People do not always search Google through the regular search results. An increasing number of searches occur in Google Images, thanks to the wealth of pictures Google offers.
For business owners it is a great opportunity to get more of their content ranking well in Google. The question is: how do I rank a picture in the top 2 lines (for instance)?
Enter a search query in Google Images (images.google.com), hover your mouse over a pictures and look at how Google describes an image.
- Size and Description
We have to ask ourselves:
- Are these factors predominant in the way Google ranks the pictures?
- What are their respective influence in the ranking algorithm?
- Are there common factors between these pictures?
- What other (not so easily visible) factors influence the rankings?
We observed the image results for some time, and found some answers which we could use to improve the ranking of our clients’ pictures. Though we can’t pretend to have deciphered Google’s ranking algorithm for pictures, we’ll share what we found and this may help you conduct tests of your own.
Bottom line: since Google Images has become a method Googlers use to search Google and find products, it behooves on us to help Googlers find our products in this way too.
Filename, URL and description
Jimbo’s Sidecars is a LocalRanker user since 2008 [a very early adopter]. Jimbo’s specialty is renovating and converting vintage Chinese sidecars into beautiful BMW sidecars. His shop is situated in Beijing, and Jimbo ships sidecars the world over. His reputation has gone completely international.
In Jimbo’s line of business, pictures are highly important as you won’t order a $10,000 sidecar from a website and send money to China before yau have solid proof that the guy is legit and that what you buy is what you gonna get.
Jimbo’s website benefits from many back-links as each of his bikes is a remarkable piece of work, and Jimbo has a following in the sidecar community. Unsurprisingly his website has been ranking #1 since mid-2009 on his main keywords.
Looking at the first row of pictures offered in images.google.com on the keyphrase “BMW sidecar” we can see 2 pictures from Jimbo’s website sitting pretty in positions 1 and 3. We looked at each of the six pictures in the first row, and tried to identify similarities in the filenames, the URLs and the descriptions.
This first image is from Jimbo’s website. The URL is a perfect match with the key phrase. The filename does not contain any of the search keywords. The description contains the word “BMW”.
The words “BMW sidecar” appear in the filename – perfect match. The URL does not feature any of the keywords. The word BMW is featured in the description offered by Google.
The URL is a perfect match. The filename only has a vague relationship with the words “BMW sidecar”. The word BMW is featured in the description selected by Google.
The filename offers a pefect match. The URL is irrelevant. The description excerpted by Google offers both words “BMW” and “sidecar”.
Note that Google does not care that the words are separated. It identifies them separately as relevant to the search query. Keep this in mind when optimizing your anchor texts: you can throw some words ‘in between’ to add diversity into your link text.
In this photo the URL and the filename are irrelevant, but the description excerpted by Google comprises both keywords.
Again, the keywords are separated. The key phrase “BMW sidecar” is not analyzes as a semantic unit (noun + modifier) by Google but as 2 distinct and separate keywords.
In this picture, the filename features an exact match. The “-” is not taken into consideration by Google. The URL is irrelevant.
The description excerpted by Google features the 2 keywords. There again, separated by a “-“, which confirm that each keywords is analyzed on its own for matching purposes with the search query.
It may also mean that “-” are irrelevant in Google’s eyes. But this is not a certainty and we will see later that the absence of separator (like “-” or “_”) can prevent Google from identifying a keyword as a possible match.
From these 6 images featured on the first result line in Google Images, we see that the presence of keywords in the filename, the URL and/or the description plays a role in the degree of relevance of the picture in Google’s eyes.
This observation has been confirmed many times, and our team always name picture files using keywords. The use of keywords in the URL is of course a given [though there are still health practitioners who believe “DrJohnSmith.com” is a good URL… even if they are not the best-known brain surgeons in the village.]
The next question is: Where does the description come from?
Google’s preferred sources for picture data
In the distant past of SEO (a few light years ago), we used to “stuff” the Alt tags of the pictures with keywords. Alt tags substitute for pictures when the pictures don’t load: at least visitors would read a description of what should be there – instead of just looking at a gaping hole. [Which leaves me wondering what excuse did we serve our clients when a picture didn’t load in the browser? Heck if I remember….]
Keywords stuffing has long been pushing daisies, and though dutiful web designers still describe their pictures with ‘alt’ tags, Google just doesn’t care anymore.
Google has long switched for contextual description, and snips the description of the picture displayed in Google Images directly from the copy of the website. OK, but where exactly from?
