Performance Max for ecommerce: Everything from feed optimisation to campaign structure

Performance Max for ecommerce is a bit more complicated (sorry, ecommerce people), but also more fun because there are more things we can do (I’ve got you covered, lead gen).

At a very high level, you might have:

• A campaign per top-level product category e.g. handbags.
• Asset group split by subcategory of that product e.g. handbags above £500.
• Optional campaign for new customer acquisition, if lifetime value and return customer rate are sufficiently good.
• Promotional periods and peak seasons like Christmas and Black Friday are particularly good times to go after new customers, because they pay you dividends into the next year.
• Dedicated remarketing campaign, which would not be Performance Max.

This is a high level example; an overview. But it’s a glimpse of what ecommerce Performance Max offers in terms of structure, conversion tracking, and audiences.

If you’re reading this, chances are you’re a seasoned ecommerce advertiser. If not, I recommend my other guides on Performance Max setup and campaign optimisation — with best practices, FAQ, and must-know information.

Otherwise, let’s dive in.

Broadly speaking, there are two types of account structure for ecommerce:

Single-category campaigns

With the single-category strategy, you might need to split your campaigns by region or state for a large country like the United States. If you have an international setup, you naturally want at least a campaign per country.

Keep it simple if you’re a low-conversion volume, low-ad spend account; have a small product range; or offer a single category e.g. lampshades. You don’t need to overthink this and can get away with one simple campaign.

Your dedicated remarketing campaign that can be dynamic remarketing. Performance Max does take precedence, but that doesn’t mean that it won’t work. It’s always a good idea to save in the region of 10-20% of your budget for remarketing. How much exactly comes down to your customer base, buying behaviour, attribution, time lag, etc. Those are rule-of-thumb numbers.

Remarketing bidding strategy depends on the platform you’re running and budget. Maximise Conversions can work, as can Target ROAS and even Target CPA. It depends what you’re doing with remarketing. Generally, start with Maximise Conversions; set a low budget and see how it performs. It’s a simple test.

Split-category campaigns

Here’s an example with bestsellers or high-sales volume products. One thing I recommend if you’re going to do this is to look at your all-channel data or back-end data for whichever category you split, because Google Ads can skew this data. A lot of people will click an advert and buy something else, so it can skew. Use all-channel data in analytics to see what the bestsellers are.

This would probably suit a Max Conversion Value bidding strategy where you use the budget to control the bidding. Set it at a level you’re comfortable with, then see how much value it can drive. Your next category or next 20 categories (however many you may have in each of your campaigns) will probably be target ROAS and an unlimited budget.

It’s not literally unlimited (well, sort of).

You control your budget with a target ROAS for ecommerce, typically. But you control your Max Conversions with budget, not bids. So an unlimited budget is essentially safer on Target ROAS, because you’re telling Google to spend the money and hit the target. It then comes down to stock availability. Common sense—if you’ve got the stock and need to make the sales, you shouldn’t limit your budget but give it some headroom.

When to split ecommerce Performance Max campaigns

Let’s get the obvious reasons out of the way: budgets, bidding, locations.

But there are many other reasons why ecommerce advertisers might consider splitting out a Performance Max campaign:

• Zero-metric situations e.g. no impressions, no clicks, no conversions
• Bundles that you want to sell separate from single-product budget and bidding
• Product launches and co-branding
• Sensitive price points (high and low)
• Seasonality
• Margins and profitability
• Discounted vs. full-price, multi-buy
• Supply conditions e.g. low-stock, overstocked

Some use cases require a bit of a different mindset, so let’s look at those.

Co-branded products

One of our clients recently worked with McLaren Racing on a co-branded product that was also a product launch. So to increase visibility, we set the ROAS really low—pretty much at breakeven—to drive up impressions for that new product, to give it a bit of a chance and a kick in the backside to do good work.

If you’re selling this kind of product and want to look good in the eyes of your partner who’s doing the co-branding review, this is a technique you can use. Split these products out, set a low ROAS, give it a bit of budget, and away you go. Or you could use Maximise Conversions as well.

Seasonality

Another forgotten split, seasonality is always a good reason. If you sell garden furniture and bring in a line of Christmas products, don’t just dump them in the main campaign. Create one for your Christmas stuff—separate budget, bidding, audience signals, etc.

What’s the best way to throw the bidding engine into a fit? Dump a one-off seasonal item into your main campaign and find out.

Kidding—don’t do that.

High-ticket products

Let’s say you sell barbecues… but also garden cabins. These items have a price difference of 20-100x, and shouldn’t live in the same campaign because of bidding, budget, and signal reasons. They’re priced differently, cost vastly different amounts to sell, and target different types of people. I’m not saying that someone who buys a cabin wouldn’t also buy a barbecue, but the other way around… that’s a bit less likely. You don’t stumble onto a website looking for a Weber to chuck your sausages on and then walk out with a $10,000 log cabin.

