SEO ·

SEO in the Age of Visual Search by Crystal-Carter of Wix

Bernard Huang

Webinar recorded on

Join our weekly live webinars with the marketing industry’s best and brightest. It’s 100% free. Sign up and attend our next webinar.

Join our newsletter

Get access to trusted SEO education from the industry's best and brightest.

We hosted Crystal Carter, the Head of SEO Communications at Wix, for a webinar on SEO in the Age of Visual Search.

Here are our biggest takeaways from Crystal’s talk:

  1. Visual search is about image recognition - search with an image and expect text

  2. Structured data is your friend - add schema markup for more information about the image including keywords

Watch the full webinar

And check out the resources Crystal shared below:

About Crystal Carter:

Crystal is the Head of SEO Communications at Wix. She is an SEO & Digital Marketing professional with over 15 years experience working with SEO and Marketing clients around the world like Disney, McDonalds, Tomy and more.

As an SEO, she identifies and implements tactics that help businesses optimise digital activity, drive sales, engagement and growth online.

She has contributed to events, webinars, and publications from Google Search Central, Brighton SEO, Moz, Whitespark Local Search Summit, Semrush, SMX, Search Engine Land, DeepCrawl, Women in Tech SEO and more.

Follow Crystal on Twitter: https://twitter.com/CrystalontheWeb

Read the transcript

Travis:

Today, we have Crystal Carter joining us. Crystal is the head of SEO Communications at Wix. She is an SEO and digital marketing professional with over 15 years experience working with SEO and marketing clients around the world like Disney, McDonald's, and Tommy, and more. As an SEO, she identifies and implements tactics that help businesses optimize digital activity, drive sales, engagement, and growth online. She has contributed to events, webinars, and publications from Google Search Central, brightonSEO, Moz, SEMrush, Women in Tech SEO, and more. Crystal, the floor is yours if you want to go ahead and share your screen.

Crystal:

Fantastic. Okay. Thank you so much for joining me all today. We're going to talk about visual search, which is a topic that I really, really like, and we're just going to jump right into it. So today, we're going to cover what I mean by visual search, what we are even talking about, and we're going to talk about how you optimize your images. We will also talk a little bit about how you optimize real life, and this is something that I think is really, really fun. I think that one of the things that has become really apparent in the last few weeks is that search is very much evolving. Right now, we're in a really, really dynamic time with regards to search and with regards to how people retrieve information, which is what they call the study of search science. They call it information retrieval science, and essentially, we want to think about that and we want to think about how people access information, how people search for information, and how visual search fits into that journey.

So they say, "What's the deal with visual search?" Essentially, the TLDR of visual search is that it's all about me or it's all about me. What I mean by that is that it's not essentially all about literally me, but it's all about me and my everyday camera roll, the same kind of camera roll that you have, which is full of pictures of places that I've traveled and fun things that I've found, fun times with colleagues and friends, really interesting things that I've seen, people that I've met, things that I've done, things that I like, things that I was interested in, silly times that I had pictures, selfies that I took.

All of these things that we have in our camera rolls, hundreds, thousands of pictures, too many pictures according to Google who's constantly telling me that I have too much, too many photos in Google Photos and that I need to pay them extra money for it, all of these different photos are now the basis of search. Not only that, but Google is using them to train their machine learning algorithms to understand search, and I'll get into a little bit more about that.

Rather than being a passive observer for all of these events, visual search is turning your camera into a primary tool for understanding the world. So this means that as a user with a regular everyday run-of-the-mill, super normal, same thing everybody has, the same camera roll, if I want to know where my mom went last week and why she's not returning my calls, then I can take her photo and I can pop it into Google search and I can copy it from Facebook, which is what I did, and I can find out that she went to New York without me. She went to see the Statue of Liberty, and not only that, but I know exactly where it is because Google's search tools can try, they use lots of different metrics, particularly for recognizing different parts of search.

So they can recognize faces for sentiment. They can recognize objects. They can recognize labels. They can add labels for descriptions. Also, they can recognize landmarks. So they can see that this is the Statue of Liberty. They know exactly where the Statue of Liberty is down to geo coordinates and they could tell me exactly where that is so that I could find it. One of the main drivers for Google's visual search is Vision AI, and they use some of this data to connect this information with Google Maps. This is something that's really fascinating.

There's other tools that use visual search as well. So for instance, if my eight-year-old asked me what's so important about this car, why did I even take a picture of this car, then I can go into Pinbot on Pinterest and I can tell them exactly, "Who we're going to call?" I can say, "That is the Ghostbusters car." Then we can tell from Pinbot and they can understand that.

