A new investigation is digging down deep into Google’s Featured Snippets.

According to the revelations, the snippets include contradictory data from similar sources, all depending on how the user phrases the search question. It’s concerning because it just goes to show how poorly the search engine can interpret data correctly.

The head of Dragon Metrics, Sarah Presch was the first to highlight how the replies are poles apart when the user puts out questions in a different manner, even if they’re asking the same thing.

For instance, looking for a link between things like coffee and hypertension could give rise to a snippet talking about the potential to raise blood pressure.

Looking for a link between coffee and hypertension gives rise to contradictory snippets from a similar article that spoke about coffee having zero long-term impacts. Similar oppositions arose when issues like politics, health, and current affairs were the topic of discussion.

The study also proved how questioning political candidates with lines including what is good and what is bad might produce very drastically different responses, despite the main leading query being the same.

This has led to some calling Google a bias machine while others spoke about how the company’s algorithms seem to prioritize material that matches what the user’s intentions might be, instead of giving rise to comprehensive and balanced data.

The results align with Google’s internal material from a few years back where engineers at the firm confirmed that some documents they could not comprehend and therefore the reply is false.

As per Google’s latest stance on this, it says those documents are really outdated so it’s better to avoid them at all costs. This does not seem too convincing as what Google ended up doing is pulling out small amounts of text depending on what people look for and then feeding them with replies that sound convincing.

Another expert who also happens to be the founder of AlsoAsked mentioned more regarding these findings. He says the company creates models to try and guess what users might appreciate but this gives rise to a feedback loop.

The results have serious implications for SEO professionals and others creating content. They mean featured snippets don’t clearly represent what the content is about. User intent also impacts how material gets interpreted and shown. Lastly, the content strategy might need changes to ensure accuracy through different question formats.

Google’s rep defended their moves by adding how users can also look for differnet viewpoints when they scroll deeper into the material. They also highlighted how features such as ‘about this result’ might assist users gauge the real data source.

So depending on the study’s findings, publishers might need to take a few actions. For starters, they should create comprehensive material that is correct, no matter how the question is phrased by the user. Secondly, they need to track how a certain Feature Snippet shows up for various search phrases. Only then can they attain success.

Image: DIW-Aigen