“When you embed text in an image, Google can not read it. That’s the old saying, at least. Technically, ‘Google’ can read and parse the text in almost any image. There’s plenty of evidence supporting Google’s OCR capability; I won’t refute it. But you should still shun putting text in images, usually.
…Says the guy who just used images of text in his last blog post.
And I’m about to do it again.
Why you should still avoid embedding text in images
Googlebot crawls your site. It finds images. Does it have the ability to opine on the contents of any image? No. And yes. Googlebot is just a crawler. Does Google scan books and make the contents available digitally? Yes. But Google is a lot of things. Just because Google Maps can give you directions to grandma’s house doesn’t mean Googlebot can. (I’m not necessarily saying it can’t, FWIW!)
I don’t know how to super-elegantly make this analogy, but Google can do a lot of things, however, that doesn’t mean that Googlebot does all of them. Another example might be Google Deep Dream. Just because Google can use AI to make weird pieces of psychedelic art from your pictures, doesn’t mean that Googlebot does that to every image on your site. (In fact, Google does that to none of your images, unless you tell it to.)
So yes, Google can look at an image and figure out what other images it’s like. Google can tell you what words are in there. Google can tell you meta information about those words. But can you, or anybody reputable tell me that Google definitely is doing that for all images on all sites by default?
Until that happens, err on the side of common sense.
‘I’m still going to put text in my images’.
Fine. Me too. But we can at least be smart about it.
When I realized in that previous blog post (about an iDrive security breach) that text in an image would be the most effective way to convey information to my very small, very specific audience, I did it. But I didn’t stop there. I also added my new text augmentation of those images to one of the available meta-data fields. Witness what control-u shows:

The text content I added to those images doesn’t appear as selectable/parsable text anywhere on the page, but it’s in the code, so Google will see it. I’ve no doubt this meta data isn’t as important as regular on-page text, but it’s far better than nothing.
What’s more, using “alt” and “description” meta data fields falls under usability best practices for visitors with vision impairment. Screen readers can read meta data. Good meta data vastly improves the user experience for blind people. So, if you don’t care about good SEO, maybe you care about good UX.

Dan Dreifort consults on search and usability. He sometimes blogs about SEO 101 bits, like this. …Mostly so he can just send people a link rather than typing things out repeatedly. Lazy? Maybe.
One thought on “SEO: Text-in-Images – Misinformation & Best Practices”