Solved: Splitting PMax Campaigns by Category or Top Products?

Hi there,

Thanks for reaching out! I'm happy to give you some initial thoughts and guidance on your PMax situation. It's a common crossroads to be at when a campaign starts doing well, and the question of how to scale it without breaking it comes up. I've seen this exact scenario with a number of eCommerce clients we've worked with.

Here’s my take on it, based on my experience.

My initial thought: Be very careful about breaking what's already working

Okay, first thing's first. You've got a PMax campaign that's outperforming your standard shopping campaigns. That is a massive win. Before you or your specialist touch anything, it’s really important to recognise that you have a high-performing asset here. The temptation to tinker and 'optimise' further is always strong, but with machine learning campaigns like PMax, it can be a really risky business.

PMax works by learning from a massive amount of data signals across all of Google's channels. It's looking at your entire product catalogue, your assets, your audience signals, and the conversion data it's collected over time. When you split a campaign, you're effectively hitting the reset button on that learning process. Each new campaign starts from scratch. It has to go through its own learning phase, which can be expensive and unpredictable. You could very easily see a drop in overall performance, at least in the short term, and there’s no gaurantee the new setup will ever get back to where you are now.

Honestly, my first piece of advice is usually to leave the winning campaign alone. Don't mess with the golden goose. If it's consistently delivering results, let it run. Your specialist's suggestion to segment by category isn't necessarily wrong in theory—it's a classic strategy from the standard shopping days—but PMax is a different beast. What worked there doesn't always translate. Your idea of pulling out top performers is even riskier, in my opinion. You'd be taking all the best players off your winning team, leaving the original campaign with only the weaker products to work with. Its performance would almost definitly plummet.

I've seen it happen before. A client has a campaign humming along nicely, getting a great return. They get ambitious, split it up to try and squeeze more out of it, and end up with two or three mediocre campaigns that never recapture the magic of the original. Sometimes the best move is to do nothing, and instead focus your energy elsewhere.

I'd say if you must split, you should flip the strategy on its head...

Right, so let's say you've weighed the risks and still feel that segmentation is the right path forward to unlock more growth. In that case, I would strongly advise against your specialist's suggestion and your own idea of pulling out the top-performing products. I would do the complete opposite.

Isolate your low-performers.

Think about it this way. Your current PMax campaign has figured out a formula to sell your top products effectively. The algorithm is working. The combination of assets, products, and bidding is a success. Why would you want to dismantle that? It makes no sense. The top products are the engine of your campaign.

The real drag on your overall ROAS isn't the top products; it's the bottom ones. The products that get some spend but never convert. They are the dead weight. By pulling *these* products out into their own separate campaign, you achieve two brilliant things:

1. You protect your winner. Your main PMax campaign is now leaner and meaner. It's free to focus all its budget and machine-learning power on the products it's already proven it can sell. You've essentially removed the dead weight, so its performance and ROAS should, in theory, actually improve.


2. You create a safe testing ground. This new campaign, which we can call the "Low-Performers" or "Incubator" campaign, becomes your lab. You can give it a much smaller, controlled budget and start experimenting aggressively without putting your main revenue stream at risk. If it fails completely, you've only lost a small, controlled amount of money on products that weren't selling anyway. If you find a way to make them work, you've just created a whole new revenue stream.

This aproach is all about risk management. You insulate your core profit centre from unpredictable changes and create a dedicated environment to solve problems. It's a far more strategic way to handle things than just carving up your best campaign and hoping for the best.

You probably should dig into *why* certain products are underperforming

So, you've created your new "Incubator" campaign with all the products that just aren't shifting. What now? Just letting them run in a new campaign probably won't change much. The reason they're not selling isn't because they were in the wrong campaign; it's because there's a fundamental issue somewhere in the sales funnel for those specific items. This is where the real work begins.

You need to put on your detective hat and do a full audit for these products. Low performance is just a symptom. You need to find the disease. I'd look at it in a few stages, based on the customer journey:

1. The Ad-Level (The Impression & Click)

Even within PMax, you have some control over assets. Are the images and videos you're using generic, or do they specifically showcase these underperforming gifts in a compelling way? A gift store is highly visual.

• -> Low Click-Through Rate (CTR)? This tells you your ad creative or copy isn't grabbing attention. The product images might be low quality, boring, or fail to show the product in a desirable context. For a gift, people want to imagine the moment of giving it. Does your photography do that? As I've told other eCommerce clients, you'd be amazed what a difference proper photography or even a simple video of someone using or unboxing the gift can make.

2. The Product Page-Level (The Consideration & Add to Cart)

Okay, so people are clicking the ad, but they're not buying. This is the most common failure point. The problem lies on your website. You need to look at these product pages with brutally honest eyes.