If I take a photo of what I think is an amazing selfie, and I tell Amazon that actually I would like more leopard print in my life, then they can help me to find more leopard print in my life using their search tool, Amazon Lens. If I meet someone at the airport and people don't believe me when I tell them that I met Xzibit at the airport, then I can use Bing's tools to show that, actually, yeah, that is Xzibit. Here's his knowledge panel. Promise you, that's him. I met him. He's great and we're besties and we're going to hang out.

Now, I'm not the only one who likes visual search. Lots of people like visual search. How many people like visual search? Well, At the beginning of this year, Google reported that there were eight billion people using Google Lens to search every single month. This is an increase of threefold since 2000, when it was said that it was around two million or two billion or so. So this is something that's actively growing.

Then one of the reasons why is because Google keeps adding more and more features to it. So now, you can not only use Google Lens to search images, but you can also use it to translate text. You can also take a photo of written text and it will take it and it will update it into written text. So if I scroll a bit of text down, then I can take a photo of it, Google will make it into actual text and then I can copy and paste that and put it into a text message or I can copy and paste that and put it into an article or something like that.

You can use it for homework. If you've got a formula and you're trying to do Pythagoras theorem and all that stuff, and you've got all this algebra and you're trying to figure it out, you can take a photo of it and Google will help you to figure out what you're doing there. You can use it for shopping. You can use it for identifying landmarks. You can use it for dining.

Then this GIF, you can see that you can also use it for multisearch, which is where you add in a modifier that helps you to see, to add more information to what you're searching. So let's say you want that shirt, but you want it in a different color, you can do that and you can also use that for near me if you take a picture of tacos or something to that effect.

Then over the last year, they've added more and more features. So previously, Google Lens was most accessible via mobile, but they've now made it much more accessible via desktop. So this is a feature. If you right click a photo, then you will get a pop out on the side, which allows you to search for Google Lens, and not only that, but when you click search there where you see that button that says search, you can pop it in, and again, you can use the modifier there.

So for instance, I took that photo of a chocolate cake or of a carrot cake and I added in chocolate and now I can see visually similar matches for that chocolate cake. I've seen other people who've used this to identify whether or not a stock image, for instance, was being used lots. So you can take a stock image on your website and you can pop it into Google Lens and you can see whether or not loads of people are using this and whether or not it's giving you an actual strategic edge by adding that into your website, and if it's not, you might be able to make some strategic decisions around what things you should choose for search.

Google has lots more plans for visual search. Here's an example of how they're using it for translating. So you can take a photo of a piece of text and then it will actively translate it and it will figure out with using AI, I saw that somebody's using AI in the chat to transcribe this meeting, but using AI, they can fill in the blanks so that you don't even see the difference between the text so you can translate that, which is something that's really, really useful if you're abroad, if you're trying to figure things out. I saw somebody who was using this online to translate Italian instructions for their English brain or English speaking brain on a washer, washing machine, which is something that was really useful for them when they were in their Airbnb, for instance.

Since I started talking about visual search, it's very interesting because Google has said the same thing. So they've said, "The age of visual searches is here." They said that in the four minutes into the search in 2022. They said, "The age visual search is here." They're adding it into lots and lots and lots of features. They are going really big on media of late and making the search more visual, more interactive, and more in keeping with the way that people are using search, are using phones, are accessing information so that they can keep in line with that and so that we can keep evolving as users.

You might say, "That's all great. That's fantastic. Good for Google. What do I actually do with any of that as a search marketer, as a business owner, as somebody who's trying to be visible on search, a content creator, whatever? How do I optimize for this brave new world with so many pictures in it?"

Well, let's start on some solid ground, right? Let's start on some well-trodden territory. So let's talk about how you optimize your images. Now, you want to lay your groundwork with image SEO. This is something that's been well-documented about how you do image SEO and there's lots of good tools and I'm sure you're all already doing image SEO across your websites, which is great, but you might be saying, "Crystal, image SEO is the same as visual search SEO," but it's not. Image search SEO is not the same as visual search SEO, and I'll tell you why because an image search SEO text queries return images, right? So you would enter the word fruit and you would expect to get a picture of fruit, but visual search SEO is about object recognition.

So what happens is the user image enters an image and they will expect to see content returned. It might be an image, it might be a text-based piece of content, it might be a video, it might be an immediate answer or a certain instant answer from Google, what have you, but it's going to be something where the image itself is the query, not a piece of text.

