The Definitive Guide to Image Search Optimisation

Last week Jeroen van Eck and Bill Slawski lamented the lack of a solid guide to image search optimisation. Bill threw down the gauntlet, and I picked it up. Here’s my Definitive Guide to Image Search Optimisation.

Image SearchFirst, what do we mean with image search? Simply put it’s the vertical search option that allows you to search for images on the web. Google has an image search, and so does Bing, and there are several other specialised image search engines around.

For the purpose of this blog post we’ll focus on Google’s image search, though I reckon most of what follows can also be applied to other image search engines.

SEO for Images

Why is image optimisation important? Well, search engines can’t see images. Now this is not entirely accurate – search engines are getting darn clever indeed – but it’s a good starting point to keep in mind when you want to optimise the images on your website.

So, as search engines can’t actually see images as we do, we need to tell search engines what those images show. This involves a wide range of on-page factors:

First and foremost your image filename is a strong indicator of what your image is showing. An image called red-shirt.jpg is probably an image of a red shirt. So make sure your filenames are properly written and contain words relevant to what your image is showing.

There’s some dispute about whether alt-texts are factored in or not, but personally I’d err on the side of caution The alt-attribute of an image is a strong indicator of an image’s topic, so you want to make sure your image’s alt-attribute is well-written and contains relevant keywords for your image.

However, for non-vital images on your web page (bullet point images for example) a blank alt-text is better. That way search engines don’t get confused about which images are important for the content of the page, and which images exist purely as support for the page’s formatting.

A caption is the text accompanying the image on a web page. This caption is a strong indicator of what an image is showing. One look at a news site and how they caption their images will tell you that captions nearly always describe what an image is showing. After the filename the caption is probably the most important single factor for determining an image’s relevance for a keyword.

Ensuring your caption is present within the same HTML container as the image (as per the example below where the caption is part of the same paragraph tag as the image) enables search engines to associate the caption and the image.

Alignment of filename, alt-text and caption
The filename, alt-text and caption of an image of a website combine to form an impression of what the image is showing. For maximum impact these three factors should be aligned, containing the same or similar keywords. The three factors can be seen as descriptions of the image, increasing in length:

  1. The filename is a very brief description of the image’s content
  2. The alt-text is a slightly longer description
  3. The caption is the full description

Optimised Image

Page Content
Search engines look beyond just the image, they also look at the page the image is embedded on. An article about race cars is likely to have accompanying images showing race cars. This means the page’s content needs to align with the image. Often with Google Image search you’ll find that the headline of a page where an image resides contains the keyword you searched for.

File Format
Your image should be in one of the three major image file formats: PNG, JPEG or GIF. Personally I’d recommend using JPG as this has additional benefits in other verticals (Google News for example), but don’t hesitate to use PNG or GIF if you feel it’s better for your images.

Image Dimensions
While on Google image search you can find images in a wide range of dimensions, if you want to break through with universal search it’s recommended to keep your image dimensions within certain parameters. An image size of at least 250×300 seems to be the accepted norm – bigger is OK, smaller isn’t.

File size
Don’t make your images too large in terms of file size. With JPEG this can sometimes be a challenge when too much compression can ruin an image’s quality. But image size is quite important, not in the least for your site’s load speed which is a ranking factor in web search.

Anchor Text
When you link to an image on your site (a large version for example) the anchor text of that link is a factor. We know search engines love links and pay close attention to anchor texts, so make sure you’ve optimised your anchor text to be relevant and descriptive.

Do note though that, if you run a news site, you shouldn’t have the images that accompany your content link to anything. Linked images do not show up as thumbnails for your news articles in Google News, as it prefers unlinked images in news items. Read more about SEO for Google News here.

Title text
An image’s title attribute is the text that shows up when you mouse-over an image. Or at least it should be – some browsers handle titles differently. The title text doesn’t convey any benefit for image search, so it’s up to you if you want to use it for anything. For the sake of clean and lean code you might want to omit it entirely.

Image Sitemap
To increase the changes all your images will be properly indexed, you can add them to your site’s XML sitemap or create a separate image sitemap for them if you want to keep things organised. The only data that is required is the image URL, but you can also add the caption, title, and geo-location of the image (where it was taken) to your sitemap.

Read Google’s guidelines on images in your XML sitemap here. I recommend using the image’s alt-text as the title in your sitemap, and to keep the caption the same as on your site.

Practice Moderation
Now with so many factors to keep in mind it’s very easy to stuff keywords in everything and hope for the best. Needless to say, this is not desirable. So practice moderation and don’t stuff keywords everywhere. Relevance is key and you want to make sure the whole package, from image filename to page content, conveys that relevance to search engines.

