Advantage+ Shopping Case Study in Pakistan

The Client

A Karachi-based fashion-jewellery brand — gold-plated everyday sets, kundan and polki-look bridal pieces, and an oxidized line aimed at university buyers — selling nationwide through its own storefront alongside a thin marketplace sideline. The brand was founded in 2019, ran a team of fourteen, and carried roughly PKR 6–8M in monthly online revenue outside festive windows, with average order value near PKR 4,200 and about 70% of orders completing as cash-on-delivery.

Like most Pakistani jewellery and fashion ecommerce, the brand’s year pivoted on corridors: the Eid corridor and the October-to-March wedding season together carried close to half of annual revenue. The six weeks around Eid — ramp, last ten days, Chaand Raat spike, and the post-Eid wedding events — were the make-or-break window, and the brand had historically treated them as a period to spend harder in rather than a period to prepare an account for.

That was the gap. The brand approached WeProms Digital for Meta ads management eleven weeks before Eid with a blunt brief: the previous Eid corridor had returned a blended 2.3x ROAS on roughly PKR 2.5M of spend — below break-even once delivery and returns were counted — and the team did not want to walk into the same corridor with the same account and hope for different arithmetic.

The Problem

The diagnostic in week one found an account that could not have scaled efficiently even with more budget:

• Structural overlap. Nine campaigns and twenty-seven ad sets, most built reactively over two years, targeting overlapping broad and interest audiences. Fourteen ad sets were effectively bidding against each other for the same Karachi, Lahore, and Hyderabad buyers, driving internal auction pressure and unstable delivery.
• A catalog feed that blocked automation. Thirty-one percent of SKUs were disapproved or unsynced. Titles were file-name artifacts — “IMG_2043” — with no style, occasion, or plating attributes; variant sets lacked item group IDs, so a choker available in three colours appeared as three disconnected products. Any automated campaign format was optimizing against a catalog it could barely serve.
• Undercounted purchases. A browser-only Pixel on a mobile-first, COD-heavy customer base was losing an estimated fifth of purchase events to adblockers, in-app browsers, and pre-payment drop-off recovered over WhatsApp. The account had been making scaling decisions against a signal that systematically understated results.
• Creative exhaustion. Six assets in rotation for five months. Retargeting frequency sat above 4.0 while prospecting went hungry, and the one comment theme repeated across ad posts — “what’s the price?” — was unanswered in the creative itself.
• No festive destination. Every click landed on the homepage, where festive shoppers browsed 640 SKUs undifferentiated by occasion and bounced.

The summary diagnosis: this was not a bidding problem or a budget problem. It was a readiness problem — the account’s data, feed, creative, and landing experience were all unfit for a corridor where Pakistani CPMs climb steeply and every efficiency point is won before the corridor opens, not during it.

Phase 1 — Diagnosis and cleanup (Weeks 1-2)

Phase 1 fixed the two things that would otherwise poison every later decision: what the account believed was happening, and what the delivery system could actually serve.

Account consolidation. We mapped all nine campaigns by audience overlap and historical CPA, then switched off fourteen ad sets that duplicated reach. Three manual campaigns survived on purpose — a small evergreenseller campaign, a cart retargeting campaign, and a catalog-free brand campaign — pending the Phase 2 build. Spend did not drop; it concentrated.

The feed rebuild. This was the real work of week two, and the pattern is the core of any product feed optimization engagement: every SKU’s title was rewritten to a pattern a buyer actually searches — style, occasion, plating, and set contents (“Kundan chandbali earrings, gold-plated, bridal set with jhumkas”) — rather than a file name. Variant sets received proper item group IDs so colourways rolled up into one product with options instead of three orphans. Dead and duplicate SKUs were pruned from 640 to 410 active. Descriptions gained plating and care facts that reduced both pre-purchase WhatsApp questions and return friction.

Measurement repair. A Conversions API deployment via a server-side gateway began sending purchase events with event-ID deduplication against the browser Pixel, recovering the estimated 18–20% of purchases the browser-only setup was losing and giving the account one purchase number to scale against. Match quality on deduplicated events rose from 6.2 to 8.4 within the fortnight.

Baseline freeze. We froze the dashboard everyone would be judged against: store-reconciled revenue, platform-reported ROAS, cost per purchase, catalog health, and AOV — pulled weekly, not daily, because corridor-week daily variance invites overreaction.

Account hygiene metric	Week 0	End of week 2
Active ad sets	27	13 (further cut in Phase 2)
Catalog SKUs disapproved/unsynced	31%	Under 3%
Variant sets with item group IDs	22%	100%
Deduplicated event match quality	6.2 / 10	8.4 / 10
Purchase events vs store orders	~81%	~99%

Phase 2 — Build the Advantage+ engine (Weeks 3-5)

With clean data and a servable catalog, Phase 2 rebuilt the account around a single scaling engine.

