Shopify Subscription ROAS Case Study in Pakistan

The Client

The brand is a Multan-based direct-to-consumer company selling premium dates and dry fruits: Aseel dates sourced from Khairpur, Dhakki dates from Khyber Pakhtunkhwa, almonds, cashews, walnuts, dried apricots, and its own trail-mix blends. The catalogue splits into two very different jobs. Gift boxes — corporate Eid gifting, wedding favours, and curated occasion hampers — drive sharp seasonal spikes and high order values. Pantry staples — the everyday replenishment products — drive smaller, more frequent orders. Revenue ran between PKR 7M and 8M a month off-peak and pushed toward PKR 14M to 16M through Ramzan, Eid, and the winter wedding season.

The channel mix was heavily paid: roughly PKR 1.9M a month on Meta and another PKR 0.5M on Google Shopping and brand search, with a genuinely strong organic social presence built on recipe content and sourcing reels from the date farms. Around 78% of orders were cash on delivery, fulfilled through third-party courier partners.

Seven months before the engagement, the team had launched a monthly pantry box: a curated 1.5kg selection at a 12% subscriber discount, positioned as the convenient way for a household to never run out. It was the right product for the category — dates and nuts are consumed on a predictable rhythm — but it had quietly stalled. Attach sat at 3.9% of orders, the dedicated subscription campaigns had been paused months earlier for “inefficiency,” and the media buyer’s dashboard framed every rupee of spend against one-time purchase ROAS.

The timing mattered as much as the ask. The engagement ran through the summer trough — the post-Eid months when gifting demand softens and paid efficiency is hardest to defend — with the winter wedding season visible on the horizon as the next scaling window. In other words, the framework had to work in the brand’s worst quarter, not its best, before anyone would trust it with the seasonal ramp.

The founder’s brief to WeProms was blunt: paid is getting more expensive every season, the subscription that should be cushioning that isn’t working, and the wedding season ramp is eight weeks out. They wanted paid efficiency restructured around what a customer is actually worth, not what a single order returns.

The Problem

The diagnosis surfaced six linked issues:

• Attach was a visibility problem before it was a demand problem. The subscription option existed only on two product pages, rendered as a small radio button below the fold. It never appeared in the cart drawer, at checkout, or after purchase. Most buyers never saw it.
• Blended CAC was climbing with CPMs. Across Ramzan and the Eid-ul-Adha wave, Meta CPMs in Pakistan rose on the order of 30% year on year, and the brand’s blended CAC moved from PKR 2,380 to PKR 3,150 while average order value stayed flat.
• ROAS was measured against the wrong denominator. Campaigns optimised to 7-day-click purchase value treated a PKR 5,700 one-time gift order and the first box of a PKR 21,600 six-month subscriber as identical events. The account was structurally blind to recurring revenue.
• The tracking signal was incomplete. A browser-only pixel was losing roughly 22% of purchase events to iOS signal loss, ad blockers, and the Pakistani staple of browsing on one device and confirming a COD order on another.
• COD returns were silently taxing acquisition. Return-to-origin ran at 21% of cold paid orders, and each return consumed two-way courier cost plus packaging. Dashboard CAC understated true CAC.
• The subscription campaign had been killed by the same broken lens. It had been paused for sub-2x ROAS — evaluated on first-box revenue while paying subscriber-acquisition costs.

Underneath all six sat one sentence nobody in the business could say with data: what is a subscriber worth, and what are we willing to pay for one?

Phase 1 — Measurement Rebuild and Cohort Diagnosis (Weeks 1-2)

Nothing structural changed in the first fortnight. The account had been making budget decisions against partial data, and every subsequent move depended on fixing that first.

Server-side tracking with deduplication. We implemented the Conversions API through a properly configured server-side purchase event, sharing event IDs with the browser pixel so duplicate purchases collapsed into single counted events, following the same discipline as our Meta Pixel and Conversions API setup practice. Against identical spend, the account could finally see about a fifth more purchase events than the browser-only configuration had reported — not a performance improvement, a visibility improvement, and the precondition for everything after it.

Custom events for recurring revenue. Alongside the standard purchase event we instrumented SubscriptionStarted and renewal orders as distinct, valued events, so subscription revenue stopped being laundered through generic purchase ROAS.

Cohort value separation. We split twelve months of order history into one-time buyers and subscriber cohorts. The numbers reshaped the whole engagement:

Diagnostic	Finding	What it changed
One-time gift buyer	PKR 5,700 AOV, ~PKR 6,400 six-month value	Gifting campaigns judged on first-order ROAS were fine
Pantry-box subscriber	~PKR 21,600 six-month value, 3.8x a gift buyer	Subscription acquisition could afford a higher CAC than gifting
Reported Meta purchases	Undercounted ~22%	Prior budget decisions were made on bad inputs
COD return rate, cold traffic	21% of dispatched orders	True CAC was materially above dashboard CAC
Subscription campaign pause	Evaluated on first-box ROAS	The pause was a measurement artefact, not a verdict

Blended CAC, honestly computed. We rebuilt CAC as blended across paid, organic social, direct, and email — then adjusted for return costs so a PKR 3,150 blended CAC reflected what acquisition actually cost. The return math was blunt: at 21% return-to-origin, every PKR 100 of nominal acquisition carried roughly PKR 8 to 9 of two-way courier, packaging, and repicking cost that no ad dashboard showed. This became the single scoreboard metric for the engagement, replacing platform ROAS.

