Contextual Search Isn’t the End of Targeted Keywords… Yet

One of Google’s more recent evolutions was the ability to contextualize content when its crawlers explore websites to determine what they’re about. Certainly from a search standpoint that’s exciting news, but it left a lot of people wondering what that would mean for keyword research.

Is there any point to targeting specific keyword phrases anymore if Google can understand lateral concepts? For instance, if you talk about car maintenance and car repairs, your site is presumably also a source of car tips in general. You could possibly rank for all 3 things without targeting the third.

In my experience there’s still a ways to go before that’s really true. Sites with a significant critical mass of content may benefit from this more than others, but I’ve seen some clear examples of where two very related search terms had totally different rankings.

Examples Where It’s Not What You’d Think

One quick example is a different site I was optimizing where we were tracking local rankings for “business coaching” and “business consulting”. For most people those probably mean the same thing, but the site had mostly been built with consulting in mind and not coaching. As contextually related as they are, ranking reports showed favorable numbers for consulting and no ranking for coaching. Not just less. None.

A difference that significant in such related words simply wouldn’t be possible if search engines ranked contextually to the degree that everyone assumed.

Sooo… for the time being that means business as usual, more or less. Optimizing content for specific keyword phrases still matters, and there’s still a direct correlation between phrasing used and rankings.

Though Google is getting better at understanding correlations between related words and whole concepts, as Moz pointed out recently, people don’t search for concepts.

Where related words does matter, however, is variations of the same word.

In the old days of SEO you’d have to make choices between singular and plural versions of words, since the repetition of that exact word was what improved rankings (for that word). That was annoyingly clunky and not very useful for searchers.

Luckily these days the RankBrain algorithm does a good job of tying searches for “staff” and “staves,” “scarf” and “scarves”. Ranking well for one means ranking for the other.

While actual keyword planning is still relevant, RankBrain does allow website contributors to write more naturally as a result. In fact, the tie-in with Google’s voice search to typed search phrases over the last few years has made major strides in its ability to understand natural language. It uses that data to make associations about what people ask and what they’re looking for, combined with the usual correlations of search phrases and the things people actually click on.

It’s an exciting step forward, and we’ll see how it continues to change the way we search.

, ,

No comments yet.

Leave a Reply

Show Buttons
Hide Buttons