One Advantage+ Shopping campaign became the growth core, carrying roughly 70% of corridor spend at ramp. The decision was deliberate: the brand’s audience definitions had been the weakest part of the old account, and Advantage+ shopping delivery — mixing broad prospecting, retargeting, and existing-customer reach inside one campaign — removes exactly the audience micro-management the team had been worst at. The full 410-SKU catalog fed it, with a product set privileging the top sixty revenue SKUs, and the existing-customer budget share was capped at 15% so the corridor stayed acquisition-weighted. The old cart retargeting campaign was folded in; one small manual campaign survived for a pre-Eid “new arrivals” engagement push, and that was the entire account.

	Before	After
Campaigns	9	2 (+1 seasonal)
Ad sets	27	1 core engine + 1 supporting
Weekly creative additions	0 (6 assets total)	4–6 new assets
Audience maintenance workload	High, manual	Minimal, delivery-managed
Budget decision cadence	Ad hoc, screenshot-driven	Written 48-hour ladder rule

The creative bank. Eighteen assets went into the launch, built occasion-first because that is how festive jewellery demand arrives: three-second hook-led Reels showing sets against outfit drapes for Mehndi, Baraat, and Walima styling; price-point statics answering the “what’s the price?” question the comment sections had been asking for months; and finish close-ups for the gold-plated line’s main objection — “will it tarnish?” — answered with the six-month re-plating policy in the copy. Six of the eighteen were micro-creator UGC pieces shot in Karachi and Lahore at a fraction of studio cost, chosen because they carried native trust into cold feeds.

Occasion landing pages. Three pages replaced the homepage as destinations — Eid gifting, Mehndi and Baraat, and Walima “reception-weight” sets — each filtering the catalog to a shoppable shortlist with WhatsApp and COD trust markers visible above the fold. This single change did quiet work later: festive visitors stopped landing in a 640-SKU warehouse and started landing in a curated corridor.

Phase 3 — Scale through the Eid corridor (Weeks 4-8)

Phase 3 was disciplined scaling. The rule was written before the corridor opened: daily budget steps up 15–20% every 48 hours whenever trailing three-day store-reconciled ROAS held at or above 3.0x, and holds — never cuts — when it dipped. Budget never moved more than once in 48 hours, protecting delivery learning through the corridor’s noisiest stretch.

The ladder ran from PKR 40k a day at the corridor’s opening to PKR 105k a day in the Eid week, easing to PKR 70k through the post-Eid wedding events. CPMs told the real story of corridor pressure — from roughly PKR 190–220 in the ramp weeks to PKR 560–680 at peak, in line with what Pakistani advertisers see as Eid competition compresses delivery — which is precisely why the efficiency gains had to come from CTR, conversion, and AOV rather than from cheaper delivery.

Creative velocity and kill rules. Four to six new assets entered weekly, themed to where the corridor was: gifting angle in weeks one and two, last-minute Chaand Raat urgency in the final week — next-day Karachi and Lahore delivery carried the message — then styling content for post-Eid wedding events. Kill rules were fixed in advance: an asset with over 8,000 impressions and CTR below 1.2% was retired; an asset with CPA above 1.3x target after three days was rotated out. No asset survived on opinion. Average CTR across the corridor rose from 1.1% to 2.3%, and retargeting frequency — previously above 4.0 — held between 2.1 and 2.9 because fresh creative kept the same audiences clicking instead of hiding.

Corridor week	Daily budget	Blended ROAS	Cost per purchase
Ramp week 1	PKR 40k	2.9x	PKR 1,520
Ramp week 2	PKR 55k	3.1x	PKR 1,470
Pre-Eid week	PKR 80k	3.6x	PKR 1,410
Eid week (peak)	PKR 105k	4.5x	PKR 1,180
Post-Eid weddings	PKR 70k	3.4x	PKR 1,340

Average order value did quiet work too. Complete-the-look bundles — earrings-plus-jhumka pairings pre-styled per occasion — lifted AOV from PKR 4,200 to roughly PKR 4,750 (+13%) without discounting, which is the cheapest ROAS lever available to a jewellery brand in a gifting corridor. The occasion pages converted at 2.6% against the homepage’s 1.7% baseline, absorbing CPM inflation that would otherwise have eroded the ladder.

Phase 4 — Wind down, reconcile, and compound (Weeks 8-12)

Corridors end, and unmanaged endings waste the peak’s gains. The step-down followed a written plan mirroring the ladder: budget reduced roughly 30% every three days to PKR 55k a day as evergreen demand took over, with the creative bank rotating back to everyday sets and the office-wear oxidized line.