So for example, if I entered a query that said dinosaur with horns, this is the kind of response I would get from Google. Now, I'm trying to figure out a specific dinosaur with horns, but on this particular one, I'm getting lots and lots of different types of dinosaurs. Now, as an amateur paleontologist who is the mother of a small boy, I can tell you right now those are all different dinosaurs. Now, they're all part of the same dinosaur family, but they're different dinosaurs.

So if I actually want to figure out which dinosaur my kid gave me, one of his toy dinosaurs, I can take a picture of that dinosaur and then I can search for it on Google Lens. Not only does it understand that it's a toy and potentially give me options to see what it is, but it also gives me the exact dinosaur that I'm talking about. Why? Because on visual search, you're giving them lots more cues. I told you about some of the things that Google's able to discern when they look at an image.

From this image, they're able to discern the texture. Because of the similar images, they're able to see the color, they're able to see exactly how many horns I'm talking about. They're able to see the profile of it. They could find similar images. They can tell that it's quadrupedal. They can tell that it has a tail. They can tell lots of different things about this particular image without me typing all of that in, which I would never do. I would never say dinosaur with four legs and three horns and a frill and a beak, who is a vegetarian, and all of this. I wouldn't type all of that in. I just wouldn't. So an image search allows you to move a lot more quickly and allows them to serve better results for their users.

So when we think about image search SEO, it is absolutely part of optimizing for visual search SEO, but it is not all of it. So you want to make sure that you have all of these things already in place, relevant file names, modern formats, alt text, and image title structure data, all of that great stuff that will make sure that you've got the good foundation, but you don't want to stop there. You want to keep going after that.

Another thing you want to think about is making sure you have your text the right way around. So visual search things use optical character recognition and they work best when the text in the images is the right way around. So running it through visual search, this is a photo of a photo that I took with some colleagues at an event. You can see that the optical character recognition is finding it a bit difficult to understand exactly which characters those characters are.

When we flipped, I tested this with something else where I was using my Google Photos, which, again, uses a lot of the machine learning tools, which power visual search across Google, and I was looking for a photo of me and my friend, Bar. I knew that we had taken a photo in our Wix T-shirts. I looked up Wix in my photos and I couldn't find it. I could find it on some of the information, but I couldn't find the picture of me and my colleague.

So instead, I went through the dates and I found this picture. You can see in this photo that all of the words are the wrong way around. So it doesn't say Wix. It says Xic, Xiw. I'm not sure. So what I did was I went through and I flipped it around. Sorry. I flipped it around so that it said it the right way around. Then when I searched Wix, there it was. They can find it, they can see it. It all makes sense.

So when you think about your images, just make sure that your photos are updated with the text within the right direction and reduce reliance on vertical text and images. So if you're thinking about sponsorships and things like that, try not to get it so it says SEO. Try to make it show so it says SEO.

Also, you want to think about relevant entities. So if you're not sure what an entity is, essentially, it's a noun, but it's a distinct element within search. So this is something that is very often defined within using schema markup, and it's also something that you can see across like Wikipedia. So everything that has a Wikipedia page, for instance, is something that is an entity, and that's some very, very top level understanding of what an entity is, but it tends to be a person, place or thing.

Google understands over five billion entities across their knowledge graph and they use lots of different knowledge points or lots of different data points to understand those things. Within visual search, they can use these entities to identify objects. So within the Wix photo uploader, media uploader, when you upload an image, they use Vision AI to auto-generate image tags.

So for instance, I uploaded this picture of some daffodils, and you can see that it's generating tags based on some of those entity elements, such as flower, plant, petal, terrestrial plant, narcissus, et cetera. When you go into Google, you see the same things showing up within the knowledge graph. So you're seeing that daffodil. So when I searched this image, what it pulled up was this knowledge panel. So it says wild daffodil, narcissus, and that's the same tags, and then you've got a plant showing there, you've got a flowering plant, that sort of thing. See, so these all match up.

If you want to see what kind of images Google can discern from photos, then you can have a look in your Google Photos, like I said, my overloaded Google Photos, and you can see the kinds of images that Google can pull out. So for instance, you can see that they're pulling out things like skateboard, things like lighthouse, things like beer, things like trains, football, but they're also pulling out things that are a little, or ducks, they're also pulling things that are a little bit more abstract like sunsets or lakes or fog. Fog is something that's a little bit more abstract as well. So this is something to think about.