Advanced Factors

Now at this point I feel obliged to point out Bill Slawski’s analysis of a 2006 Microsoft patent regarding the ranking of images for a search query: How Do Images Get Ranked in Image Search?

What stands out here is the possibility of search engines going beyond the on-page factors and looking at things such as link relations between images, the number of times an image is used on a site, number of websites that use an similar image, and click-through behaviour of users on image search SERPs.

I highly recommend Bill’s post and the associated reading material, but I think you shouldn’t worry about it too much. Make sure you get your on-page image optimisation factors right, and you’ll be there most of the way already. And having pictures unique to your site doesn’t hurt either.

The Future
There’s some talk about the use of the EXIF image file format for SEO purposes. EXIF is an extension to existing file formats such as JPEG and TIFF and it’s already being used by many digital cameras. Basically with EXIF you can embed metadata about your image in the actual image file itself. This metadata can then be read by programmes such as image sharing sites and, yes, search engines.

While this metadata is mostly technical information about the image such as details about the camera it was taken width, specific camera settings, and compression and resolution info, there’s a chance in the future other SEO-relevant information can be included such as description text. Geo-location info is already part of EXIF, and this could be used by image search engines to connect images to geographic locations.

Right now I wouldn’t worry too much about EXIF, but if you’re serious about image SEO it’s a good idea to keep an eye on it.

To wrap things up I’ll leave you with some words of wisdom from the big G itself: the Google Webmaster Central page on image search.

[Update 31/8: input from Bill Slawski and Bob Gladstein added in the post]
[Update 03/9: info about image sitemap added]
[Update 09/9: EXIF details added]
[Update 28/9: linked image info for Google News added ]

About Barry Adams

Barry Adams is one of the chief editors of State of Digital and is an award-winning SEO consultant delivering specialised technical SEO services to clients worldwide.

30 thoughts on “The Definitive Guide to Image Search Optimisation

  1. @Ani the problem with those traffic figures is that it doesn’t catch the traffic you get from universal SERPs on Google web search – which can be quite substantial. This traffic will show up as plain Google web search traffic and not as image search traffic, which can skew the percentages.

    But yes image optimisation should not be a high priority, unless you run a site thriving on images, such as a photosales site.

  2. As far as I can tell, an image’s title attribute isn’t indexed by the search engines, and I find that as a usability factor, it’s usually better to leave it out. You’ve already got the text surrounding the image explaining its context. For people who can see the image, it’s usually self-explanatory, and for people on screen readers, there’s the alt attribute (and for some, the file name).

    In the example you used, a title attribute of “Image Search Optimisation” doesn’t really give the user any useful information that they’re not already getting from the caption, “The Definitive Guide to Image Search Optimization”. A CMS like WordPress defaults to giving images identical alt and title attributes, and I normally delete the title.

    I personally only use the image’s title attribute for images that anchor links and may require some further information: that the target document is a PDF, it’s in another language. I’d add a similar title attribute to a text link if I felt it was needed.

    I’m not suggesting that it’s somehow wrong to include the attribute. It’s certainly not spam, and if you tried to spam it by stuffing it with keywords it would just be ignored anyway. I just don’t think that it adds anything of value most of the time.

  3. Hi Barry,
    great article, thanks.
    I have only one small detail to add : it could become usefull to include images in the XML-sitemap. I think today it isn’t important for the ranking but for the speed of indexation (may be). May be the sitemap-mention could be more importent for the ranking some days…
    Perhaps in some cases it could be usefull to make “unusual” pictures. While checking the key “mona Lisa” during the last two years I recognized that Google more and more prefers “unusual versions” like cartoons or fun pics. Could be because of higher clickrate, could be because of a kind of “ranking-mixture”: different versions of a picture may be more interesting for users than 20 nearly identical images on the first 20 positions. And as we see in the “Similar images feature” google is able to find the degree of consensus.
    Best wishes,

  4. @Bob I agree that the title text has little value, but I like seeing mouse-over texts on images, and I use Firefox – hence why I recommend using it. 🙂 But maybe I’m funny that way. There’s sure no harm in omitting the title attribute.

  5. Hi Barry,

    I think you’re off to a pretty good start here.

    Regarding alt text, one of the problems I see on many sites is that they try to choose an image for a page first, and then try to come up with alt text that’s often less about the image itself and more about the keywords chosen for a page. And then they try to stuff fairly meaningless images, such as decorations or bullets with keywords as well.