Reconciliation. Across the six-week corridor the account spent approximately PKR 2.9M against prior Eid’s PKR 2.5M (+16%) and drove close to PKR 10.0M of store revenue versus PKR 5.7M the year before — the 3.4x blended corridor ROAS against 2.3x, with roughly 2,100 orders at a PKR 1,380 average cost per purchase. Platform-reported figures were reconciled against store data weekly, which mattered twice: once when in-app browser attribution lagged during the Eid week and the dashboard was trusted instead of panicked over, and once when a returns spike in post-Eid week two was caught early enough to swap the sizing-ambiguous bridal sets out of top billing.

The compounding assets. Three things outlived the corridor. The feed structure now maintains itself against a written title pattern, so the next catalog drop inherits cleanliness instead of debt. The creative bank’s Eid-period winners were archived with performance notes — a styling-Reels library the next corridor starts from rather than a blank page. And the ladder rule, kill rules, and step-down plan were documented as a one-page corridor playbook, so Eid ul Adha and the winter wedding season begin from a system rather than a scramble.

Final Results at 90 Days

Metric	Before (prior Eid)	After (this corridor)	Change
Festive-week ROAS (peak)	~2.8x best week	4.5x	+61%
Blended corridor ROAS (6 weeks)	2.3x	3.4x	+48%
Corridor revenue	~PKR 5.7M	~PKR 10.0M	+75%
Corridor spend	~PKR 2.5M	~PKR 2.9M	+16%
Cost per purchase	PKR 1,840	PKR 1,380	−25%
Average order value	PKR 4,200	PKR 4,750	+13%
Ad CTR (corridor average)	1.1%	2.3%	+109%
Landing-page conversion rate	1.7% (homepage)	2.6% (occasion pages)	+53%
Catalog disapprovals	31% of SKUs	Under 3%	Fixed

The arithmetic behind the headline is ordinary and checkable: sixteen percent more spend, converted at a 48% better corridor ROAS because purchases cost a quarter less and orders carried thirteen percent more value. These figures are illustrative outcomes built from common patterns WeProms sees across Pakistani jewellery and fashion ecommerce — not an audited third-party statement — and they exist so a growth team can sanity-check what a prepared account should plausibly return from an Eid corridor at this AOV and margin.

What Made This Work

1. The feed was fixed before the budget rose. Advantage+ shopping delivery can only optimize against products it can serve. Repairing titles, variant grouping, and disapprovals in week two meant the engine spent the corridor learning buyer preferences instead of working around a broken catalog.
2. One engine replaced twenty-seven competing ad sets. Consolidation ended internal auction pressure and gave delivery stable signals. The team stopped micro-managing audiences — historically their weakest skill — and spent that attention on creative and landing experience, where their instincts were strong.
3. Scaling was governed by a written rule, not corridor adrenaline. The 48-hour ladder with a hold-don’t-cut clause kept the account compounding through noisy days. The single most common self-inflicted wound in Pakistani festive accounts is a panic budget cut during a two-day ROAS dip that resets learning exactly when CPMs peak.
4. Creative was planned by occasion, not by audience segment. Mehndi, Baraat, Walima, and last-minute gifting each received hooks that answered the actual question in the comments — price, tarnish, delivery time. CTR doubling was a creative outcome, and in Meta’s auction, creative rather than targeting does most of the heavy lifting.
5. AOV worked alongside efficiency. Pre-styled bundles raised order value thirteen percent without a discount, which compounded with cheaper purchases into the corridor ROAS. Most festive accounts chase volume discounts that do the opposite.

What Teams Can Apply

For Pakistani ecommerce brands approaching a festive corridor:

1. Audit for overlap before you scale. List every campaign and ad set and ask which two are chasing the same buyer. Consolidation costs a week and pays for the whole corridor in auction efficiency.
2. Treat your feed as campaign infrastructure. A catalog with file-name titles and broken variant grouping caps every automated campaign format’s ceiling. Rewrite titles to how buyers search — style, occasion, attribute — two months before the corridor, not during it.
3. Write the ladder down before the corridor opens. A rule of 15–20% steps every 48 hours while three-day reconciled ROAS holds above your threshold, with holds instead of cuts on dips, outperforms judgment calls made during CPM spikes.
4. Build occasion destinations, not homepage traffic. A curated Eid or wedding-event page converting at 2.6% against a homepage’s 1.7% is a 53% efficiency gain that no bidding change can match.
5. Plan the wind-down as carefully as the ramp. Corridor demand decays over days, not weeks. A written step-down keeps the account from carrying Eid-week budgets into ordinary Tuesdays.

WeProms Digital has applied this corridor-preparedness framework across Pakistani ecommerce brands in jewellery, fashion, beauty, and home goods. The occasion mix, creative bank, and ladder thresholds shift with each catalog and margin — the sequence of feed, measurement, consolidation, creative, and governed scaling is what stays constant.