What I also find interesting about this particular set of information is that they're also pulling out Lego. Lego's very interesting because Lego is able to exist both as an entity, but also the word Lego is written on every little piece of Lego, which means that they've got a one-two punch when it comes to visual search, which is something that's very, very interesting.

So what I'm saying is, when you're thinking about the entities or when you're thinking about your images and how you optimize your images for visual search, make sure that you are including relevant entities to your images. So if for instance, this is a photo of me in Riverside, California next to a very famous hotel, which is essentially a landmark. So Google, if I search this photo on Google, they understand that that's a landmark. If I was a real estate agent, for instance, in Riverside, California, it would behoove me to make sure that I had some images of myself in well-recognized places in Riverside, California on my website so that Google could understand what was going on and could have more signals that my website was based in Riverside and that sort of thing. So think about the entities that are present in your images that are on your website. So make sure that you have relevant landmarks, objects, faces for sentiment within your images.

To sum up, when you're thinking about optimizing your images, make sure that you're relying on standard image SEO, readable texts, and relevant images because this will make sure that you've got lots of good activity there.

So how do you optimize real life? Okay? This is something that I think is really, really fun, and this is something that I think is a really exciting space for SEOs to be considering, but it's also something that brand people have been thinking about for a long time. So this is something where I think that if you're working in a larger organization, you can use this to help bridge the gap between some other parts of your organization.

So one of the things that's most important to consider about visual search is that the user is a big, big factor, and there's about eight billion users, potential users of visual search around the globe. So that's a lot of variables. If you want to think about how to control those variables in order to optimize for visual search, you want to dust off those guidelines, those brand guidelines, and you want to make sure that you have rock solid IRl, tight, inconsistent, real-life branding.

So let's go back to Xzibit. So there's a lot going on in this photo. I think I'm super cool in this pic. I was in my early 20s, and I was at the airport and I was like, "OMG! It's Xzibit. I have to get a picture with Xzibit." The reason why he's in this presentation is he's the only person in my camera roll who has a knowledge panel, but here we are. Okay.

So in this photo, I think I'm winning because I've got this great picture with Xzibit. Xzibit thinks he's the star of the show because he's super famous and people want to take pictures of him even when he's at the airport and all of that sort of stuff, but in this photo, what's really interesting is that in terms of visual search, the real winner, the real star of the show is actually Peet's Coffee. The reason why Peet's Coffee is winning in this particular image is because like Lego before, they are being recognized in this image on a few different points.

So what we see from here is that Peet's Coffee in this image has texts that's easily readable and that's high contrast and very clear for Vision AI to understand. What we also see is that their logo is being identified really clearly and with very high certainty. If you Google them, it's also really interesting because you don't get lots of images of really crafted latte art and lovely cups of tea and things like that. What you get is a masterclass in IRL branding, and it's phenomenal. Peet's coffee is not messing around. They have not come to play. They've made sure that they're super consistent across all of these things and they are being rewarded for it. You can see that their logo is consistent all the way across. You can see that they've got high contrast all the way through, and they're being rewarded for it on search.

So what I would recommend is that you be more like Peet. So what they've done here is they've got really tight consistent branding, but they've also made sure that they're using all of the same logos in real life that they're using in their structured data, that they're using in their knowledge panel, that they're using in Google Merchant, they're using in Google business profile. It's all the same. If you're from someone where this could be potentially significant, where you're potentially in a lot of B2C spaces, where you put your logo on a lot of stuff and somebody's saying, "Oh, we should mix things up. We should change things around," don't. Don't do that. Keep it simple, keep it consistent. Be like Peet. Make sure that you've got everything super tight and consistent and it's working really well for you, and don't forget about uniforms and merch. Case in point.

This is a team that I used to do a lot of work with. This is Optic Solutions. We had an event and we decided to get T-shirts or, sorry, we started to get sweatshirts. The sweatshirts were not in our traditional brand colors. The sweatshirts were in different colors. If you searched that sweatshirt and Google search, Google's like, "I don't know what this is, I don't know who these folks are. I don't know what's going on there." However, if I search for the print merch that we had that was in our normal brand colors, that was using the normal logo, that was using the same thing, they know exactly who it is. They know exactly who it is. You know why? Because Google's Vision AI, Google's AI visual search recognition thing, they also process information for colors. They also understand the colors.

