Why a store's 3.0 ROAS was really 0.73, and how it reached 1.8

Background

The client is a direct-to-consumer online store selling home and lifestyle products across India, with a catalogue of about 140 products and an average order value of ₹1,460. It runs on Shopify, accepts prepaid and cash-on-delivery orders, and spends on Google Ads and Meta Ads.

The owner was told by the ad dashboards that campaigns were working. Google and Meta together reported ₹9 L of revenue from ₹3 L of spend, a return of 3.0. Yet the bank balance told a different story: stock was selling, but cash was not growing.

The challenge

The reported return of 3.0 looked healthy. The real return, measured from the store's own order data after cancellations and returns, was 0.73. That is ₹73 of revenue for every ₹100 of ad spend, well below the break-even point.

The business sells at a 62.5% contribution margin after product cost, shipping and payment fees. At that margin, the break-even return on ad spend is 1 divided by 0.625, which is 1.6. Anything below 1.6 loses money on every order, before overheads.

• Two purchase tags were firing on the thank-you page, so some orders were counted twice.
• Meta and Google both claimed credit for the same order, because each used a different attribution window.
• Cash-on-delivery orders that were cancelled or returned still counted as revenue in ad dashboards.
• Campaigns optimised toward the platform's own numbers, so budget kept flowing to the wrong places.
• The product feed had missing titles and categories, and Shopping ads showed irrelevant queries.

Objectives

The owner's goal was simple: make ad spend profitable. We translated that into measurable targets.

• Reconcile ad platform revenue with store revenue, so every report uses one trusted number.
• Raise verified ROAS above the 1.6 break-even within 12 weeks.
• Reduce cost per verified order.
• Keep ad spend flat at ₹3 L per 12-week period, so any gain came from efficiency, not a bigger budget.

Strategy

We started with measurement, because you cannot fix what you cannot see. Then we changed how budget was allocated.

1. Establish one source of truth

We used Shopify order data after cancellations and returns as the revenue figure. Ad platform numbers were treated as directional only. Every weekly report showed both side by side, along with the gap.

2. Repair tracking

We removed the duplicate purchase tag, standardised event names, passed order values accurately, and set up server-side events alongside the browser pixel to reduce lost signals. We also excluded cancelled and returned orders from reporting.

3. Rebuild the account structure

Brand search, non-brand search and Shopping were separated, so brand traffic stopped flattering the numbers. On Meta, prospecting and remarketing audiences were split and excluded from each other, which reduced overlap.

4. Clean the product feed

We rewrote product titles to include the words shoppers use, filled missing attributes, and excluded low-margin products from paid campaigns.

5. Use margin to decide

Each campaign was judged against the 1.6 break-even ROAS. Campaigns that could not reach it after a fair test were paused.

Implementation

The work ran over twelve weeks with the same total budget as the previous twelve.

Before and after

Results

The most important result is one that looks like a step backwards: platform-reported revenue fell from ₹9 L to ₹7.5 L, while verified revenue rose from ₹2.19 L to ₹5.4 L. The platforms now report less because duplicate and inflated conversions were removed. The store now earns more because budget moved to campaigns that produce real, delivered orders.

Summary table

Business impact

Before the work, the ₹3 L ad spend produced a verified loss of about ₹1.63 L after product, shipping and payment costs. After the work, the same spend produced a verified contribution of about ₹37,625. That is a swing of roughly ₹2.01 L over twelve weeks without raising the budget.

Cost per verified order fell from ₹2,000 to ₹811. The owner now has a weekly number they can reconcile with their own books, and a clear rule for pausing campaigns: if verified ROAS stays below 1.6 after a fair test, the campaign stops.

There is also a quieter benefit. Because reports now agree with the bank, the owner can plan stock and cash with more confidence, and discussions about marketing are about decisions rather than whose dashboard is right.

Lessons learned

Assumptions and method

This is a modelled example, not an audited client result. The storefront, website and dashboard images are designed mockups, not screenshots of a client account. Assumptions: ad spend of ₹3,00,000 in each 12-week period; average order value of ₹1,460; a 62.5% contribution margin after product cost, shipping and payment fees (so break-even ROAS is 1.6); 150 verified orders before and 370 after; verified revenue equals verified orders multiplied by average order value; contribution equals verified revenue multiplied by the margin, minus ad spend. The four gap components add up exactly to the difference between reported and verified revenue before the work. Real results vary with margin, return rates, product mix and seasonality. We do not promise any specific outcome.

Keep reading