I recently reviewed the Google Ads account for an Ecommerce brand running around $10k/month across European markets. They had roughly 1,600 SKUs, an average order value of about $278, and a gross profit margin of approximately 38%.
On the surface, decent numbers. But once I looked under the bonnet, there were tens of thousands of dollars in potential profit being left on the table every year. And the scary part? Almost every issue I found is something I see in the majority of Ecommerce accounts I audit.
So let me walk you through exactly what was wrong, and what I recommended to fix it.
The Single Campaign Trap
The biggest problem was one of the most common mistakes in Ecommerce PPC: running a single Performance Max campaign with all 1,600 SKUs lumped together.
When you do this, Google treats every product identically. Your hero products that print money at a 5x ROAS are sitting in the same campaign as products that have burned through $800 in spend at a 1.03 ROAS. The algorithm has no way to differentiate, so it spreads budget around without any strategic logic.
Even worse, roughly 2,000 products were "zombies", meaning they had received zero ad spend. Google had never even tested them. That is a massive pool of untapped potential just sitting there gathering dust.
The ROAS Rollercoaster
The second major issue was how they were managing their target ROAS. It was all over the place.
They had lowered their ROAS target too aggressively at one point, which drove high revenue but essentially broke even or lost money. Then they over-corrected by raising the target too high, and revenue fell off a cliff.
This is a pattern I see constantly. People panic when revenue drops, slash their ROAS target, then panic again when profitability tanks.
On top of that, they had miscalculated their breakeven ROAS. They were not accounting for 21% VAT, which meant their actual breakeven was closer to 2.8x rather than the 2.63x they had been using. That might sound like a small difference, but when you are spending $10k/month, that error adds up quickly.
They had also been optimising for add-to-cart instead of actual conversions, which was sending the algorithm completely the wrong signals.
The Visibility Problem
There were also some critical blind spots in how they were reading their data.
They had never added impression share columns to their reporting. Without those, you have no idea whether you are losing impressions because of budget constraints or because of rank (i.e., your bids are not competitive enough). Those two scenarios require completely different responses.
They also had no custom column showing gross profit after ad spend. So every decision was being made on revenue and ROAS alone, with no visibility into whether they were actually making money.
And they had never reviewed the channel-level performance breakdown within Performance Max. When I pulled up the data, Display ads were achieving only a 2.2 ROAS and YouTube ads were at 1.73 ROAS, both well below breakeven. Meanwhile, Shopping was carrying all the weight, and Search text ads were actually outperforming Shopping but receiving minimal spend.
The Four-Campaign Fix
Here is the campaign structure I recommended, moving from one campaign to four based on product performance.
Campaign 1: High ROAS (Hero Products)
Products that have spent above a meaningful threshold (say $100+) and achieved above your ROAS threshold (say 300%+) over a 120-day window. For these winners, you actually set your target ROAS lower than their current performance. The goal is to spend more on what is already working, accepting a slightly lower ROAS in exchange for higher volume and more total profit.
Campaign 2: Low ROAS (Underperformers)
Products that have spent significantly (say $200+) but are below your ROAS threshold (say under 250%). For these, you do the opposite: set the target ROAS higher than their current performance. This forces Google to either improve efficiency or reduce spend on these products. This was my recommended first move because it delivers the quickest efficiency gains.
Campaign 3: Low Spend
Products that have not spent enough to reliably judge their performance. Individual product ROAS is unreliable at low spend levels, so these stay in the existing campaign structure until they accumulate enough data.
Campaign 4: Zombies
Products with zero spend that Google has never tested. The key here is to run these on Maximise Conversion Value with no target ROAS, giving Google freedom to explore.
But critically, do not dump all 2,000 zombies in at once. Hand-select the top 100 products most likely to succeed, optimise their titles before adding them, use a moderate-to-low daily budget, and then graduate the winners to the appropriate campaign based on their performance.
The Implementation Sequence
The order matters here. I recommended:
Phase 1: Create the Low ROAS campaign first. This is the highest leverage move because you are immediately cutting waste on your worst performers.
Phase 2: Create the Low Spend and Zombie campaigns simultaneously.
Phase 3: Consider the High ROAS campaign once the system is stable.
Quick Wins You Can Steal Today
Before touching campaign structure at all, there are a few things you can do right now that take minutes and deliver immediate value.
Build a Gross Profit Custom Column
Create a custom column in Google Ads using this formula: (Conversion Value x Gross Profit Margin %) - Cost. This lets you see actual profit directly in the interface rather than making decisions based on revenue alone.
Set Up a Dummy Search Campaign
Launch a paused search campaign with a fake ad group and keyword. This unlocks additional impression share columns (Search Lost IS for Budget and Search Lost IS for Rank) that are otherwise hidden. Once you can see these, you will know exactly why you are losing impressions and how to respond.
If you are losing impression share due to budget, simply raise your daily budgets. There is no ROAS tradeoff; you are just capturing demand you were missing. If you are losing due to rank, you need to accept the ROAS vs. volume tradeoff and decide accordingly.
Test Higher ROAS Targets
Their average was around 3.7x. I recommended testing 4.0-4.5x for one to two months. Higher ROAS means less revenue but potentially more profit. And looking at their historical data, their most profitable months were consistently their highest ROAS months.
Feed Optimisation and Search Opportunities
Two final recommendations that are lower priority but still valuable.
For product feed optimisation, maximise your titles to 150 characters where possible. Treat it exactly like Amazon listing optimisation: stuff in as many relevant keywords as will fit. Prioritise this work on your highest-spend products first, since those are the ones actually receiving impressions.
For search campaigns, their PMax data showed that search text ads were outperforming Shopping on ROAS. There is a real opportunity to pull converting search terms out of PMax and build dedicated exact/phrase match campaigns around them. ROAS typically improves 100%+ when you make this move. It is higher effort, though, so I would tackle it after the campaign structure changes are bedded in.
ROAS Adjustment Guidelines
A few rules of thumb for adjusting ROAS targets safely:
For an account like this, 50-80% adjustments in either direction are within a safe range. The learning period is measured in conversions, not days. Expect volatility for 30-50 conversions after any change, with stability returning after roughly 50 conversions on the new settings. Always ignore the most recent 7 days when evaluating true ROAS performance because of data lag. And keep an eye on the bid strategy report, which shows learning periods and when changes are actually taking effect.
Conclusion
The core takeaway from this case study is that a single Performance Max campaign for all your products is almost certainly leaving money on the table. Segmenting by performance into four distinct campaigns (heroes, underperformers, low spend, and zombies) gives Google the right signals for each product group.
Your breakeven ROAS is not your profit-maximising ROAS. Aim for at least 100% above breakeven for real profit maximisation. Impression share data tells you exactly where your growth potential lies: budget-limited means easy wins, rank-limited means tradeoffs. Zombie products represent untapped potential that you need to force Google to test. And always move your ROAS targets in opposite directions for different product tiers: lower for winners (more spend), higher for losers (better efficiency).
Finally, do not forget the fundamentals. Account for VAT in your profitability calculations, check your channel-level performance within PMax, and measure learning periods in conversions rather than days.