How to Fix Failing Google Ads for Ecommerce: The PMax Feed-Only Strategy for Beauty and Cosmetics Brands

I recently had a consulting call with a beauty and cosmetics ecommerce brand, and the conversation was so packed with lessons that I wanted to share the key takeaways with you. Because frankly, the mistakes they were making are ones I see all the time, and the fixes are surprisingly straightforward.

Here is a quick snapshot.

This brand sells around 600 to 700 SKUs, targets busy mothers, and has two customer segments: high-value customers purchasing full makeover packages ($150 to $300) and single-item buyers ($19 to $21 each). They were crushing it on Amazon with a 4 to 4.5x ROAS, but Google Ads was a disaster. Their ROAS was sitting at a painful 0.3 to 1.1x despite spending $100 a day for months.

So what went wrong, and how did we fix it?

Let me walk you through everything.

They Gave Google Everything It Asked For

This is the number one mistake I see ecommerce brands make with Performance Max. They followed Google's recommendations to the letter: loaded up 30+ images, videos, headlines, and descriptions. They gave Google every single asset type it requested.

And Google rewarded them with a 0.3x ROAS.

Here is the thing. When you hand Google that many variables to play with, the algorithm has no idea what to prioritise. Your budget ends up scattered across YouTube, Display, Discover, Gmail, Maps, and Shopping. For a $100 per day budget, that is a recipe for failure. Too many channels, too many creative combinations, and nowhere near enough data for any single element to optimise properly.

On top of that, their product titles were generic Shopify defaults rather than keyword-rich titles like they were using on Amazon. And their Merchant Centre was occasionally showing 15,000 products instead of their actual 600 to 700 SKUs, which is a data feed issue that needed sorting out.

The Fix: PMax Feed-Only

My primary recommendation was to strip everything back and run Performance Max in feed-only mode.

What does this mean? You attach your product feed to PMax, but you do not provide any videos, headlines, descriptions, or images. None of it.

When you do this, Google can only run Shopping ads. It cannot show your ads on YouTube, Display, Discover, Gmail, or Maps. You are forcing all of your budget into the single highest-intent channel available.

This dramatically simplifies your setup and your optimisation. Instead of trying to figure out which of your 30 images and 15 headlines are working or failing, you have one lever to pull: your product feed. It focusses your spend on the lowest-hanging fruit, which is people actively searching for products like yours.

Start With Your Top 20% of Products

The next piece of advice was equally important: do not throw your entire catalogue at Google from day one.

Start with your top 20% of SKUs only. Your best sellers. The products you already know people want to buy. Include all colour variants of those top sellers, because each variant counts as a separate product and gives you more opportunities to appear in search results.

This is a critical point. If you have a lipstick that comes in 12 colours, each colour variant should be a separate product listing. More variants means more chances to show up when someone searches. This brand was already doing something similar on Amazon, so applying the same logic to Google Shopping was a natural step.

Only expand to additional products after you have proven profitability with your best sellers. Let PMax prove itself before you add complexity.

Your Product Titles Are Your Biggest Lever

If there is one thing I want you to take away from this, it is this:

Your product titles in Google Shopping are essentially your keyword targeting.

Google matches search queries to the words in your product titles. If your titles are generic, you are invisible for the searches that matter.

The fix is to create a supplemental feed to customise your titles without changing anything on your actual website. Google Shopping allows up to 150 characters in product titles (compared to 200 on Amazon), so you have plenty of room to stuff in relevant keywords.

Think about how your customers actually search.

If you sell a "Rose Glow Foundation," your Shopify title might just say exactly that. But your Google Shopping title should include terms like "lightweight foundation for mature skin," "dewy finish," "full coverage," and whatever other attributes your customers search for. The same keyword-stuffing approach that works on Amazon works brilliantly on Google Shopping.

One more note on images: only your primary product image shows in the main Shopping placement. Secondary images only appear if someone clicks through to the Google Shopping tab, which is rare. So focus your image optimisation efforts on that primary image.

Bidding Strategy: Keep It Simple

Do not set a Target ROAS at the beginning. I know this feels counterintuitive, but hear me out.

Control your performance through your daily budget instead. Here is how that works in practice:

If your ROAS is high (say 10x), you can raise your daily budget to scale up. More budget means more spend and more sales, but your ROAS will naturally decrease as you scale. Only increase your budget when you are comfortable with that ROAS coming down.

Add a Target ROAS later, once you have established enough data for the algorithm to work with properly. Trying to set targets before you have conversion data is like asking someone to hit a bullseye blindfolded.