By the end of week two, no creative had been touched and no campaign restructured. But the business could finally price a subscriber against a gift buyer, and that ratio — 3.8 to one — was about to drive every bidding decision.

Phase 2 — Attach Path and Store Restructure (Weeks 3-5)

With measurement trustworthy, we rebuilt the purchase path so the subscription stopped depending on buyers finding it themselves.

The cart drawer became the primary attach surface. Any cart containing a pantry SKU now showed a one-tap toggle: switch the items to monthly delivery, save 12%, and get free delivery on every box. This placement matters because it reaches buyers at maximum intent, after they have already chosen the products, without asking them to re-enter anything.

Anchor pricing on product pages. On replenishment SKUs the monthly price became the default selection, with the one-time price shown struck through beside it. Gift SKUs kept one-time as the default — forcing subscription framing on an Eid corporate gift would have damaged conversion — with a softer “send this box monthly” suggestion instead.

Post-purchase one-click attach. The order status page offered a single-click conversion of the just-placed order into a monthly box. Card and wallet buyers attached instantly; COD buyers confirmed a recurring cadence with one tap and a verification call before each dispatch.

A gifting-to-pantry bridge. Gift-box buyers — the brand’s highest-volume seasonal segment — were offered a follow-on pantry subscription framed as keeping the household stocked after the occasion. This deliberately used the gifting spike as a subscription acquisition engine rather than a dead end.

Each surface carried its own measurable conversion: the cart-drawer toggle converted on 11.6% of carts containing a pantry SKU, the post-purchase one-click offer attached 6.2% of eligible one-time orders, and the gifting bridge ran weakest — corporate gift buyers converted to pantry subscriptions at under 2%, mostly when the sender rather than the recipient took up the offer. We kept it because those converts were the cheapest subscribers in the program, acquired at near-zero incremental media cost, but we stopped investing in its creative once the pattern was clear.

Confirmed-cadence COD. Because most Pakistani buyers pay on delivery, the subscription was built as a commitment rhythm rather than an auto-charge: a confirmation call two days before each monthly dispatch, with an easy skip option, and a prepaid quarterly plan at a deeper discount for the card-comfortable segment. Renewal payment never became the reason to churn.

By the end of week five, attach had moved from 3.9% to 5.8% of first-time orders — before the paid restructure had even begun, and driven almost entirely by traffic the brand was already paying for.

Phase 3 — Paid Restructure Around Recurring Value (Weeks 4-8)

Phase 3 ran in parallel, restructuring the account the way a shopify marketing agency engagement should: consolidate for signal, then let cohort economics — not legacy structure — set the bids.

Consolidation for learning. Twenty-four mostly starved ad sets collapsed into six campaigns with three jobs: gifting prospecting on hero SKUs, pantry-subscription acquisition aimed at subscriber lookalikes and warm audiences, and tight retargeting with attach-forward creative. Concentrating budget gave each campaign enough conversion volume to exit the learning phase properly.

Value-based bidding keyed to cohort value. The gifting campaigns stayed on purchase optimisation with server-side purchase values. The subscription campaign optimised to SubscriptionStarted once it cleared roughly sixty events per thirty days, with bid guidance derived from the PKR 21,600 six-month subscriber value rather than the first box’s revenue. This is the specific mechanism that made paying above the old CAC rational: a PKR 3,400 cost to start a subscriber is expensive against a PKR 2,800 first box and cheap against six months of margin.

A creative grid with local texture. Four angles — sourcing provenance from the Khairpur farms, everyday snacking and energy, gifting occasions, and subscription convenience — crossed with three formats: UGC unboxing reels, static stat cards, and carousels. Sourcing-reel creative filmed on the farms consistently outperformed studio product shots, a pattern we see across Pakistani food D2C. Attach-forward retargeting creative explicitly showed the cart-toggle mechanic, teaching the offer mechanically rather than conceptually.

RTO control as a paid-efficiency lever. Cold-prospect COD was restricted to metro and tier-one city delivery zones, checkout required a verified phone and address, prepaid orders earned an extra 5% off, and courier confirmation calls ran before dispatch. Returns on cold paid traffic fell from 21% toward the mid-teens within the phase, which flows straight into true blended CAC.

