AI marketing has been the topic of 2023. But how can AI be used in marketing and digital advertising? It's been all over LinkedIn feeds and all over the podcast space. But how will it change the way we operate in the digital advertising space?
Listen on Spotify
AI has taken a direction through ChatGPT, Google Bard, and a lot of other generative AI tools, that runs parallel to the direction that a lot of advertising platforms have been taking over the past 5 years: automation and moving a lot of the heavy lifting of advertising tasks.
First, what is Black Box Advertising? It refers to the trend with advertising platforms in the digital space leaning heavily into automated campaigns. While this automation takes a lot of the tasks that marketers would normally have to worry about away, it also takes a lot of the campaign customization and testing option away as well.
On one hand, AI algorithms can do things like bidding changes and targeting updates a lot faster than than a human. On the other hand, a lot of the levers marketers use for optimization are being removed in favor of just letting the AI do it.
Before Black Box and heavy automation, whoever had the best bidding software or the best bidding tools would win the day. Then it was whoever had the best targeting strategy could differentiate. Today a lot more of those functions have been moved AI algorithms.
For example, with tools like Google Performance Max, Meta Advantage Plus, and Amazon Automatic Campaigns, you can put money in and get money out at an expected return. And that is amazing for companies. But the levers and options a marketer needs to really optimize them and turn a good campaign to a great one have been taken away. So, you have to build your own levers.
These AI-driven algorithmic campaigns boil down to a simple equation of data + prompt = output performance. That data component is something brands are always going to have control over. They have the control over what data they're feeding into these systems.
The second piece of that equation, the prompt, is where Google, Amazon, and other ad platforms really shine.
But that equation is only going to be as good as the data that you put in. The main avenue for data for eCommerce brands is the product feed - the biggest and most important repository of data that you have. The best part; it's built right into your website and eCommerce platform.
Bad product feeds can result in a number of side effects which can be incredibly limiting for eCommerce campaign performance:
What that ends up meaning is that customers see what's on the tin - what's being advertised to them - is not what the product actually is. Then they're very unlikely to return to your brand.
There are two parts to get the maximum output:
So, it's absolutely critical to getting ahead with these new kind of automated campaigns that you have your data in line and accurate.
It's very common for eCommerce brands to simply use the built-in Shopify plugin or another off-the-shelf solution to connect their feeds to Google. And those platforms do a good enough job. But they lack some of the depth and hamper the ability to generate accurate data. More importantly, they lack the ability to provide data that tells a compelling story.
There are three major components to a good feed:
If you build your feed to include these things, you'll unlock a lot of extra reporting features to get better metrics into your business health.
Accurate Product Data is Key to Getting the Most Out of Your Feeds
Make sure that your product listings include all relevant data like color, size, material, or anything that could differentiate it.
This will give you a level of data granularity that is very difficult to get out of a basic off-the-shelf feed plugin. Feed management is going to be an incredibly important issue as we start to see AI black box campaigns mature.
As ad platforms move more and more towards automation, AI might start being implemented to generate the search headlines, potentially removing another customization lever for advertisers.
And you can see a future when some of the image generation AI could start to produce visual creative for campaigns based on what they know works.
We could be looking at a future when ad platforms are using AI to build entire display ads - copy and graphics - programmatically. Creative is an ad customization lever that we have currently. But these may also be pushed into the black box as ad platforms push more towards automation for advertisers.
This all brings up bigger questions for companies who will want to keep a level of control over their brand image. Display ads and the copy used therein are an extension of a company's brand. Marketers spend lots of time thinking about brand voice and visual style and it's yet to be seen if an AI can replicate that.
It's unlikely that feed management is going to get taken over by AI. The one thing that brands are always going to have control over is the data they're providing to Google and other ad platforms. So that's where we should focus.
Now is the time for eCommerce brands to start looking at their data and how their products are being cataloged in data feeds. This is what we mean when we say, "Use your own data to build your own levers." Because no matter what happens, AI algorithms and ad platforms need your data to do what they do. And you control that data.
The Project: A Finch client was looking to make the switch from standard shopping campaigns to Google Performance Max campaigns. We built a new product feed for them leading into a big Q4 rush. We made the switch to a highly segmented Performance Max campaign structure using the data that we provided in the feed.
They had a catalog of roughly 30,000 products. We were able to slice that down to specific segments and push budget towards specific product groups and specific product categories.
The Results: ROAS jumped compared to the previous shopping campaigns by about 42%. That was our green light to go all in. We ended up increasing their spend by 5X for the Q4 rush. That resulted in an increase in revenue of 7X - from the $200,000 range up to the $2.5M range. Just a massive increase.
The Conclusion: If implemented correctly, these feed platforms have the ability to work faster and perform much better than a human can when making snap decisions for your ads. But marketers have to pay very close attention to the data that is going into these campaigns to get the best returns.
Managing your product data feed is a very continuous process. It takes constant monitoring, especially if you have a large product catalog.
Knowing your audience and what is important to them is important. This information will guide you through how to segment your catalog and how to write the copy in a way that resonates.
At Finch we're always talking to clients about how to better optimize their feeds. A feed refresh cycle can be anywhere from once a week for larger catalogs to once a month for smaller catalogs.
eCommerce brands can do this on their own if they have the bandwidth, but there are great tools and partners out there to help as well. Finch provides feed management in our Tier 1 support package. We utilize a software called DataFeedWatch, which is an incredible tool brands can use to manage their feeds. It is a hands-on tool that takes a lot of attention to get the most out of, but it's worth it.
Productsup is another great vendor out there for feed management as well as Feedonomics - they both have service offerings to manage and optimize your feed for you.
And if you're on a tighter budget, Google Merchant Center, where you upload your product feeds into Google, offers built-in tools that you can use to to set custom labels and specific product rules. It's not the most robust and it definitely has the highest learning curve. It can be tough to implement and you have to be a little bit creative. But if all else fails, you can at least use the available levers in Google Merchant Center to optimize your feeds.
Amazon can be a little bit trickier when it comes to feeds because it's all based on Seller Central which is handled more internally. You can, however, generate feeds for Amazon using data feed tools. We haven't seen as much impact from feed optimization on Amazon. We recommend a different set of tools for optimizing your Seller Central account and your product catalog there.
Meta is catching up very quickly in the realm of optimizing product data feeds, especially with their Advantage+ campaigns which are alleviating some of the pain caused by the iOS 14.5 privacy updates.
Exporting out of Feednomics or DataFeedWatch will allow you to support Meta, Bing, and Google. It's best to have one feed exporting out to Google, a separate feed that's exporting to Facebook, and another to Bing. Determine the differentiators around what customers are looking for on those individual platforms. Then build separate feeds running to customize the messaging and data to make sure you're getting the maximum performance out of every channel.