Run Branded Search as a Separate Campaign

Here is something that surprised this brand owner: feed-only PMax has no search component. It only runs Shopping ads. So if you want to capture branded searches, you need a separate campaign for that.

And yes, you absolutely should capture branded searches.

I know what you are thinking. "People searching for my brand will find me organically anyway." But branded organic CTR is only 30 to 50%, not 100%. That means 50 to 70% of people searching for your brand name are not clicking on your organic listing.

A branded search campaign captures that remaining traffic, and the economics are fantastic: expect 10x ROAS or better, with cheap clicks and sky-high conversion rates.

What You Do Not Need to Worry About

This is my favourite part of the conversation, because so many advertisers waste time and energy on things that barely move the needle. Here is what you can safely ignore when starting out:

• Customer lists and Klaviyo syncing have minimal impact at this stage.

• Retargeting campaigns can be handled separately later, if at all.

• New customer acquisition bidding is a feature to save for later.

• Negative keyword lists are not as important as commonly believed.

• YouTube advertising requires video creative and is not essential.

• Individual product search campaigns unless a product has 10,000+ monthly searches.

Focus on the fundamentals first. PMax feed-only plus branded search. That is your entire Google Ads account to start.

Use Strikethrough Pricing, Not Coupon Codes

If you run promotions, strikethrough pricing is far more effective than coupon codes for Shopping ads. Google automatically picks up Shopify sale prices, so if you mark a product down from $20 to $17, that strikethrough shows up right in the Shopping ad.

Coupon-based discounts require manual setup in Merchant Centre promotions, which is fiddly and less visually impactful. Strikethrough pricing drives higher click-through rates because shoppers can see the deal instantly without needing to apply a code.

Realistic Timeline Expectations

Expect breakeven ROAS within two months maximum. For Shopping ads, it is often achievable within one month. A 3x ROAS is a reasonable target by month two. And if you are aiming for scale, a 1.2x ROAS target is actually healthy and allows you to push significant volume.

If you are not seeing breakeven by month two, something is fundamentally wrong and needs investigating.

The Two-Campaign Formula

To bring it all together, here is your starting Google Ads setup:

1. PMax feed-only with your top 20% of products, no Target ROAS, budget as your primary control lever

2. Branded search as a separate campaign, expecting 10x+ ROAS

That is it. Two campaigns. Simplify your approach, prove profitability with your best sellers, invest in keyword-optimised product titles through a supplemental feed, and expand only after you have the data to back it up.

Conclusion

The biggest lesson from this consulting call is that simplicity wins. This brand had been spending months and tens of thousands of dollars following Google's full PMax recommendations, and it was not working. The fix was to strip everything back: feed-only PMax with top-selling products, a separate branded search campaign, keyword-optimised product titles through a supplemental feed, and patience to let the data build before setting ROAS targets. Use your daily budget as the primary control lever, focus on strikethrough pricing over coupon codes for promotions, and ignore the advanced features until the fundamentals are delivering results. Two campaigns are all you need to start proving that Google Ads can work for your ecommerce brand.

Customer Acquisition in Performance Max: Bidding Equally vs Prioritising New Customers

I received a great question recently from someone in an Ecommerce community. Their agency told them to bid equally for new and existing customers in Performance Max, saying it was a "new feature" and the better approach. But they weren't convinced.

Neither am I, frankly.

Let me walk you through the 3 different approaches to handling new customer acquisition in PMax, why your revenue reporting changes with some of them, and which option I actually prefer (spoiler: it's probably not what your agency recommends).

Does Bidding for New Customers Actually Help?

Short answer: Yes.

We've found that specifically targeting new customers does help increase acquisition rates. The question isn't whether to do it, but how to do it in a way that doesn't create a reporting nightmare for you.

When you choose the option to bid for new customers within your existing campaign, you have to assign a value to a new customer. Google then adds this value into your revenue ("Conv. Value") column in Google Ads. This is where things become contentious.

Some people call this "fake revenue" and argue it makes your Google Ads data "no longer true." They have a point, but I think they're missing something important about how we should think about data in the first place.

Option 1: Bid for New Customers in Your Existing Campaign

This is the most straightforward approach. You simply toggle on the new customer acquisition setting in your existing PMax campaign and assign a value to new customers.

The upside? It's simple. One campaign, one setting, done.

The downside? Your revenue figures in Google Ads will now include this extra assigned value. If you set new customers as worth an extra $50, every new customer conversion will show $50 more in your Conv. Value column than the actual transaction value.

Here's something important though: this extra revenue can be segmented out in your reports. If you want to bid for new customers but still see reports on revenue and ROAS that look just like they did before, you can do this. It makes the reports a bit more fiddly to set up, but it's entirely possible.

