Skip to main content

The algorithms that control how often and to whom a web post is displayed are constantly evolving. Google wants to show relevant posts for each individual search, and it is a reasonable assumption that there is a feverish activity to constantly increase the accuracy. Advertisers and post makers are keen to grab extra views with various tricks. When generative AI, known as SGE, is used to improve search results, the rules of the game change fundamentally and it is difficult to underestimate the extent of this change.

One new development is that Google has started to read text in photographs. Suddenly, labels and packaging can be part of your SEO efforts. How should you think about this? We can't get away from the fact that Google wants to make money, they run a business to make a profit. Google wants to sell ads. But it is reasonable to continue with the assumption that Google wants to fill each SERP (Search Engine Return Page) with the most relevant results for each individual search. Even though Google wants to get revenue from many ad impressions, it is precisely relevant impressions that matter. Google simply becomes a poorer marketplace with search results that are not relevant. For example, using AI to generate images with extra text, added to drive clicks, will be counterproductive.

You might want to add: "also available in red and green". This might catch a few extra searches and might also be of interest to the searcher. But if you add "also available as sugar-free" to the label of a drink that is not sugar-free, well, you are probably on thin ice. In addition, adding clickable information to an online image that is not found on the actual packaging in the store quickly enters a legal gray area.

One speculation here is that it will become increasingly important to coordinate data on labels and packaging and signage with the information available online. Now that AI-generated images have been widely adopted, I suspect that the opposite will happen: labels and packaging will be available in different variants to drive traffic. What happens then is that the credibility drops and with it the expected relevance. For example, a restaurant featured on the Google Business Profile that states that it does not serve vegan dishes may "accidentally" publish an AI-generated image of a sign on the street that mentions vegan dishes. Will Google's algorithm see through this or at least recognize that something is not right here? My reflection is that there is a lot to be gained by consistently synchronizing information and ensuring that there is no contradictory information or ambiguity. Avoid different types of packaging, labels and so on.

In the US, Google has introduced a tool called Product Studio, which aims to simplify the improvement of product images, drive traffic and generate sales. One feature that may be of interest to many small businesses is the ability to easily remove a distracting background from a product image. However, it is ultimately our responsibility as advertisers to use this tool correctly and ensure that it drives relevant search results. And more generally about the future of online advertising: if we want to capture relevant clicks, we need to bait with accurate and consistent info.

Read more about Search With Generative AI here.