Today we’re joined by a global leader in SEO, content marketing, and data analytics. She’s here to pull back the curtain on a common struggle for e-commerce marketers: taming the “black box” of Performance Max campaigns for large product catalogs. We’ll explore how La Maison Simons transformed its advertising strategy by moving away from outdated category-based segmentation. This conversation will delve into the power of performance-driven automation, the impact of a shorter analysis window, and how a consistent cross-channel strategy led to a staggering jump in return on ad spend, all without increasing their budget.
Before this shift, you segmented by category, which let best-sellers consume the ad budget. Could you share the daily challenges this “black box” approach created and how it specifically impacted the discovery of promising but overlooked products in your catalog?
It was a constant source of frustration. Every day felt like we were feeding a machine that only rewarded the popular kids. We’d see a single best-selling sweater just gobble up the entire budget, while we knew we had hundreds of other fantastic, high-potential items sitting completely invisible in the catalog. It wasn’t just about wasted potential; it was about a lack of control. We were stuck making minor, manual adjustments, essentially guessing, while the algorithm just kept reinforcing the status quo. Promising new arrivals or sleeper hits never got a chance to prove their worth because they were starved of the initial visibility needed to generate data.
Your new model uses dynamic segments like “Stars,” “Zombies,” and “New Arrivals.” Could you detail the specific performance thresholds that define these groups and walk us through the automated process of a single product moving from a “Zombie” to a “Star”?
Absolutely. The beauty of this system is its data-driven logic. A “Zombie,” for example, is a product with low clicks and poor ROAS that we might have previously ignored. Instead of writing it off, the system automatically gives it a small, controlled budget to see if it can find an audience. If, within our analysis window, it starts gaining traction—let’s say its click-through rate improves and it generates a few sales—it moves into a testing phase. If it continues to perform and crosses a key ROAS threshold, it gets promoted to a “Star,” unlocking a larger budget. The entire process is automated; a product can literally go from being invisible to a top performer without a single manual campaign adjustment from our team.
Switching to a rolling 14-day analysis window from a 30-day one is a key tactical change. Can you share an anecdote where this faster reaction time allowed you to either capitalize on a sudden trend or quickly cut spend on an underperformer?
The agility we gained was a game-changer. I remember one instance where a particular style of jacket was suddenly featured by an influencer on TikTok. With the old 30-day window, we wouldn’t have caught that surge in interest for weeks, long after the trend peaked. But with the 14-day analysis, the system picked up the spike in clicks and conversions almost immediately. That jacket was automatically reclassified and its budget was increased, allowing us to ride the wave of that micro-trend. Conversely, we’ve seen products that initially performed well start to fade. The shorter window allows us to spot that decline and reallocate the budget away from it before it can drain resources for a full month.
You applied this same segmentation logic across Meta, Pinterest, and TikTok. What unique adjustments, if any, were needed for these different platforms, and how did this cross-channel consistency help drive the 14% increase in average order value?
While the core logic of “Stars” and “Zombies” remained the same, the creative and bidding strategies were tailored slightly for each platform’s user behavior. For instance, Pinterest is about inspiration, so our “New Arrivals” there focused on strong lifestyle imagery, whereas Meta was more direct-response driven. The cross-channel consistency was the key to lifting our average order value. A customer might discover a “Star” product through a Google Shopping ad, see it again in a dynamic retargeting ad on Meta, and then see it styled in an idea pin on Pinterest. This repeated, consistent exposure across their journey builds trust and desire, making them more likely to explore and add more items to their cart, which directly fueled that 14% AOV growth.
The ROAS jump from ~800% to ~1500% without increasing ad spend is incredible. Besides improving product visibility, what were the other key drivers behind this efficiency gain, and how did your team’s day-to-day focus shift away from manual tweaks?
The efficiency gain was multifaceted. A huge driver was the drastic reduction in wasted spend. We were no longer pouring money into products that weren’t resonating. But the biggest shift was in our team’s focus. Instead of spending hours each week manually adjusting bids or moving products between campaigns, our marketers could finally zoom out and think strategically. Their days became about analyzing the performance of entire segments, understanding why certain products were becoming “Stars,” and planning broader creative strategies. This shift from reactive, manual labor to proactive, data-informed strategy was just as valuable as the ROAS lift itself; it allowed us to be true marketers again, not just campaign managers.
Do you have any advice for our readers who feel like their Performance Max campaigns are a ‘black box’ and want to regain strategic control over their large product catalogs?
My advice is to stop letting the algorithm make all the decisions. You can regain control by feeding it smarter signals. Start by classifying your products based on performance, not just category. Create dynamic segments like “Stars” for your winners, “Zombies” for underperformers you want to test, and “New Arrivals” to ensure newness gets a fair shot. Automate the movement of products between these segments based on fresh, real-time data from a shorter, 14-day window. This isn’t about fighting automation; it’s about using automation to enforce your strategy. When you do this, you’re no longer just handing over the keys; you’re giving the machine a detailed map and telling it exactly where you want to go.
