[Part 1/3 – Structures] How Retail Brands Can Adopt Modern Search Best Practices To Improve Performance

As match types have expanded we have seen the slow decline of granular structures in 2021. Our retail clients are seeing the best results by moving away from granular structures such as SKAG (single keyword ad groups) to MKAG (themed multiple keyword ad groups).

Furthermore, we’re now seeing strong results by consolidating match types as opposed to splitting these out into their own ad groups or campaigns.

Brands must embrace more automation with responsive search ads (RSA) becoming the only search ad type. They must ensure all ad sets contain RSA and that they’re optimised for ad strength and asset performance.

Finally, brands should test Performance Max campaigns across all of Google’s inventory. We’re seeing this campaign type work well as a sweeper alongside standard activity.

Move from Granular Structures (SKAG) to Consolidated Structures (MKAG)

Best practice account structure used to be all about Single Keyword Ad Groups, or SKAG. 

This was a highly granular account structure wherein you controlled the ad copy at a keyword level, so you’d have the flexibility to show completely different ads to every single keyword in your account.

Each keyword had its own ad group and the best practice was to set the ad headline to the keyword so that when a user saw an advert it matched exactly what they were searching for.

SKAG structure

When we set up SKAG for clients from MKAG, we used to see 15-20% uplifts in performance. Now we’re seeing 15-20% uplifts by moving from SKAG to MKAG.

Over time this has ceased to be best practice. There are two main reasons for this. 

Firstly, as Google has expanded match types to include the same meaning variants, you can no longer create these very granular structures. For example “blue socks” means the same as “navy socks”, so navy socks could trigger either the “navy socks” advert or the “blue socks”, resulting in one of the structures being redundant.

Secondly, it silos volume resulting in limited learning. With RSAs however, there are over 48,000 different combinations of 15 Headlines and 4 Descriptions, which Google continually rotates in order to find the optimal ad for Click Through Rate (CTR).

It doesn’t take a PPC expert to realise that you need to give each RSA as much volume as possible in order for them to test and learn and find the best ad copy! Google has suggested that you have at least 3000 impressions per week for them to work optimally.

Seeing as volume comes through at the keyword level, this means keywords need to be grouped together into ad groups so you can maximise the volume through your RSAs.  

Hence the new best practice of Multiple Keyword Ad Groups – or MKAG.

MKAG structure

These work like the structure above where you group same meaning variants together at ad group level.

These consolidated ad groups then contain Responsive Search Ads that are being tested against each other.

Consolidate Match Types

Splitting out match types into their own campaigns or ad groups has long been best practice. 

Ensuring coverage is exhaustive but mutually exclusive by excluding exact match from phrase match keywords to ensure search queries are matched into the right keywords.

However, as Google has now introduced query level bidding, where bids are made at the search query level, there is no longer a need to do this.

We have tested consolidating match types into a single ad group and seen conversion rates increase by 13% on average across our retail clients.

A knock on benefit from consolidating match types into a single ad group is that you drive more volume through your RSA (responsive search ads), improving their performance as they are data intensive.

Use a Responsive Search Ads First Strategy

This year brought Google’s announcement that Expanded Text Ads (ETAs) would sunset in June 2022, to be replaced by the much more data-intensive Responsive Search Ads (RSAs).

Responsive Search Ads work by providing Google with up to 15 headlines and 4 descriptions and Google’s AI determines which combinations should be served based on the search query. 

Google will allow advertisers to “pin” headlines so that you can still create static ads which lenders for example may find useful if you want to show your APR in one headline for compliance reasons. But typically the best practice is not to do this.

Don’t pin headlines. Data from Optmyrz has shown that ads that don’t pin headlines have outperformed pinned headlines in terms of CTR, CPA and CPC metrics.

There are three main actions that you need to take to prepare your account for the transition to RSA.

Firstly, you need to look for opportunities to consolidate lower volume ad groups to drive more data through RSAs as they don’t share data with each other. 

Consolidate similar ad groups. Google has suggested that ad groups should have at least 3000 impressions per week for RSAs to work effectively.

