Advanced Feed Management Features

AI has been the word on everyone’s lips the last few years. Since ChatGPT launched in November 2022, AI technology has expanded at an incredible rate and is now everywhere, built in on your mobile phones and integrated into cars. 

Google has pushed this technology with the launch of its Performance Max campaigns, replacing smart shopping, by using AI to adapt ad formats, bidding, create new assets and generate headlines and descriptions among other things. 

While there is much debate on whether AI can do these jobs better than a trained professional, what is clear is Performance Max is the future of shopping ads and is here to stay. With that in mind, what else can AI do to help drive your business forward, and what other tools can help with managing your PMax activity? 

FeedGen

Managing your feed, and ensuring descriptions, titles and other attributes are sufficiently filled out can be a very time-consuming task and feel never-ending when you are managing many products.  

Feed Gen is an open-source product that uses generative AI to create compelling descriptions and fix feed quality issues within your product feed. Not only does FeedGen analyse the text data in your feed, but it can also analyse your product images, to amend errors or add value to your descriptions.  

For example, if you have a black t-shirt, but have inputted “White” in the colour attribute, Feed Gen can analyse the image, see that the product is black and amend your feed to change “White” to “Black”. If the t-shirt is also, for example, long-sleeved and crew neck, Feed Gen can pick up this information to add those important keywords to your description. 

Implementing FeedGen, therefore, could lead to the following 3 benefits: 

1. Improve your performance, as higher quality content increases coverage and boosts click-through-rates as a result of better search matching. 

1. Greater feed quality, by filling in missing attributes, leading to a reduction in rejections and improve overall feed quality. 

1. Save significant time spent manually optimising the feed, by automating with FeedGen, allowing for more time on value-added tasks within your business. 

We would like to note, that FeedGen is not an official Google product and it will also not be suitable for everyone. What we have noticed so far is this tool works far better with product feeds that require improvements and have pre-existing issues with rejections and poor-quality feeds. For those of you with pristine, high-quality feeds, already in place, you may find the qualities of FeedGen to be minimal. 

However, this technology is still in its infancy. In the future, we would like to see AI-generated content for feed management improve and expand into optional attributes (we are thinking additional image links, unit pricing measurements, energy efficiency classes and beyond!) as well as the basic mandatory fields. 

For more information visit Feed Gen. 

FeedX 

Do you like testing changes to your product feed, to find the best strategy for optimal performance? It can often be frustrating to fully understand the impact your test has had, if there is a lot of anomalous data, caused by seasonality, promos and other external factors. How do you know if the improvement you have seen in performance wouldn’t have happened anyway? 

This is where FeedX can come in, to help better understand the impact of tests you run. 

FeedX looks at your historical performance to identify appropriate runtimes, trimmings etc. for your test. Then as you start your experiment your feed is split between control and treatment, using a supplemental feed for treatment items, uploaded onto Merchant Centre. 

If you are running a crossover experiment, then halfway through the test swap over the control and treatment groups, to observe every item with and without the treatment. 

You also have the option to remove your highest-performing products from the test, as these can sometimes hurt experiment sensitivity. 

Once the experiment has completed FeedX analyses your performance to see the impact on impressions, clicks, conversions and conversion value. With this data, FeedX can provide a predicted annual impact of these changes (for example a 10% increase in revenue). 

For more information visit FeedX. 

Feed Segmenter 

Segmenting your products is one of the most important things you can do to achieve effective performance across shopping, paid social media, and beyond. There are a lot of good reasons to segment your products, such as: 

• Seasonal demand – ensure you are investing in the right products at the right time of year. For example, swimming trunks in summer and woolly coats in winter. 

• Overstocked inventory – Have a lot of stock you need to clear and want to push it hard? Segment your inventory in the most profitable way. 

• Long sales cycles – Some products (especially high-value items) often see a longer than average conversion window between first click and purchase. Segmenting these products can help to ensure they are not deprioritised.  

There are a few different ways you can segment your products, such as using Listing Group Rules through Google Ads, using GMC Feed Rules and now, for those of you with large inventories there is Feed Segmenter, which harnesses BigQuery. 

Feed Segmenter brings in first-party attributes, such as inventory levels and conversions and significantly reduces the effort required to manage segmentation within your feed. Based on the Google Cloud, it utilises BigQuery data transfer to bring in those first-party attributes. Managed through Google Sheets, it allows for non-technical users to set up multiple filtering rules, without having to use the GMC. 

Filtered products are then labelled with a custom label and made available to the GMC as a supplemental feed. These changes once pulled through onto Google Ads will now be visible and can be selected in your listing groups custom labels. 

For more information visit Feed Segmenter.  

Mad Pmax 

As you may already know, PMax uses the power of Google AI to drive conversions across shopping, display, YouTube and a range of other Google channels. These campaigns are comprised of Asset Groups, that require the user to input headlines, descriptions, images, videos, sitelinks, audience signals, listing groups and more. Multiple asset groups can be assigned to each campaign, for occasions such as Black Friday, Mother’s Day, Valentine’s Day and any other significant date you require a change in assets. Assuming you are running a dozen or more campaigns, this leads to a substantial number of asset groups you are required to manage at once. 

Mad PMax utilises the simplicity of a basic Google Spreadsheet with Google Cloud to make changes across asset groups in a single place. Simply make amends to the sheet, with your latest offers and with a single click those changes are uploaded through to Google Ads. 

There are various cases when Mad PMax can help save significant time, such as: 

• Replicating PMax campaigns at scale 

• Upload PMax assets at scale 

• Create asset groups at scale 

• Prevent PMax setup errors 

To get started however, you will need Google Cloud Project, a Google Ads Developer Token and Terraform Deployment. 

More information on Mad PMax can be found here. 

If you want support or advice on any of the above or if you’re not using a feed management tool, such as Feed Manager, to optimise your feeds, get in touch with us at info@productcaster.com today.