How We Achieved 8.5x ROAS for a Melbourne E-Commerce Store

In August last year, a Melbourne homeware retailer came to us after 14 months of running Google Ads in-house. Their monthly spend was sitting at $4,200 and their return on ad spend had never moved above 2.1x across that entire period. They were generating revenue, but the margin left after ad spend was thin enough that scaling the budget felt pointless. More spend, roughly proportional revenue, no real improvement in profitability.

Nine months later, the same monthly budget was returning a verified 8.5x ROAS. The Google Ads channel had overtaken their wholesale revenue for the first time, and two of their product categories had moved to a waiting list.

Here is a complete account of what changed.

The audit: what was actually wrong

The account had been built using Google's Smart campaigns, which hand almost all optimisation decisions to Google's algorithm. Smart campaigns were spending the majority of the budget on broad search traffic for terms like "home decor" and "gift ideas". Neither of those converts reliably for a mid-range homeware brand competing against Kmart and Adairs on the same results page.

There was no separation between brand and non-brand keywords. When someone searched the store name directly and clicked a paid ad, that click was being counted as a paid acquisition at the same cost-per-click as a cold audience click. Brand clicks inflate ROAS significantly because they convert at a much higher rate. This was making the account look better than it was while hiding how badly the actual acquisition campaigns were performing.

The Google Shopping product feed had no custom labels, no margin information, no supplemental feed adjustments and no exclusions for out-of-stock products. Google was actively spending budget showing Shopping ads for products the store barely stocked and for a handful of items that were actually sold at a loss once shipping was accounted for.

Attribution was running on Last Click. This was crediting purchases entirely to the final ad click before conversion, completely ignoring the research and consideration steps that happened before the final session. When we switched to data-driven attribution, our understanding of which campaigns and keywords were actually contributing to revenue changed significantly.

The campaign restructure

The first change was separating brand keywords into their own isolated campaign with a conservative Target CPA bid strategy. Branded clicks are cheap, they convert at a very high rate and they represent people who already know the store. They belong in their own campaign so their strong performance does not distort the data from non-brand campaigns where the actual acquisition work happens.

We rebuilt the Shopping structure using Performance Max, but with a deliberate approach to asset groups. Rather than one PMax campaign with all products and all assets in it, we created separate asset groups for each major product category: kitchen, bedroom, outdoor and gift sets. This allowed budget to follow category-level ROAS data rather than pooling everything where Google had no meaningful signal about what to optimise for.

The product feed required substantial work. We added custom labels to segment products into three margin tiers. Google's Shopping algorithm optimises for conversion value, and it needed to understand that a $14 item with $2.50 of net margin should not receive the same bid priority as a $95 item at 60% margin. Without this signal, the algorithm was optimising for revenue volume rather than profit.

We built a negative keyword list of around 380 terms across 11 categories. Searches containing "cheap", "wholesale", "free shipping template" and a long list of specific competitor brand names had all been serving ads. This alone reduced click volume by 19% in the first 30 days while revenue from the account held almost flat. That is the clearest possible indicator that a large share of the budget had been paying for traffic that was not going to buy.

Fixing the landing page problem

Just under half of all Shopping ad clicks were landing on the store's homepage rather than on the specific product being advertised. This is a common problem with Shopping campaigns when product feed URLs are not set up correctly at the item ID level. A customer who clicks on a specific linen cushion listed at $89 and arrives at a general homeware homepage has to do additional work to find what they clicked on. A meaningful portion of them do not bother.

We corrected the feed URLs to ensure every product ad deep-linked to the correct product page. We also made targeted improvements to the product pages themselves. The key changes were compressed image files for faster mobile load times, product photos showing items in styled real-room settings rather than flat white backgrounds, and a cart abandonment exit pop-up offering 10% off an email capture on first order.

Results at 30, 60 and 90 days

By day 30, average cost per click had dropped 22% as the negative keyword work filtered out the low-quality traffic. ROAS moved from 2.1x to 3.4x on the same budget. The account was profitable for the first time in over a year.

By day 60, the Performance Max campaigns had trained sufficiently on the restructured asset groups and improved product feed to begin targeting much more accurately. Total revenue from the Google Ads channel was up 38% compared to the equivalent period in the prior year, with ad spend down slightly as we paused the worst-performing product categories pending further margin analysis.

By day 90, ROAS reached 7.2x. The kitchen accessories and gift bundle categories were consistently performing above 9x on their own. We scaled the daily budget for those two categories by 40%, and within the following six weeks the account-wide ROAS stabilised at 8.5x on a total spend 40% higher than the original monthly figure.

What this means for Melbourne e-commerce businesses

This account was not fundamentally broken. The products were good, the pricing was competitive and the website was functional. The problem was almost entirely structural. Poor keyword controls, no margin-aware bidding, an unoptimised product feed and the wrong campaign architecture were collectively wasting a large proportion of a budget that should have been generating a strong return.

8.5x ROAS is not an unusual result for a well-managed homeware or gift e-commerce account. It is a reasonable benchmark for a category with decent margins and a product that photographs well. The distance between what this client was achieving and what was possible was not created by a bigger budget. It was created by fixing the structural problems that were burning most of the existing one.

If your Google Ads account is generating clicks but delivering disappointing conversion numbers, or if your ROAS has been flat for several months despite ongoing optimisation, the problem is almost always in the account architecture rather than the product or the market. A structured audit usually identifies the specific bottlenecks within the first two to three days of analysis.