orders. The WhatsApp layer is part of the broader lifecycle work we describe in our email marketing automation service, applied here specifically to recovery.

Structured A/B testing. We tested the elements that move recovery revenue: send timing on the cart series (30 minutes versus 60 versus 90), discount depth and type in the incentive email (percentage off versus flat rupees off versus free shipping), subject line framing, and WhatsApp-versus-email-versus-both sequencing. Each test ran to a pre-set sample size before a winner was locked. Notably, the no-incentive reassurance email consistently outperformed an early discount — buyers who returned on reassurance had a lower return rate and higher AOV than those who returned on discount. This is the finding that shaped the incentive ladder.

Incentive ladder. Rather than leading with a discount (which trains buyers to abandon in order to receive it), we built a ladder: reassurance and social proof first, friction removal second, and a measured incentive last and only for carts that did not respond. The incentive was also tiered by AOV and by whether the buyer had purchased before — a returning buyer received a softer incentive than a first-timer, because they already trusted the brand. This protected margin while still recovering the hard cases.

Mobile-native email design. Because 78% of opens were on mobile, every recovery email was rebuilt single-column, thumb-scrollable, with a single primary call to action above the fold and product imagery that loaded fast on mobile data connections. Load speed on recovery emails is a quiet but real factor — an email whose images take four seconds to render on a mid-range Android over a patchy connection loses the reader before they see the product.

COD confirmation optimisation. We tested the timing and tone of COD confirmation sends across email, SMS, and WhatsApp, and coordinated with the call-centre team so that automated touches and agent calls did not collide (two contacts within an hour annoyed buyers and reduced confirmation). The coordinated cadence lifted confirmation rate while actually reducing agent workload per confirmed order.

Phase 3 results (by end of week 8):

Metric	Start of phase	End of week 8
Cart abandonment recovery	8.6%	10.9%
Pending COD order confirmation	9%	15%
Email revenue share	13%	19%
Email-attributed revenue (monthly)	PKR 2.1M	PKR 3.0M
WhatsApp recovery read rate	Not live	78%

Phase 4 — Measure and Compound (Weeks 8-12)

The final phase turned the recovery engine into something durable and measurable at the revenue level the brand actually cared about. Recovery rate had already moved substantially; the task now was to prove the revenue impact, lock in the iteration rhythm, and protect margin against discount dependency.

Revenue attribution dashboard. We built a dashboard that tracked recovered revenue by flow (cart, browse, checkout, COD), recovery share of total revenue, COD confirmation rate, per-send revenue, and the discount cost of recovered orders. This replaced the brand’s previous gut-feel view of email — “it sends, we think it works” — with a defensible accounting of which flows drove which revenue and at what discount cost. The 21% email revenue share and PKR 3.4M monthly figure are what this dashboard made visible and trackable over time.

Discount-cost tracking. A danger with recovery programmes is that they appear to drive revenue while quietly eroding margin through discounts. We tracked the discount cost of recovered orders separately, so the brand could see net recovered revenue, not just gross. This is what let them confidently increase acquisition spend knowing recovery was profitable, not just top-line flattering.

Iteration cadence. We established a monthly recovery review: recovery rate by flow, revenue share trend, discount-cost trend, COD confirmation movement, and one new test. This cadence is what kept recovery rate climbing past the build phase rather than settling at its Phase 3 level. The 11.4% monthly recovery figure matured in this phase as WhatsApp recovery reached full coverage and the incentive ladder stabilised.

Discount dependency reduction. Over weeks 8-12 we deliberately reduced the share of recovered orders that relied on a discount, leaning harder on the reassurance and friction-removal emails that testing had shown convert without margin cost. This is the part of a recovery programme that separates a healthy engine from a discount treadmill.

By the 90-day mark, the engine had done what acquisition spending alone never could: 11.4% of abandoned carts were being recovered every month, email revenue share had climbed from 7% to 21%, and email-attributed revenue had grown from PKR 1.1M to PKR 3.4M — all without an increase in ad spend.

Final Results at 90 Days

Metric	Before	At 90 days	Change
Cart abandonment recovery (monthly)	4.1%	11.4%	+7.3 pts
Email revenue share	7%	21%	+14 pts
Email-attributed revenue (monthly)	PKR 1.1M	PKR 3.4M	+209%
Browse-abandonment conversion lift	Not running	+24%	New channel
Pending COD order confirmation	9%	17%	+8 pts
Recovery flow open rate	19%	38%	+19 pts
Discount cost share of recovered rev.	n/a	11%	Tracked

Each result traces to a specific phase: tracking repair and COD mapping in Phase 1 enabled the rebuilt cart, browse, and checkout flows in Phase 2; WhatsApp integration and the incentive ladder in Phase 3 drove the recovery-rate acceleration; and the attribution dashboard and iteration cadence in Phase 4 made the gains durable, measurable, and margin-aware.

What Made This Work

1. Tracking before creative. The decisive move was repairing the event layer and COD mapping before building any flow. Recovery fails silently when events fire inconsistently or when a whole order state (pending COD) is invisible to the platform. Trustworthy tracking is the prerequisite that made everything else measurable.

2. The channels Pakistani buyers actually use. Adding consent-captured WhatsApp recovery was the largest single engagement lever. Email alone under-serves a market where the default communication channel is WhatsApp and where COD confirmation happens informally. Meeting buyers on the channel they already live in beat every creative change we tested.

3. Cart, browse, and checkout treated separately. These are three distinct abandonment points with distinct intent levels, and they warrant distinct treatment. The browse flow alone contributed meaningful revenue simply because its addressable audience was the largest and it had previously been ignored entirely.

4. The incentive ladder, not the discount. Leading with reassurance and social proof, and reserving incentives for carts that did not respond, recovered revenue while protecting margin. The finding that reassurance-led returns had lower return rates and higher AOV than discount-led returns reshaped the whole programme.

5. COD confirmation as a first-class recovery flow. In a market where most orders are cash on delivery, the pending-COD queue is one of the largest revenue leaks, and most brands treat it as a call-centre problem rather than a recovery problem. Automating it in coordination with human callbacks recovered orders that were already “won” but slipping away.

What Teams Can Apply

For Pakistani fashion and apparel ecommerce brands that want to recover revenue at checkout:

1. Audit your event tracking before you build flows. If “viewed product” and “started checkout” do not fire reliably on mobile, your recovery flows are operating on partial data. This is the single most common recovery blind spot.

2. Build browse abandonment, not just cart abandonment. The browse audience is larger and warmer than most teams realise. Ignoring product viewers who never carted means ignoring the biggest recoverable segment you paid to acquire.

3. Add WhatsApp to recovery if you have consent. In Pakistan, WhatsApp recovery reads and converts at multiples of email. It is the highest-leverage channel addition for any fashion brand running recovery on email alone.

4. Lead with reassurance, not discounts. Test trust-building copy — fabric, sizing, COD and returns — before you reach for a discount code. Recovery programmes that depend on discounts train buyers to abandon; programmes that reassure do not.

5. Treat pending COD orders as a recovery flow. Map the COD lifecycle so unconfirmed orders enter automated recovery in coordination with your call centre. This is a Pakistan-specific leak that most dashboards do not even surface.

WeProms Digital has applied this recovery framework across Pakistani ecommerce brands in women’s apparel, menswear, footwear, jewellery, and accessories. The specific channels, copy, and incentive ladder change with each category — but the tracking-first, multi-channel, reassurance-led approach stays the same.