    I’ve also seen a page ranking at 14th in Google web search for the term “blue arrow” on the sole basis of alt text for a handful of blue arrow link decorations (it had a pretty high PageRank). I usually recommend that people use blank alt text for images that have no meaningful connection to the actual content of a page, such as bullets or decorations.

    And it’s often not a bad idea to develop an idea or concept of an image that accompanies a page’s content before actually creating or finding the image itself if possible, so that the image adds meaning to what you present on a page, and so that an appropriate alt description naturally includes the keywords you’ve selected for the page.

    Regarding captions, your example shows something that I consider a very good practice, but you didn’t spell it out in your post. I like to include a caption and the image in the same HTML container, so that a search engine can get a good idea that the caption and the picture are related. You’ve done that with a paragraph element in your example.

    There’s a chance that a search engine will look at other text from a page to associate with an image as well, including the page title, and possible text within a certain distance from the image itself, but you make it more likely that the caption will be considered by using that HTML container.

    The MIcrosoft patent has a number of good ideas of how images might be ranked higher in image search results based upon on page factors, site-based factors, and off-page factors such as links to the images, but there are also a few other Microsoft patents worth exploring.

    One that is somewhat similar to the one in my post that you linked to discusses how including an image on a page can be helpful in the ranking of that page. It may sound kind of obvious on its face, but I think that’s important to keep in mind when creating a page or a blog post – the image you use can help the page it appears upon rank better in search results. Many of the features Microsoft describes in that patent are similar to the ones described in my post above.

    Another Microsoft patent describes how they may choose a “dominant” image on a page or a segment of a page, which could be a concern when you either have a page with multiple sections involving multiple topics, or you have a number of images and the search engine wants to choose between them to possible display in search results (something Microsoft often does with news stories).

    From what I’ve experienced, I think Bob is pretty on point with the use of title attributes for images and links.

  6. @Bill: thanks for your input, it’s much appreciated! I updated the post to cover some of your comments as well as Bob’s.

    @ Martin: personally I’m not sure including images in an XML sitemap is a good idea. Search engines might index your images faster, but they’d do so without the proper context – i.e. the caption, page content, etc. All a search engine would see would be the actual image file, devoid of any on-page contextual factors. I don’t think that’s necessarily a good idea.

  7. Thanks Barry, good detail.

    For the filename of the image, I thought it best to divide the words up with a dash or hyphen (e.g. yellow-tennis-balls.jpg), instead of having them as one word (e.g. yellowtennisballs.jpg) or divided with a underscore (e.g. yellow_tennis_balls.jpg).

    What is your preference?

  8. Well, that’s comprehensive! It’s going to be time consuming on large, graphic heavy sites (online shops, etc), but the payoff may well be worth it.

    I wonder if anyone’s checked to see if visitors landing on a site from an image search actually convert?

  9. @Andy In my experience they hardly convert. I’m talking about traffic from Google Image Search (not universal) in the old interface here. Most people searching in Google Images suprisingly are searching for… images. Unless your goal is to sell stock images converting will be very unlikely for these visitors. Still, this doesn’t consider images in the regular SERPs and is based on a few specific websites.

  10. What about pointing search engine spiders to your image locations through an XML SITEMAP? Isn’t that the key to all of this? I think it is absolutely critical and an easy win for those wanting their images indexed.

    In short Google is indexing enormous amounts of content. Everything you do to make its life easier will benefit your website in the long run.

    For those interested in getting their images into a search index, please have a look at the following:

  11. @Uttam yes you’re right, the sitemap was an omission as has been pointed out to me by several folks. 🙂 I’ve added a paragraph on image sitemaps to the post.

  12. Great article Barry – thanks.

    I have one bullet you could think of adding.
    I have a couple of times experience that having pictures on the CDN actually ended in the pictures not been index at all.
    So the hosting does matter as well I pretty sure that loadtime just as onsite must have an effect

    Best from DK

  13. Lovely read, thanks!

    In my experience unique images are very important also, especially in Google News. Imagine 10 sites/blogs writing about Apple’s quarterly results. All using Apple’s wonderful logo. The bloggers/reporters all used Google’s Image Search to find it.

    If you, as tech blogger, use Bing’s Image Search you’ll most likely get a different variant of the logo thus making it unique. Combine it with a great descriptive annotation and et voilà!,or.r_gc.r_pw.&biw=1236&bih=722&um=1&ie=UTF-8&tbm=isch&source=og&sa=N&tab=wi

    Also, I imagine hotlinking could be compared to building backlinks, Has there ever been any tests on this?

  14. Very good written story. It will be supportive to anyone who employees it, including myself. Keep up the good work – can’t wait to read more posts.

Comments are closed.