PLA Query Segmentation - How to Drive Stronger Google Shopping Performance with PLA Query Segmentation

Product Listing Ads (PLAs) for Google Shopping are powerful, but don't let you target clicks by individual keywords, like AdWords. PLA query segmentation can bridge this gap, cut your costs and drive stronger overall performance for Google Shopping.PLA Query Segmentation - How to Drive Stronger Google Shopping Performance with PLA Query Segmentation

PLA query segmentation is an important workaround for targeting shoppers with buying intent. As you probably already know from our PLA blog series, Google Shopping doesn’t offer a clear way to segment product listing ads (PLAs) based on keywords, the way AdWords does for text ads. However, PLA query segmentation is a way to bridge this gap, acting as a surrogate to bid up on higher-converting keywords without specifically targeting them individually. The result?

  • No longer wasting spend on poorly-converting queries
  • Higher conversion rates (CVR) due to increased capture of high-intent clicks
  • Higher overall sales

How does PLA query segmentation work? 

Since the main goal of Google Shopping is to drive sales, we use PLA query segmentation to effectively segment our audiences by the purchase intent signaled by their keyword choices. For example, a general search like “Women’s hiking boots” generally implies low purchase intent. The purpose of this query is likely just browsing and gathering information. Accordingly, Google returns a wide variety of products that fit this query: 

PLA Query Segmentation - The general search for head term "women's hiking boots" returns varied results.

Conversely, a query like “Women’s North Face Storm III waterproof hiking boots” suggests a searcher who has already done their initial research, and may be more ready to buy. 

PLA Query Segmentation - More-specific search for a brand term returns fewer results, but perhaps with more purchase intent.

Of course, because it’s so much more specific, a query like “Women’s North Face Storm III waterproof hiking boots” will, of course, have a lower search volume than a more-general, high-traffic head term like “women’s hiking boots.” 

However, if you’re running PPC campaigns for a major retailer selling this type of footwear, you know you need your PLAs for this brand and style of shoe to get the best placement possible - you’re competing against generalist retailers, specialty apparel outfits and resellers on eBay, among others. 

Beyond ranking for high-traffic, general head terms, you also need to target searchers running queries that show high purchase intent - otherwise, you risk missing out on opportunities for your products to appear most prominently in search results for high-converting queries that would have driven more actual sales. This is why we need PLA query segmentation - a bidding strategy that helps segment queries for purchase intent. So how does it work?

PLA query segmentation: Vary bids vs. implied purchase intent

Unfortunately, PLAs don’t let you adjust bids based on query - only on product groups. PLAs get query-matched and served for a variety of different relevant queries that you can’t directly control, other than excluding those queries and specific keywords from your campaign targeting entirely. This is why we need PLA query segmentation, which uses priority settings at the campaign level by classifying queries into segments.

Our PLA query segmentation strategy boils down to:

  • Using negative to filter out specific, high purchase-intent keywords from general, non-brand campaigns
  • Setting these general, non-brand campaigns at high priority but low bids to decrease wasted spend on top-funnel searchers unlikely to buy
  • Setting highly specific, brand + high purchase-intent campaigns at low priority but high bids to ensure you capture bottom-funnel searchers ready to make a purchase

To do this properly, we have to do our PLA query segmentation, separating the high purchase-intent keywords from high-traffic, low-converting head terms. We then have to structure our campaigns properly with regard to priority and bidding.

Segmenting based on intent with query types

As you probably already know, text ad campaigns on AdWords have numerous keyword matching options that are largely within your control, including exact match, phrase match, broad match, broad match modifier and negative match. For AdWords campaigns, this lets you bid higher on exact match keywords than on broad match queries with higher traffic but lower CTR.

With text ads, you’re able to bid higher on exact match keywords than you would for broad match queries that have a lower clickthrough rate (CTR). With PLAs, you can’t target keywords directly, but as part of PLA query segmentation, you can use negative keywords and priority settings to nudge clicks to the right campaign for certain search queries. 

Here are 3 different query types that signify varying stages of the path-to-purchase: 

Query type

Examples 

General queries 

Men’s tennis rackets, Women’s hiking boots

Brand-specific queries

Wilson tennis rackets, women’s North Face STORM III waterproof hiking boots

Product-specific queries (containing SKU ID, manufacturer part number, UPC code, or highly specific product identifying information implying strong purchase intent)

Wilson tennis racket MPN WRT73901U1, Women’s North Face Storm Waterproof Hiking Boots Gray Size 6.5 On Sale

Say you’re the PPC manager for Wilson Tennis, and you set bids at $1 for the product “Pro Staff 97 Tennis Racket.” Generally, brand-specific and product-specific queries are all relevant to your PLA for this product. Here’s the problem: You spend $1 no matter if someone searches for “Wilson Pro Staff 97 Tennis Racket” or “men’s tennis racket.” 

That’s why it’s important to segment queries for your PLAs. If you take a look at cost per click (CPC) for your general queries, brand-specific queries and product-specific queries, you might find they’re quite similar. But it still makes sense to invest more in high purchase-intent keywords, as the revenue generated per click will likely be higher.