When you think about your Google merchant profile, they ask you specifically for your color scheme. They ask you for the codes. They ask you for the colors that are associated with your brand, and that goes along with helping them to understand and to discern where your brand is and where your logo is when they're finding you on search, and this is something that can be really useful for you.

Also, think about SEO for real life sponsorships. So this is something that can apply to big businesses and small businesses. A lot of big businesses are going to be doing this on a bigger scale, but smaller businesses are doing this as well with sponsoring little league or sponsoring the fun run or sponsoring, I don't know, a bake sale for the Girl Scout team or Girl Scout Club, whatever, what have you, but pay attention to this because it's super useful because if anybody taking photos of it, it's something that can benefit you on search.

So a few years ago, wix.com sponsored the Yankees, which is great, and was fantastic. If you look at this, we've got some really good brand positioning here. We're behind the batter's box. All looks great, all looks fantastic. If you run it through visual search, you can see that we're doing really well. So we've got three logos here on the left. We're 97% certain that that's wix.com, and in the middle we've got 95% coming in at wix.com. You can see that where the batter's box is, where the batter is actually in the way, the certainty drops because they can't see the whole thing and they're not quite sure. So this is something that's really, really, really interesting.

The other thing that I think is interesting about this also with regards to visual search is that they're understanding the logo here. They're not just thinking about the text. The text is great, but the logo itself is processed as an entity, as the entity as a whole rather than just the words. So this is really, really, really interesting and it's something that you should think about when you're thinking about your sponsorships and making sure that your logo is something that is searchable and something that is discoverable.

The other thing that's really interesting about this is that because Wix sponsored the Yankees stadium, we're also seeing a lot of general PR in general above the line coverage across the web. What we see is we see that you start to get searches or you start to get images that show up across Google related to our sponsorship. Now, if you search Wix Yankees stadium, at the top you get the kinds of results that you would expect to see under the image tab. You would expect to see pictures of Wix Yankees stadium articles related to Wix Yankees stadium, but as you scroll further down, you start to see other things. You start to see articles where they're pulling up images under the tab Wix Yankees stadium that are accurate. They're reflecting Wix's sponsorship of the Yankees stadium, but none of these pages include text or text cues or all tags or HTML or anything like that that says anything about wix.com. None of them say anything in the text about wix.com.

What we're talking about here is visual search image recognition because all of these images have visual search recognition or all of these images have the wix.com logo in them. The one here of Giphy is very interesting because Giphy is absolutely tapped into the vision search AI algorithms or, sorry, vision search AI machine learning tools. So they have generated a category for wix.com based on that entity, based on their visual search recognition of that particular entity.

So whether it's the little league, the bowling league, major league, whatever you're doing, be SEO conscious of your sponsorships and think about placement as well. So this is a photo that I took of the Staple Center a few years back. You can see that when you run it through Vision AI, they're recognizing lots of different sponsors and lots of different elements, but they're recognizing all of these as text. The only thing that's being recognized as a logo is Kia.

One of the reasons why I think it's being recognized as a logo isn't just that it's the biggest or, well, it's very big, but it's not just that it's Kia, it's not that just that the logo is particularly good because there's a logo underneath the backboard as well, but I think it's also because it's very, very clear. So it's not cluttered up. There's a lot of negative space around that particular logo. So if you're getting sponsorship, make sure that you've got pride of place and you've got pride of positioning so that you're making sure that when people are taking photos of it, that people are able to find you on search and that it's able to benefit your brand and your brand entity.

You also want to think about making sure that you're avoiding having your visuals in high traffic areas. So for instance, on this event, which was a great event, they had sponsorships at the bottom, but you can see they're a bit obscured there. So when people are taking photos of that, they're not in the same way as the batter's box. You're not going to be getting as much visual search value out of those particular images as you could. So make sure that you're creating areas for people to take photos and that they're very, very clear. So when you think about that, make sure you're on plain backgrounds with good lighting and prioritize unobscured placements.

The other thing I also say is worth thinking about with regards to visual search is thinking about 100% that pic. When I say that, I mean those photos that people take that you know everybody takes. Everybody takes that photo. People come back from Paris and they're like, "I went to Paris and I saw the Eiffel Tower," and there's a photo of them on that place and they've got the Eiffel Tower on their hand or whatever it may be. There's loads of these places where everybody gets the same photo, and I'm no different. Here's me at the space shuttle endeavor in Los Angeles at the California Science Center. I love this spaceship. It's my favorite spaceship. I don't know why I have a favorite spaceship, but anyway, I went there and I was like, "Oh, my God! It's a spaceship. I need a selfie," and that's what I did. I made sure I got myself a selfie.

