Every now and then you might be able to spot something going wrong at Google, with unexpected results and data leaks popping up. Here are some of the most significant times that Google’s got it wrong, along with what we can learn from them.
Search engines are notorious for keeping their cards close to their chests when it comes to how they work in terms of crawling, indexing and ranking web results. Recommendations passed onto SEO experts are often cryptic or vague such as “create great content” or “don’t be evil”, and changes to search algorithms are constant, making it difficult to catch up.
However, every now and then you might be able to spot something going wrong at Google, with artificial intelligence turning out to be not so intelligent at all, or confidential data being leaked.
These mistakes make it possible for SEOs to gain valuable insights into just how Google works, and what to do to stay one step ahead. Here are some of the most significant times that Google’s got it wrong, along with what we can learn from them…
In this blog you’ll find:
Mistake #1: Lies about legs
An increasingly visible element on search engine results pages is the ‘quick answer’ box, where users can find out the answer to a common query without clicking into a web page. Great news for users, particularly with voice search on the rise, but bad news for the websites which provide the content who are missing out on valuable visits.
Results are scraped directly from the page and served up in the quick answer box, but these can’t always be trusted. This is best illustrated by asking Google how many legs a particular animal has (the following are all real results as of June 2019):
Four legged whales:
And the same for snakes:
Better order some extra shoes for that horse.
So, what went wrong here?
Google’s bots pulled the answers from generally recognised sources like Quora and Wikipedia. But, without being properly sense checked and moderated by a human, the result has become ridiculous. What can we learn from this?
Google trusts certain sources over others
Sometimes too much, as we’ve already seen above.
It takes a lot to build a trustworthy website, but once your authority is as high as it possibly can be in your niche, you’ll often be picked as a provider of answers and images for ‘position 0’ snippets.
Answers from sources like Quora and Wikipedia, are generally a ‘last resort’
Research suggests that sites which offer specific knowledge to answer a question is preferred – so questions about cars should be answered by a car-focussed site, and questions about horses should be answered by a horse-specific site.
If you’re finding that the quick answer results for your area of expertise are taken from general sites like Quora, then creating your own reliable and authoritative content to provide a better answer will soon see you taking the answer box for yourself.
Google can misinterpret a page
When pages answer one question (how many legs do early whales have?), but contain all of the keywords for the set query (how many legs does a whale have?), then it can mean you start ranking for topics beyond your main focus.
So, make sure your content is as understandable as possible with structured data.
Google can make mistakes with spellings and variations
Thanks to the way Google’s semantic search works, the term ‘fore legs’ has been misinterpreted as ‘four legs’. Once added to the ‘two back legs’, the answer becomes ‘six’.
If your site’s content isn’t completely clear, you could be attracting irrelevant traffic. When you’re writing about nails, make sure you clarify whether that’s fingernails or metal nails.
If you’re a recruitment agency writing about catering roles, for example, make sure there’s no confusion with catering rolls.
As a bonus, here’s the terrifying truth about Peppa Pig:
Mistake #2: Image interpretation
Despite the popularity of Google images, mistakes are still pretty common when it comes to deciphering what your image contains. We’ve reverse image searched some recent stock photos from unsplash.com to show you what we mean:
1. A Skyscraper:
Google thinks this is HD (Home Depot), and brings up lots of fun facts about the big orange home improvement supplier.
2. A red and silver rooftop in Slough, just down the road from the CLD offices:
‘happy mothers day architect’ apparently. Pretty specific, but pretty far off the mark.
And as for this image, we know that it’s obviously an aubergine / eggplant. But right click on it and select “Search Google for this image” and you’ll see that Google has wildly misinterpreted things (it might be an idea to clear your search history afterwards…)
What do we learn here?
Mark up your images!
Add captions, ALT text and descriptions so that your content can be interpreted correctly. Adding all of these to your image improves the relevance of your page overall, a valuable ranking factor to help you boost the positions of your product pages, blogs and other pages.
Pay attention to image ‘quality’
When using images for products or logos that you want to have a search presence for, make sure that your image is in focus and has a plain background so it can be understood clearly.
Use original images
Stock photos are unlikely to provide any visits to your sites and won’t give your website the unique identity that helps it to stand out.
Mistake #3: Multinational Mix Ups
With Google being such a global player in the world of search, there are times when those local differences aren’t appreciated, and users end up with unhelpful results. Whether it’s different products in different countries which use the same name causing trouble or irrelevant news results creeping in, there are times when entire results pages are way off.
Take this example – a search for “Burrito” in the UK is full of US results:
Looks like there’s not much burrito based breaking news hitting the headlines in the UK.
We also see American results creeping into UK product searches. Looking for a new car? The Ford Fusion is very different in Europe to America.
A European Ford Fusion – Practical, reliable and lovable in an Eeyore kind of way.
An American Ford Fusion – Tough and brash.
Another scandal is Google’s results for searches of ‘chips’ in the UK. If you were expecting the classic partner to fish, made from fluffy deep-fried potatoes, you’re in for a shock – the first page of results is almost entirely dedicated to a US comedy cop show from the 1980s, CHiPs:
No matter, surely adding ‘potato’ to your search will bring up the right kind of chips for the UK audience…?
Google fails again, as these are quite clearly crisps. In a similar vein, a search for ‘jelly’ doesn’t bring up a sweet, translucent gelatine dessert - or even jars of jam which are known as jelly in the USA.
It’s a far more international affair as the results for jelly are dominated by a Dutch YouTuber:
Why do these things happen?
Poor international targeting by websites
Many multinational corporations are lagging behind when it comes to setting up their international targeting correctly. In fact, research from SEMRush on Search Engine Journal suggests that 75% of websites struggle to set up their HREFlang tags. These tags are important as they indicate to search engines which version of a page should be served up for which language.
At the time of writing, we noticed that both ford.com and fritolay.com were both lacking HREFlang tags which increases the changes of irrelevant results from these domains leaking into UK results (as we’ve seen with the wrong Ford Fusion and Potato Chips). Making sure you’ve got the correct set up for your global strategy means you’ll be getting the right results in front of the right users.
Location of links
Ford and Frito-Lay both have backlinks from sites all over the world, which makes it more difficult to identify their target location.
We ran both sites through our backlink tools which revealed that Ford had over 1000 German and 700 UK domains linking to the site. For Frito-Lay, links came from the UK, Denmark, Greece and Portugal despite their North American focus.
If you’re trying to get your site ranked in a particular country, backlinks play a large part. Gaining natural links from sites based in the same location as your target market rather than just looking at link metrics can help you achieve long lasting rankings.
How to beat Google
When Google’s mistakes are coupled with omissions or issues with the leading brands’ sites, it can be possible for smaller sites to quickly and easily do things better and take their place.
As mistakes become rarer and artificial intelligence becomes more sophisticated, the focus will soon shift again. Concentrating on making your site the best it can be in terms of content, linking, mobile friendliness and technical quality should help you rank well over the long term - despite the shifting sands and temporary blips of Google.
Talk to us to get your organic search strategy sorted and start beating Google today.