Structuring your campaigns for PLA query segmentation

To get started on PLA query segmentation for your keywords based on purchase intent, you’ll need to create three separate campaigns for the same product: General, brand-specific, and product specific. 

The best place to look when defining your segments is your past search query performance data in AdWords. You could also look at data from just your Shopping campaigns, or account-wide dimensions. Either way, it’s important to focus on isolating keywords that have a strong conversion rate over time. 

Using PLA query segmentation for branded and product-specific keywords makes sense it:

  • You see a lot of relevant product-specific queries in your search term report, and...
  • There’s a significant difference in conversion rate and CPA between branded and product-specific queries. 

In this instance, it would be worthwhile to create a bidding strategy that prioritizes product-specific queries over branded queries. You then use negative keywords will as a filter to avoid targeting the same queries with each campaign.

Here’s how you can set up each of your campaigns, including priority settings, bid adjustments, and negative keywords: 

Search terms

Priority settings

Negative keywords

Bid

Generic queries (e.g. men’s tennis rackets)

High

Brand-specific and product specific keywords

$1.00

Brand-specific queries (e.g. Wilson tennis rackets)

Mid

Product-specific queries

$2.00

Product-specific queries (e.g. Wilson Pro Staff 97 Wilson tennis racket MPN WRT73901U1)

Low 

N/A 

$3.00

Since your product-specific campaign is the lowest priority, the only negative keywords you need to exclude are keywords you would normally exclude from your campaigns overall, namely keywords with the undesirable combination of high cost and low conversions.

Here’s a walkthrough for setting your campaign structure up in Adwords:

  1. Create your campaigns - If you already have a Shopping campaign setup, you can simply duplicate it to create a brand and non-brand campaign, or like in the example above, create non-brand, brand, and product-specific campaigns.  

    If your research shows that branded keywords have better conversions and CPA (which, logically, they should), then make sure you retain your original shopping campaign to serve as your branded campaign. It will have the most historical conversion data - an important resource you don’t want to lose. 
     
  2. Add your negative keywords - You need to separate negative keywords out for your PLA query segmentation by adding product-specific queries as negative keywords to both your brand specific and non-brand campaigns. You’ll also need to set negatives for brand-specific keywords in your general campaign.
    Just like you with any other campaign, it’s also a good idea to include any low-value keywords as negative keywords for each campaign. 
     
  3. Assign priority settings - Your priority settings will tell AdWords which campaigns to serve first. In AdWords, go to the campaign you want to edit, then:

    Click the “Settings” tab
    Select “Shopping Settings” (Advanced)
    Next to “Campaign Priority,” click Edit.

    Then you’ll see the option to change your priority settings to high, medium or low. 

    Your non-brand campaign should have the highest priority setting so Adwords references it first. Your negative keywords will make sure that more-relevant search queries are passed on to your brand-specific or product-specific campaigns. 
     
  4. Adjust bidding - Now you can reallocate your spend to focus on high-converting keywords, hopefully driving more sales in the process.

    Your non-branded campaign should have the lowest bid. Again, the reasoning should be clear - non-brand head terms have extremely high traffic but extremely low CVR. There’s no point in letting these top-funnel keywords eat up your budget if they’re not going to drive sales. Your branded campaign(s) should have slightly higher bids.

    You can do this directly from AdWords Editor:
    - Go to “Recent Changes”
    - Find your campaign 
    - Go to “Keywords and Targeting”
    - Select “Product Groups” from the drop-down menu 

    Next, you can sort the data by Max CPC, highlight the values, then set your bid higher. Post changes and you’re done. 

With these changes, your more-specific campaigns should have higher conversion rates than your general, non-branded campaign. However, since the keywords in these pre-filtered campaigns are more specific, they will also get significantly less traffic. The lower traffic should enable you to bid higher here to ensure you close those deeper-funnel conversions with shoppers who are much more ready to make a purchase. To make better decisions on ideal bid levels for different business cases, we recommend looking into a predictive advertising solution.

Takeaways

PLA query segmentation isn’t necessarily intuitive, but its results - cutting wasted spend on non-converting head terms and driving more sales through high purchase-intent queries - are worthwhile. We recommend regularly reviewing query patterns over time to find new ways to target high-intent users, such as using segmentation for in cases of seasonality (bidding up on hockey supplies in the winter and bidding up on beach blankets in the summer).

About the Author

Andrew Park

Andrew Park is a content marketing manager at QuanticMind. A UC Berkeley graduate and lifelong Bay Area resident, Andrew has done tours of duty in editorial, PR and marketing, and now works with the QuanticMind team to communicate the importance of data science and machine learning in digital advertising.

More Content by Andrew Park
Previous Article
Enterprise Bid Management Solutions - Is Automation a Threat to Marketers?
Enterprise Bid Management Solutions - Is Automation a Threat to Marketers?

Enterprise bid management solutions are complex, but offer time-saving automation. But what are bid managem...

Next Article
The Customer Journey Is Changing. Is Machine Learning the Answer?
The Customer Journey Is Changing. Is Machine Learning the Answer?

EBOOK

A Beginner's Guide to Machine Learning & Artificial Intelligence

GET THE GUIDE