It's super easy to search to find out which spaceship this is because the California Science Center has made sure that they've put that in their knowledge graph. So it's in their knowledge panel and you can see it. They know that people are looking for the space shuttle and they know that everybody walks in and they get the same photo of the space shuttle and they've put it on their website. They put it on their website. It's the same photo. They know that that's what people do. They come in and they take a photo of the space shuttle. So that means that every time anybody searches for them, they're like, "What was Crystal doing? Where did she see a space shuttle?" They search that picture and they can find this one, they can find this place immediately because they made sure that they put that picture on their website.

You might say, "What if I don't have a spaceship? What if I don't have a spaceship? I'm a little old and nobody gave me a spaceship. What should I do?" I would say, if you don't have that pic, make one. It doesn't have to cost the earth. It doesn't have to be this big. It doesn't have to be a literal spaceship. You don't eat federal funding to make one, but I would say that it's worth doing. This is a small business. They do nails and they do hair and it's super girly and they spend a lot of time making their shop really, really Instagrammable and they do hair and nails and all that stuff and it's super, super fun.

If you drop that pic into Google search and they're a small business, if you drop that pic into Google search, Google image search, then they show up and Google knows exactly where they are because they made sure that they put that into their Google business profile. They made sure that that picture is really, really present and it's easy to search and it's easy for Google to find.

So make sure that you have that picture on your website, and if you're not sure what that is, if you're a bigger business that might have a knowledge graph entry, then go and find out what image is on your knowledge graph and make sure that you're reflecting that on your site. If you're not sure what it is, then you can go into user-generated content sources to look for candidates. So Facebook, Instagram, TripAdvisor, Yelp, you can see what pictures people are taking. It might be that you have a big statue outside of your house. I know that Bob's Big Boy used to have the guy with the burger, for instance, outside the door. If you've got something like that, there's probably lots of people taking that photo. Make sure you have that photo on your website.

You can also think about mobile brand representatives. So for instance, if you're talking about the Red Arrows or you're thinking about somebody who's a mascot, if you have a van, that's something that goes around a lot. If you think about the UPS, the UPS truck, that is an iconic vehicle, and that's something that people might be taking pictures of regularly and it's got your logo on the site, make sure that that has a place on your website.

So I would also say that it's worth encouraging user-generated content. Now, I get a lot of grills. I've been grilled for this, but I'm a level six Google guide. I know people get jealous about this and stuff, and I'm a level six Google guide. It's okay. You can get there too. I have the socks to prove it. It's great. The thing that's great about being a level six Google guide is I get to tell the world all about my burritos, which is fantastic.

When I do this, I do this not just for fortune and fame and glory, but I also do this for points. They give me lots of points and I like those little gold stars and it's great. I'm not the only one who's a Google guide. Google guides are generating over 20 million contributions every day on Google Maps because everybody wants to share their burritos and everybody wants really cool socks. Google uses these pieces of content as datasets for their search.

So for instance, if you're searching near me using multisearch, then Google will be referencing photos that they've received from Google guides on Google Maps to find similar information to that when people are searching for food, for instance. So for instance, I dig my burrito and I do it near me, then Google will go through all of the images that they have of burritos that have been submitted to them on Google Maps by Google guides, and they will try to return a result for me.

So this means that it's worth thinking about your Google business profile as a visual search dataset as an opportunity for people to find you on visual search. So if there are things that are important for you that you want to show on visual search, make sure you include those in your Google business profile photo collection so that Google can find you. Also think about how you manage that. So you want to entice Google guides. Google guides, they're very good at taking photos, labeling them, following up because you get prompted to label them, add them, and they know how to optimize the images because they get more points for everything they optimize.

So if you want to entice Google guides, think about adding good owner photos because Google guides will get points to say, "Oh, your phone's been seen by lots of people," and that makes Google guides happy. So make sure that you are enticing them with good owner photos. Make sure that you're liking and responding to top photos. Make sure that you're applying to users and display your reviews on your website. It is something that I've seen a lot where basically if you want to get more reviews on your website, a really good way to do that is to show that people are reviewing you on your website.

There's a couple of really good widgets that you can get that you can just add and it'll show that like, "Oh, we've got 105 reviews on Google business profile and we have seven out of five stars, 4.5 stars overall." Then also remember to prompt users at location. So you can get little stickers on the door that say, "Review us, leave a photo, take a photo," that sort of thing. There's a few great places for this. Obviously, there's Google business Profile, but others are available. So TripAdvisor, Yelp, Instagram, and Facebook can all be sources for some of these images, and make sure that you're thinking about it strategically.