For our first test to identify where Google selects the words describing an image, we entered “realtor in tucson az” as our search query in images.google.com. This picture is the first one served by Google.
Note that the filename andthe URL of the image, and its description all feature the keyword “Tucson”. The description specifically features the words “Tucson Arizona real estate“.
Consistent with previous findings, Google rates this image very relevant to the search query “realtor in tucson az“.
Google identifies “realtor” with “real estate” and “az” with “arizona” which confirms that certain keywords are fully interchangeable in Google’s database. Work on your text links with this in mind.
We then went to the page where Google found the picture.
Note where the text of the description shows: above the picture, immediately next to it. Note also the presence of the keywords (yellow circles) around the picture.
The second image to appear in images.google.com under the same search query is found by Google on the website Realtor.com.
There again, consistent with our previous findings, the URL and the description of the picture feature the keywords of our search query.
We have seen that “Realtor” was interchangeable with “real estate“, just as “Arizona” and “az”. “tucson arizona” match our search query.
Note that the “,” between “tucson” and “arizona” has no impact on Google. Note also that “real estate” would usually be writen “real-estate” but this does not prevent Google to identify the word with “realtor“. The “-” has no bearing on the search results. Lastly, note that our query featured the word “in” and that Google ignores it completely.
Where did Google find these words “Find Homes For Sales in Tucson, Arizona”?
Here is the page on which Google found the picture. Note the position of the words: above the picture, close to it. Note also the presence of keywords in the proximity of this picture (yellow circles).
The third picture in the results of images.google.com shows the same pattern as in our previous tests. The keywords match closely our search query.
The same pattern repeats itself. The keywords of our search query are detected as a match by Google in this picture.
The next screenshot shows that the description associated by Google to this picture was also at the proximity of the picture. The site shows it on the left of the picture: in terms of coding, “left” of the picture and “above” the picture are the same. The text could have been placed visually above the picture, this would not have affected the way Google looked at it.
Again, note the presence of keywords in the description and around the picture.
For these 3 tests, we can see that Google likes the keywords to be placed above and around the picture, very close to it. We also see that if we want Google to describe our picture the way we intend it (rather than leaving it to Google to select some random sentence in our page to describe the picture), we should place our descriptive text above the picture and close to it.
Is it important to describe the picture the way we intend?
Place yourself in the shoes of a Googler. You see rows of pictures in a page, and some of them seem more interesting than others. Before you click on any of them though, it is likely that you will hover your mouse over those more interesting, and read the caption. We have been trained since an early age to read words associated with pictures: in vocabulary books first, in school textbooks then, later on in advertisements, then in newspapers.
A quick glance at the caption will tell you if you should continue and click on the picture to go to the website and see in a bigger size. If the caption is interesting and relevant, the image wil win your click more often than not. You may already have decided to click on the image to follow it to the website, but the caption will comfort you in your decision.
Direct marketers know this. Newspaper editors know this. You will be looking for the captions of the photos shown to you. You will want to know the context of the photos. You will want to get the idea of the story told by the picture BEFORE you read the real story.
This is the reason why newspapers like USA Today put half the picture in the upper part of the first page (“above the fold“) and the rest of the picture and its caption in the lower half of the page. When you see the newspaper on the stand, and spot half the picture at the center of the upper part of the page, you tend to pick up the paper and unfold it to get the full picture and read its caption. At this point, the paper is in your hands, not any more on the rack. Does that make you more likely to buy it? You bet it does.
In the third test example, the picture picked up by Google is actually the second picture on the page, NOT the first.
Check out the first picture found in the body of the page: note the presence of keywords next to the picture. All seems good. So why did Google shun this picture and selected the second one in the page?
Look at the picture in Test #1. This is the exact same picture. This is a perfect example of a very important Google policy: duplicate content is not welcome.
“Duplicate content” is text, photos, videos that Google already found on the web and already indexed in its database. Google is not interested in serving twice the same content to Googlers. Google has become sufficiently intelligent to recognize content that only duplicates, even with alterations, similar content already in existence.
The same goes for photos. Photos can be identified by color, color variations, size, copyright mentions embedded (watermarked) into the file. Here, Google has identified that this photo presented exactly the same characteristics as the photo used in Test #1. And even if the keywords describing the photo and placing it in context were relevant, Google just did not select it.
Bottom line: use your own photos… Or be shunned.
I will soon publish a Part II to this article. Meanwhile, get to work to optimize your site.
To your success!