Option 2: Create a Duplicate Campaign for New Customers Only

The second approach is to duplicate your existing PMax campaign and set the duplicate to bid only for new customers. This gives you the benefit of being able to bid more aggressively for new customers without changing your revenue data in Google Ads at all.

Here's how to implement it:

1. Create a copy of your current PMax campaign

2. Set the new campaign to focus solely on new customer acquisition

3. Give the new campaign a lower ROAS target so it can bid more aggressively

This keeps your data clean. Your original campaign continues tracking revenue as it always has, while your new customer campaign operates with its own distinct metrics.

There's a catch though. This approach only makes sense if your existing PMax campaign receives at least a few hundred conversions per 30 days. Much less than that and you probably don't have enough conversion data to make a duplicated new customers campaign worthwhile. Google's machine learning needs sufficient data to optimise effectively, and splitting an already thin data set across two campaigns can hurt performance rather than help it.

Option 3: My Preferred Approach (The Unpopular One)

This is an unpopular opinion, but I actually think the best option is to bid for new customers in the existing campaign and assign them a higher value. This assumes you have a good estimate of how much more a new customer is worth to you versus a returning one.

Yes, I know. This is the option that adds "fake revenue" to Google Ads and supposedly makes your data "not true" anymore.

But here's my take on that:

There is no "true" number.

What you have are multiple different stories you can tell about the data, and each number is "true" for the story it tells.

Understanding Data as Storytelling

Let me give you an example. Google Ads conversion tracking, when using the Google Ads website conversion pixel (not GA4 importing), tracks every conversion where Google Ads was a touchpoint anywhere within the customer journey. By default, the story it's telling is: "This is how much revenue there was where Google Ads contributed at least one click to the overall customer journey."

On the other hand, a third-party tracking tool that tracks all your digital marketing, when it reports how much revenue was due to Google Ads, might be using last-click attribution. That tells a different story: "This is how much revenue there was where Google Ads contributed the last click in the overall customer journey."

Neither of these is "wrong." They're just telling different stories about the same data.

When you add a new customer value to the revenue column, you're not making it wrong (again, assuming you have good data on how much a new customer is worth to you above a returning one). Instead, you're just changing the story. The new story becomes: "This is how much value there was where Google Ads contributed to the customer journey, factoring in the long-term value of acquiring new customers."

If you can wrap your head around this new story, then I think doing this is the optimal way to optimise for new customers in PMax.

Which Option Should You Choose?

It really comes down to two factors:

Your comfort with rethinking data interpretation: If you're comfortable understanding that your Google Ads revenue figures now tell a different story (one that includes customer lifetime value considerations), then Option 3, integrating new customer value directly, is the most effective approach.

Your need for simplicity: If you prefer a straightforward approach where your Google Ads data looks exactly like it always has, then Option 2, creating a duplicate campaign, achieves much of the same benefit. You can bid more aggressively for new customers by giving that campaign a lower ROAS target, without having to completely rethink how you look at your data.

What I wouldn't recommend is simply bidding equally for new and existing customers and calling it a day. Your agency might be right that it's a newer feature, but "newer" doesn't mean "better left alone."

New customers are almost certainly worth more to your business than returning ones when you factor in their lifetime value. Your bidding strategy should reflect that reality.

The Bottom Line

If you have strong data on what a new customer is worth to your business over time, use it. Either assign that value directly in your existing campaign (my preference) or create a duplicate campaign with more aggressive bidding targets.

Just make sure, if you're going the duplicate route, that you have enough conversion volume to support two campaigns. A few hundred conversions per month is the minimum I'd recommend before splitting your efforts.

And remember: whichever approach you choose, you can always segment your reports to see the data presented in different ways. The goal is to optimise for what actually matters to your business, not to keep your Google Ads dashboard looking tidy.

Conclusion

To summarise the key points from this newsletter:

• Bidding specifically for new customers in Performance Max does help increase acquisition rates

• Option 1: Bid for new customers in your existing campaign. Simple but changes your revenue reporting (which can be segmented out)

• Option 2: Create a duplicate campaign for new customers only. Keeps data clean but requires sufficient conversion volume (a few hundred per month minimum)

• Option 3: My preferred approach is to assign a higher value to new customers in your existing campaign. This changes the "story" your data tells but is the most effective method if you understand the new narrative

• There is no single "true" number in data. Different tracking methods and attribution models tell different stories, all of which can be valid

• Your bidding strategy should reflect the reality that new customers are typically worth more than returning ones when factoring in lifetime value