We receive a steady stream of queries from users whose sites our keyword analysis engine Qirina refuses to index due to paucity of identifiable text content. The website owner will have, for instance, a gallery of daguerrotypes, or artisinal merchandise for sale, and Qirina will return an error that the site is in non-English or uses unnatural language, or that there is insufficient content for analysis.
This shouldn't be so suprising. Search engines are driven by text. A lot of fancy number crunching, including linguistic identification and lexical analysis, may be going on, but the base technology hasn't changed since the early 1990's: the search engine looks at your keywords and attempts to discern what it is about based on the incidence frequency of the keywords.
How well your site is ranked, of course, is another matter, but the determination of which category of keywords to rank it in still depends almost entirely on the keywords on your site.
Site owners who have just pictures (and no keywords) on their front page are making life hard for the search engines. Ditto for site owners who have ecommerce sites with nothing but a list of product names.
Similarly, if no natural language is used, and all the search engines find is a list of keywords, it makes it a bit hard to tell if your site is legitimate or a spammy site stuffed with keywords.
Good ecommerce sites include a little bit of natural language text. It may even help human visitors. Here's an example of an ecommerce site the search engines may have a hard time with. It looks great but there's precious little natural language on the front.