Secondly, you should audit your account and ensure that every ad group within your account contains at least one RSA (although you’re able to have up to three for testing).

Thirdly, you should make sure you fill out your ads completely. 

Use all 15 headlines and 4 descriptions. According to data from Optmyzr RSAs that have 13-15 headlines returned 90% more conversions than RSAs with 3-5 headlines. 

Data Credit – Optmyzr

So, as part of your audit ensure that every RSA is fully completed with 15 headlines and 4 descriptions.

Finally, you need to test your RSAs aggressively. This doesn’t work in quite the same way that ad testing has traditionally worked where you create two ads, change a single variable and see which performed best. 

With RSA you should instead look to test individual assets (headlines and descriptions) that make up each RSA.

There are two main metrics that Google gives you to help you improve performance. Ad Strength indicates the effectiveness of a relevant ad, and the specific action item(s) which can improve the strength of the ad such as making the headlines more unique or adding more headlines. 

Increase ad strength to increase reach. According to presentations by Google, every increase in ad strength corresponds to approximately a 3% uplift in clicks. 

So going from ‘poor’ to ‘average’ should lead to about 3% more clicks, and going from ‘average’ to ‘good’ will drive another 3% in clicks.

Firstly, by providing you with Ad Strength from poor, to average to good that relates to the overall set of assets in the same way they do for RSA.

The second indicator of performance looks at the asset level performance. Google will provide you with an “Asset Performance Rating” once you’ve got 5k impressions over a 30 day period, that runs from low through to good then to best. 

This allows you to see how individual headlines and descriptions are performing and you should focus on removing the low performing assets and replacing them with new ones to test if they increase the rating.

Launch Performance Max Campaigns 

Performance Max is Google’s latest automated technology which launched to all advertisers in November 2021. 

Performance max allows retail brands to find additional conversions across all of Google’s Inventory including Shopping, YouTube, Display, Search, Discover, Gmail, and Maps.

It works by using Google’s smart bidding technology based on a specific conversion goal. They can be set up easily using a set of text, images, and video assets.

We’re seeing strong performance with this campaign across our retail clients and would recommend that brands test it out. 

The best way to think about these campaigns is not to think about them taking over from your existing activity but more as a way to find new customers beyond that in a controlled manner.

They work in a similar way to how you may already be using Dynamic Search Ads (DSA) in that they sweep up missed search terms that might have been missed by your keyword coverage but for all of the other ad formats available within Google’s ad inventory. 

From our early observations, we’re seeing them largely focus on Display and Discovery placements.

Performance Max sweeps up profitable conversions in the gaps between your standard Search, Shopping, Display, Youtube and Discovery activity.

With Performance Max there are three main levers that you can pull to optimize your ads. 

Firstly, you can test different bid strategies & bid.

Secondly, you can test different conversion goals in your customer journey to see which delivers the best overall profitability. 

Finally, you can (and should) extensively test your text, images, and video assets. 

We would suggest taking your best-performing assets from your search, video and display and discovery campaigns to create your first Performance Max campaign.

Google provides feedback on ads’ performance in two main ways. Firstly, by providing you with Ad Strength from poor, to average to good that is related to the overall set of assets in the same way they do for RSA.

They also provide asset-level performance for each of your headlines, descriptions, images and videos so that you’re able to see their performance running from low, to good to best.

You should focus on providing a full and varied set of assets and ensuring that you cycle through assets removing the low-performance ones and replacing them with new assets to test.


To thrive in 2022 retail brands will have to test ever more automated solutions such as Performance Max to continue to find performance gains. 

They must let go of granular structures and start to embrace more consolidated ones now keywords act more like audiences and focus more on strategy over granularity.

They will need to embrace automation in their ads with RSA being the default ad type and will benefit from adopting early and extensive testing in the first two quarters of 2022.

About Wes

Wes is the Managing Director of DemandMore. He is columnist for several leading marketing publications including Campaign Magazine, Econsultancy and Search Engine Land where he shares his expertise in search marketing.