• -> Lots of product page views but no 'Add to Carts'? This is a massive red flag. It means the ad did its job, but the page failed to convince the customer. I've seen this hundreds of times. The likely culprits are:
  • Product Photos: I mentioned this for ads, but it's even more critical here. Are there multiple high-res images? From different angles? A lifestyle shot showing scale or use? A photo of the packaging? Remember, they can't touch it, so your photos have to do all the work.
  • Product Descriptions: Is there one? I've seen stores with no descriptions at all. This is a conversion killer. Is it just a list of dimensions and materials? That's not a description; it's a spec sheet. You're selling a gift. You need to sell the emotion, the story, the reason *why* this is the perfect gift for someone. Professional copy here can make a world of difference.
  • Price & Value: Is the price clear? Is the value proposition clear? If it's an expensive item, are you justifying the cost? If it's cheap, does it look too cheap to be a good gift? Maybe a special offer, a bundle, or highlighting 'free shipping' could be the nudge people need.
  • Trust and Credibility: This is huge for a non-brand-name gift store. Why should I trust your site with my card details? Your whole store needs to project trustworthyness. This means having clear reviews or testimonials (and if you don't have any, you need a strategy to get them). It means having an easy-to-find contact page, a clear returns policy, and professional design. A cluttered, slow-loading, or amateur-looking site will send potential customers running for the hills.

By going through this diagnostic process for each of your low-performing products, you'll start to see patterns. Maybe all your jewellery has bad photos. Maybe all your "for him" gifts have terrible descriptions. Once you identify the problem, you can form a hypothesis and start testing solutions in your new Incubator campaign.

You'll need a tailored approach for different gift categories

This brings me back to the idea of 'categories'. Your specialist was onto something, but maybe not in the way they thought. It's not just about grouping by product type (e.g., 'mugs', 't-shirts'). It's about grouping by the *buyer's intent* and the *gifting occasion*. The person buying a novelty mug for a colleague's secret santa is a completely different customer to someone buying a personalised necklace for their partner's 30th birthday. They have different motivations, different price sensitivities, and they'll respond to completely different messaging.

Your main PMax campaign is likely doing well by finding the broad, easy-to-reach "gift buyers." But it's probably struggling with these more niche products because it's serving them the same generic ads and sending them to the same generic audience signals.

In your new "Incubator" campaign, this is your chance to get specific. I would structure it using Asset Groups based on these gifting personas or occasions. For example:

• Asset Group 1: Anniversary Gifts
  • Products: All the underperforming jewellery, keepsakes, etc.
  • Creative: Images and videos showing couples, romantic settings. Headlines like "The Perfect Anniversary Surprise" or "A Gift They'll Cherish Forever."
  • Audience Signals: People with an 'Anniversary in 30 days' life event, people interested in romantic dinners, wedding-related interests.
• Asset Group 2: Gag Gifts / Office Humour
  • Products: All the quirky, funny, underperforming mugs and desk toys.
  • Creative: Fun, bright, humorous images. Headlines like "Win the Office Secret Santa" or "The Leaving Gift They'll Actually Remember."
  • Audience Signals: People who work in certain industries (e.g., 'office administration'), interests in things like The Office (TV show), memes, comedy websites.

By giving PMax these much more specific, relevant assets and audience signals, you're giving the algorithm a fighting chance to find the *right* people for these niche products. You're moving from a scattergun approach to a laser-guided one. This level of detail is how we've managed to drive such high returns for eCommerce clients, like the 1000% ROAS for a subscription box client or the 691% return for a women's apparel brand. It comes from deeply understanding the specific audience for a specific product and tailoring the entire message to them.

This is the main advice I have for you:

As you can probably see, getting more out of your ads from this point on isn't just about settings in Google Ads. It’s about deep strategic work: funnel analysis, customer psychology, and structured, methodical testing. Tbh, this is where most ad specialists fall short. They know how to operate the platform, but the real growth comes from the strategy that informs those clicks.

This process of diagnosing and fixing underperforming assets is exactly the kind of detailed work that can turn a struggling product line into a profitable one. It takes time and a particular kind of expertise, but it's the foundation for sustainable scaling. We've seen this time and again, whether it's driving a 633% return and a 190% increase in revenue for a cleaning products brand.

If you feel this level of analysis could be beneficial, and you'd like an expert eye to go through your account and map out this kind of strategic plan in more detail, we offer a free, no-obligation initial consultation. We could have a proper look at your campaigns and product funnel together and give you a clear roadmap for what to do next.

Hope this helps!

Regards,

Team @ Lukas Holschuh