Metric	Before	End of Phase 3 (Week 8)
Subscription attach rate	3.9%	7.1% of first-time orders
Blended CAC	PKR 3,150	PKR 2,480
Meta ROAS (30-day click)	2.3x	2.9x
COD return rate, cold traffic	21%	15%
Subscription share of revenue	7%	14%

Phase 4 — Season Pacing and Compounding (Weeks 8-12)

The final phase built the operating rhythm that keeps these gains from decaying after the engagement window.

Season pacing. The August corporate-gifting bump around Independence Day was paced with a budget ladder that protected the subscription campaign’s learning, and a wedding-season ramp plan was documented with pre-agreed scaling triggers: raise daily budgets by a maximum of 20% per step once cost-per-subscription-start held stable for four consecutive days. Scaling by laddering, rather than doubling spend into the season, kept the efficiency gains from being auctioned straight back.

A weekly scoreboard. One dashboard, reviewed weekly: attach rate by SKU, blended CAC including return costs, ROAS by campaign, subscriber pause and churn rates, and return rate by delivery zone. Creative refresh ran on a standing cadence, with retirement rules set in advance — any ad below the ROAS floor after PKR 150,000 of spend was paused without debate, which kept creative decisions fast and unemotional.

Churn containment. The subscriber portal gained pause, swap, and cadence-change controls surfaced in every renewal confirmation. A paused subscriber is cheap to reactivate; a churned one has to be re-bought at full acquisition cost. Most pauses resumed within two cycles, and voluntary churn stayed low enough that the subscription base compounded rather than refilled.

Compounding effect. By week twelve, subscription cohorts carried 18% of monthly revenue. That is the structural change behind the CAC number: when roughly one rupee in five arrives from a base that does not need to be re-acquired, blended CAC falls even at flat spend — which is exactly how it reached PKR 2,240 without a meaningful budget increase.

Final Results at 90 Days

Metric	Before	Day 90	Change
Subscription attach rate	3.9% of first-time orders	7.9%	+103% (doubled)
Blended CAC	PKR 3,150	PKR 2,240	-29%
Meta ROAS (30-day click)	2.3x	3.2x	+39%
Subscription share of store revenue	7%	18%	+11 pts
COD return rate, cold paid traffic	21%	13%	-8 pts
Subscriber vs one-time 6-month value	Not measured	3.8x	Now a bidding input

These are illustrative outcome ranges reflecting patterns WeProms sees across Pakistani D2C food and consumables brands, not an audited result for a named company. They give a growth team a realistic shape for what a measurement-first, attach-led paid restructure can produce.

What Made This Work

1. The subscriber was repriced before anything was rebuilt. Until the cohort split put a PKR 21,600 six-month value against a PKR 6,400 one-time value, every bidding decision was made against the wrong number. The 3.8x ratio justified paying more for subscription starts than the old CAC ceiling allowed — the single highest-leverage insight of the engagement.
2. Attach was solved on traffic already being bought. Moving the offer into the cart drawer and post-purchase flow doubled most of the attach gain before paid restructure began. The store was already paying for these buyers; it simply was not showing them the offer.
3. Signal quality preceded spend decisions. Recovering roughly a fifth of lost purchase events through server-side deduplication meant value-based bidding learned from complete data. Bidding changes made before that fix would have been optimisation against noise.
4. RTO was treated as a paid metric. Restricting cold COD to deliverable zones, verifying contact details, and nudging prepayment cut returns by eight points — a direct, durable reduction in true CAC that most dashboards never surface.
5. Seasonality was planned, not survived. The budget ladder and pre-agreed scaling triggers let the brand enter the wedding season with efficiency intact instead of auctioning away the gains in a spending scramble.

What Teams Can Apply

For Pakistani D2C brands selling consumables on Shopify:

1. Compute one-time versus subscriber value before touching bids. The ratio, not the platform dashboard, sets what you can pay to acquire a subscriber. Without it you will keep pausing your most valuable campaigns for looking inefficient.
2. Put the attach offer where the intent already is. The cart drawer and the post-purchase page convert because the buyer has already decided. A subscription that lives only on a product page is a bet that buyers will explore.
3. Judge the account on blended CAC, not platform ROAS. Platform ROAS flatters last-click paid and ignores organic, email, and returns. One honest blended number beats five confident dashboards.
4. Fix the conversion signal before scaling spend. Server-side tracking with deduplication is not an advanced tactic in this market — it is the entry fee for value-based bidding to work at all.
5. Manage COD returns like a bid multiplier. Every return carries courier and packaging cost straight into CAC. Zone restrictions, verification, and prepaid incentives are paid-efficiency work, not just operations work.

WeProms Digital has applied this measurement-first, attach-led framework across Pakistani consumable brands — dates and dry fruits, supplements, health foods, personal care. The cadences and offers change with each category’s consumption rhythm; the sequence does not. The same disciplines underpin our work across digital marketing for food brands nationwide, from the Multan and South Punjab sourcing belt to Karachi and Lahore metro demand.