So in this example, we’ve got varying profit margins and price points, and we’re controlling the high-margin product with a low target ROAS. We’ve got our other categories with a mid to high target ROAS, and then we’ve got a prospecting campaign for some big value, high-price items that need a bit of nurture.

We’ve got a new customer, low target ROAS campaign running to drive volume into this account. This is not set in stone; you can expand out with two more categories or campaigns—or 200. God help you if you do that, but the option is there.

Discounted products

Something people often forget is to split discounted and full-price products. You typically have a higher ROAS target for items on sale, and a lower target for what’s not on sale and new.

This example includes a new customer campaign as well, and it could get really fancy. You could split your new customer campaigns into discounted versus full price as well. That gives you an additional layer of budget and bidding control for new customers.

You might look at this going, “Hold on a minute. You put a higher target ROAS on the discounted products?” Well, yes. Discounted products will typically increase your conversion rates, often significantly, which means you can set a higher ROAS. You might not have to; you might find that a level ROAS with your non-discounted products is fine. This is why I’ve marked it with an asterisk: It’s a stock-level player as well.

If you’ve got a very large stock of an item, you could go the opposite route: a lower ROAS to ship those discounted products, especially if it’s an end-of-line item you’re trying to get rid of. So you’d go inverse with a low target ROAS and just get the stuff bloody sold.

Items on sale are a balancing act that you might want to test. If you’ve got limited stock, you might want to actually try and go for a higher ROAS so you don’t lose out as much on the discount of the item.

Most people either dump the discounted products in the same feed and campaign as their ordinary merchandise, and don’t touch the budget and bidding—or they might just give it more budget, which is completely suboptimal.

This is better.

How to optimise your Google Ads product feed

Please don’t skip feed optimisation.

The critical things are product title and pricing. Category does have an impact, and if you don’t add categories to products, Google will auto-categorise. It does an okay job, but I wouldn’t rely on it entirely.

Product titles should be keyword-centric, not brand-centric unless you’re selling specific brands. I see this mistake quite often where advertisers stick their own brand at the start of a product title. And then they wonder why they’re getting so much brand traffic to their products.

Google is a comparison shopping engine, and most people are price-sensitive. So pricing will directly impact auctions and therefore, impression share. From a potential customer’s perspective, that will attract you more clicks. Pricing is your lever in a competitive market, either to gain market share or reduce it—if that’s what you need to do for whatever reason. That’s why margins start becoming a bit more important if you do have products with a variety of them.

Managing ecommerce budgets in Performance Max

With ecommerce budgets, you split it 80/20 or 90/10 between Performance Max and dedicated remarketing. The remarketing budget may get eaten up by Performance Max, but see how you get on.

If you run Standard Shopping, assume that as soon as you put Performance Max in place, it will steal from those campaigns. So if you’re thinking about migrating, assume the worst. I recommend—particularly if you’ve got a large product count—that you try a low-risk category on Performance Max first. Just don’t dump your non-converting categories on it and then wonder why it’s wasting money. Give it a chance; be fair to it.

Another interesting thing—and the reason why I like running dynamic remarketing with Performance Max—is because if it does kill it off, you at least know by how much. It gives you a sense of what Performance Max could be doing in terms of remarketing, because it’s rather hidden; you can’t specifically report on it.

Performance Max bidding strategies for ecommerce

If you’ve got a new account in the ecommerce space, you probably want to start with Maximise Conversion Value and switch to a Target ROAS a few weeks to a month later. If you’ve sat there for a month and your spend’s gone haywire, and you’re under where your ROAS needs to be, either reduce your budget or stick a target ROAS in. You control your ROAS by adding or removing budget, so if your ROAS is positive, increase your budget gradually—a safe, arbitrary 10-20%. Can you increase your budget by 50% or more? Yes, but it’s far more nuanced and you’d need to do some calculations and predictions. If you want to understand the Google Ads auction in more detail, then watch my video.

If you’re migrating from Standard Shopping in an existing account, use Target ROAS and set it a bit higher than current performance. The reason you can do this is because your Standard Shopping already has data. Another reason for increasing your ROAS target is the assumption that Performance Max will slightly outperform initially, because it’s using remarketing and generates a higher ROAS for roughly the same cost. So it gives you a little bit of a bump—maybe 5-10%, nothing magical.

But it will annoyingly go after branded terms, so make it work a bit harder for you if you migrate from Standard Shopping to Performance Max.

Performance Max optimisations unique to ecommerce

It’s okay to just exclude poor performing products. It’s not the most optimal optimisation; you really should fix those products. But sometimes, certain categories or products simply don’t form. You just might not be competitive on pricing, for example. You should give these neglected products a chance elsewhere.