So here's an example of one that I used, for instance, where I took a photo of some food, and this is an example of how you can use those near me searches. You can see some of the ways that Google is sending that information back. So think about your user-generated content and how you can make sure that you have content that reflects the kinds of searches that people are going to be making. So connect with Google guides. Encourage users of visual search reviews across user-generated channels including Google search, but also thinking about things like TripAdvisor and other places as well, and making sure that you're handling that strategically.

So to sum up, visual search is growing rapidly, but don't forget that the photo is the query. Okay? So it is about image recognition. It's about object recognition. So make sure that that's very clear for users. Also, optimize images with standard image SEO, with readable text in your photos and also with relevant entities. Then you can think about optimizing GIRL with really tight and consistent branding, making sure you're encouraging the creation of user-generated photo opportunities.

You know when they say, "This is a scenic viewpoint," if you know that you have a scenic viewpoint in your place and make sure that you also have that reflected on your website, look for SEO opportunities and IRL sponsorships. Again, this doesn't necessarily need to be an every single day activity, but if you're an SEO and you're working with a wider team, make sure that you're also thinking about maybe adding that into your general brand guidelines and then maybe including it as a checklist. Also, think about user-generated content for Google business profile. If you do all that, you should be ready to roll. Thank you very much and let's have some questions.

Travis:

Awesome. Great job. That was very, very awesome. Also, I never really thought about the importance of text directional images, but it makes a lot of sense that machines would read very similarly to us. Yeah, we have a couple questions. I think the first one would be, "You mentioned including the relevant landmarks on your website. Do these images need to be taken by you, the business owner or can you leverage existing images?"

Crystal:

So yeah, I think if you're not taking the images yourself, then I think it can be less advantageous for you from a search point of view because it's not unique. So with search, generally speaking, things that are unique tend to perform more distinctly. You're essentially trying to make sure that Google knows that you're unique, that you are a source of information that they can't get somewhere else, and so must come to your website. So the more unique content that you have on your website, the more of a destination you are. If they see that lots of people have the same image across lots of different websites, then it's more likely that they will send it to somebody else. Whereas if you have something that's your own, then it's more likely that they'll send them to you if it's of value.

Also, when it's your own image, then you can tailor it so that it meets your needs. Obviously, let's say if you would have a picture of the Eiffel Tower or something like that, let's say you were a French patisserie based in Omaha, right? So you're selling lovely French pastries, but you're not in France. I think it's fine if you're using a picture of the Eiffel Tower just to give a symbolic reference even if it's not your photo, for instance, but if you can, use unique images. I think that's more advantageous.

Travis:

Nice. Then James sent in a question, "I'm a professional photographer that shoots product reviews for a digital marketing company. Besides the image SEO basics and the pictures are presented clearly, what else should I be thinking about since I have access to high quality equipment?"

Crystal:

So I think composition matters. So composition matters a lot with regards to visual search recognition. I think that one of the things that's tricky for artists with regards to visual search is that visual search rewards very minimalist, less busy photos. So I have an article that's on Moz that talks and it's like a thing about my teapot, right? So I have my teapot and I have it where it's ... Oh, actually, here's a good example. So I have a teapot where you can see the handle like this, and then I have a teapot where you can see the handle like this. When I have it like this, Google can recognize that it's a teapot, and when I have it like this, they're not sure, right?

So I think that when you're making your images, and particularly if there's commercial intent behind them, then make sure that you're thinking about the composition so that they're clearly identifiable. If you're not sure, then you can test them using image recognition software. So Vision AI is a tool from Google, and you can drop in a few images and you can see them. Also, if you're using Google Photos, you can search your photos. So you can upload 10 photos, 10 similar photos, and you can search the photos. Let's say you were taking photos of guitars. You could upload all of the photos that you took of the guitars. If you search for guitars, then you can find, and if you see that only six of the photos are recognized as guitars, then maybe those are the six ones that you should use and maybe the other ones shouldn't be used. If you see that none of them are recognized as guitars, then that's tricky.

So you want to make sure you've got good lighting, good composition, and I would say test your images. There's a few tools that you can use. So Vision AI, it's an API and you can use that to tag things at scale. There's lots of other tools that will allow you to tag images at scale, and those tags will tell you whether or not Google or other machine learning tools can recognize those images, which is really useful. So think about composition, and if you change your composition, it will change how Google understands your images.

Travis:

Nice, nice. We got another question. "Since Google recognizes the content of an image to some degree, should the image description, alt text go beyond a straightforward description of the image?"

Crystal:

So I would say that the alt text should probably be around the core elements of the image. So for instance, if it's a guitar, for instance, it might be a Fender Stratocaster guitar with amp or something. I feel like that's the core element of the image. They will be able to discern from their images that it's a blue Fender Stratocaster guitar with an amp. It might be that you can think about others. You can also add descriptions. You can also add texts. All those text cues are still valid, but Google's able to add in some of those other elements.

The other thing you might want to think about with regards to image recognition, and going back to the other one, is logos, visibility of logos. It's a bit tacky. Don't get me wrong, it's a bit tacky, but Google can recognize logos. A friend sent me a photo of their lunch and I was like, "Oh, I'm going to run this through Google Lens," and I didn't even realize, but the top of the plate had a tiny little logo for the restaurant. So I found out exactly where she had her lunch because of the tiny little logo on the plate at the top. So think about that as well when you're thinking about your images.

I think also within schema markup for your images, I think that that's where you can really go to town on some of the attributes. So for an image with schema markup, you can write in lots of descriptions. You can add in keywords if you want to. You can add in lots of different elements to give information about the image. So I think that that is something where you can add more text if you really want to go down the route.

Travis:

Nice. That answers Eric's question. He was asking, "Does it help image recognition if you add keywords to the alt text?"

Crystal:

Yeah. The alt text should be relevant. I think that it should be relevant. Again, if we're going back to the guitar one, it shouldn't just be rocking out. The text shouldn't be rocking out or it shouldn't be photo 375. That's not useful. So just make sure that it's relevant and has the core terms. Google's able to discern a lot across a photo and stuff. So I think that I should make sure it's relevant and has the core terms.

Let's say this photo of the guitar and the amp, let's say the article is about the amp and not about the guitar. Put priority on the amp, right? So you would say, I don't know, Gibson amplifier with a Fender Stratocaster guitar rather than saying Fender Stratocaster guitar with a Gibson amplifier. You would say the first one. Think about how it works with the article holistically and put priority on the thing that's most relevant to the article overall.

Travis:

Cool. This might be the last question, but are there any fun strategies that you've seen other companies use, maybe like the B2B space, how they use image search for brand discovery at Wix or other companies?

Crystal:

Yeah. So for Wix users, as I say, we have Vision AI built into Wix, which was one of my favorite discoveries when I first joined the team. I was like, "Are you kidding me?" Yeah, you upload a picture and it auto generates tags. The other thing that's really great is when you have that picture in Wix. So I upload all my pictures and I don't always pay attention to the tags and stuff until I'm ready to use it. So sometimes I'll just upload them all and then I'll sort it out. Now, people, that's bad practice. That's lazy, but on my personal site, when I'm doing stuff, I'll just upload it and then I'll sort out the tags later, but what you can do is in that case is I can just search in my photos that are in my media library on my Wix site, I can just search for daffodil and it'll pull up the picture of a daffodil. I can search for smiling and it'll pull up a picture of people smiling across that, which saves a lot of time and it's really, really valuable.

There's also tools that allow you to add visual search to your retail experience. So I showed Amazon Lens, which allows you to ... So let's say I go out somewhere and I'm like, "Oh, my God, I love that jacket," and my friend's like, "I don't know. My husband got it for me. It was a birthday present. I don't know where it came from." I could take a photo of her jacket and I could search it on Amazon Lens to find where she got it from. So Amazon Lens had built that into their tool. Pinterest has a builtin to theirs as well. I think eBay has an image search tool as well. There are lots of businesses who specialize in allowing you to add similar capability to a business that's not Amazon, obviously.

Travis:

That's awesome. That's so helpful. Well, cool. Thank you again for joining, Crystal. That was a fun and insightful presentation. Everybody learned a lot. Everybody, definitely give Crystal a shout out on Twitter. Thank you for spending time today. Crystal, is there anything you wanted to share before we give everyone their day back?

Crystal:

I would just like to say please check us out on the Wix SEO Learning Hub. We share lots of really interesting insights and lots of different ways. There's some image search optimization tools there. Yeah, check them all out.


Written by
Bernard Huang
Co-founder of Clearscope
©2024 Mushi Labs. All rights reserved.
Terms of service, Privacy policy