So say you’ve got no-impression or very low-impression products. You might split those out and give them a different budget, bidding strategy, audience, assets, etc. In doing that, you can fully populate asset groups with properly tailored assets. If you’re selling hats and handbags, and you’ve got them in the same asset group, how are your images going to be specific? You’d have to include both in the same image. Then it looks like a bundle or combo offer.

And do consider the “on sale” campaign, because it’ll perform differently. You don’t want to mix your discounted products in with your regularly priced ones. That will hurt your ROAS and screw up the bidding, to be frank.

Improving ecommerce conversion quality

In ecommerce, you’ve got macro and micro conversions.

Macro conversions are transactions where someone gives you money. Micro conversions are indicators such as basket ads, individual page views, and product or category views.

Enhancing that data allows Google’s bidding algorithms to find more potential customers like the ones who are most profitable to your business. Considerations here are around lifetime value and new vs. returning customers, which you can use in the bidding systems with Performance Max. You could use it in Smart Shopping, previously.

Ecommerce conversion quality is about preparing for profit. Many advertisers use a fixed target ROAS, which is fine as long as you’re feeding back profit margins into your account i.e. labelling the feed.

In this example, we’re assuming we’ve spent 10 (dollars, pounds, whatever) to get one conversion. We’ve got three categories with the margins.

• Category one: 100 conversion value, 10x ROAS
• Category two: 50 conversion value, 5x ROAS
• Category three: 40 conversion value, 4x ROAS

It appears category three is the loser here, so you’d reduce the bidding target, lower the budget, or pause the category because your ideal in this example is 5x ROAS.

When you consider margins, the story completely changes. Note now that the losing category three is actually more profitable than the higher-ROAS category two. The actual return on investment (not just the ad spend) is higher. You deal with this by labelling your product categories or even your products based on margin, and then you adjust your target ROAS per category where possible to suit the actual end ROI accounting for margin levels.

This is something people will often miss or skip, but it’s one of the reasons why you would segment a Performance Max campaign. You can’t control the bidding or set bidding at asset group levels, so you have to split out by category if you have varying margins and profitability. If this is you, put pen to paper and if you see the same situation happening, that’s a good sign you should be splitting.

Running Standard Shopping alongside Performance Max

Yes. If you’ve got a really complex inventory and more budget to play with, it can still work okay. Split your feeds and your large inventory and budget between Performance Max and Standard Shopping. You could do a branded campaign as well—a blended strategy where Standard Shopping is brand-focused. Another option is a catch-all “bottom feeder” campaign with low manual bids.

If priorities are really important to you, then you can’t use Performance Max, because it’s not an option. But for most people, Standard Shopping isn’t really viable because if you’re not actively managing, optimising, and honing it, then you won’t beat Performance Max.

I shouldn’t have to say this, but if you’re not going to put the work in, then just roll over. Let Performance Max do its thing.

Naked ecommerce campaigns

What I call “naked shopping” is a no-asset, no-image, no-text, no-video Performance Max campaign. It’s essentially a converted Smart Shopping campaign and yes, they still work and perform.

What you do is create a Performance Max campaign with nothing but a product feed, and it runs a bit like a Smart Shopping campaign would do. If you find that you can’t do this because Google won’t let you, just pause or remove those assets (or build the whole thing in Google Ads Editor).

Now obviously, Google doesn’t call it “naked shopping”, but this method is in their official support docs. So they do acknowledge that this stripped-back setup is a thing, though they don’t recommend doing it. And generally, when Google doesn’t recommend something, they will squish it eventually.

Product type reports

These are quite useful if you go based on your ROAS i.e. conversion value divided by cost. You can do the easy street route of excluding a category in an asset group, or split out and improve the assets for that underperforming category. You can even do that for a high-performing category, because you might be able to get even more out of it.

The other thing this report helps you identify is those poor categories. Move them to a new campaign, give them a bit of love with some better assets, and perhaps a higher ROAS target. If they’re way under ROAS like these two examples above—the two red conversion value costs—then you can give them a higher Target ROAS, to try and force the bidding engine to do what it can with them. You’re also going to give more budget to your better performers if you pull these out of a campaign that’s doing well.

Next steps: Always stay testing

Yes, we all miss Smart Shopping. But it’s not coming back, and so we’ve got to learn how to get results with what’s available—and that’s Performance Max. I hope this guide has helped you better understand some of the intricacies specific to ecommerce.

When set up correctly, you can absolutely get great results. But it takes time and testing (and Google’s favourite, money) to figure out what works best for your account.

And if you get stuck, be sure to refer back to my God Tier Performance Max training video for a complete rundown of how to set up, optimise, and take control of these campaigns.