Arrrgh… Google Still Isn’t Recognizing Underscores as Word Separators in URLs

May 19
2009

Although it isn’t a primary “signal” like the title tag or anchor text, keywords in your URLs can help with your Google rankings. But ONLY if Google can see the actual words in the URL. Turns out that separating the words in a URL with hyphens allowed Google to see the individual words, but using underscores did not. And this, unfortunately, continues to be the case today.

Not quite two years ago at WordCamp, Matt Cutts made the following statement that Google was imminently going to be treating underscores as word separators:

The interesting thing is we used to treat underscores as if they were like word A underscore word B, we would glom that together and we would index that as A underscore B, so if you just searched for the word A, we wouldn’t return your post. Ah… We’re in the process of changing that. We might have already changed that. So dashes and underscores are almost exactly the same.

You can hear the above statement for yourself in this video of Matt’s talk, at around the 17 minute mark.

I excitedly wrote about it in a post for the News.com Blog, since historically keywords separated by underscores didn’t look like separate words to Google, and this would save a lot of folks a lot of time if they were embarking on a URL rewriting project to fix their underscore problem.

Unfortunately I jumped the gun a bit, because Google still has not made the switch to recognizing underscores as word separators like they do with hyphens.

Your next question might be “But are you sure??” Yup. When I spoke to Matt in February at SMX West, he confirmed that underscores were NOT treated as word separators. According to Matt, this change is still in their queue but unlikely to happen before summer. My interpretation: don’t hold your breath, it’s between summer and never. ;)

Why didn’t they roll out that change? Certainly it’s clear it’s not a priority. Google engineers are focused on improving relevancy and improving the searcher’s user experience. I would guess that this particular tweak to their algorithm isn’t going to do much for their users.

So, in your URLs, keep favoring hyphens over underscores for the foreseeable future.

And here’s one gotcha to be aware of: don’t use an underscore to separate a lookup ID from hyphenated keywords. For example, a URL like http://www.example.com/1234_nike-pegasus-running-shoes.html may at first glance appear to be search engine optimal, but the keyword “nike” is not visible to Google as a separate word. The keyword is actually understood by Google to be “1234_nike”, not “nike”.

By the way, although I favor the hyphen, there are other word separators accepted by Google, such as the dot (.), the plus sign (+), and the “escaped” space character (%20).

"Thin Slicing", a Powerful SEO Tactic

May 19
2009

In my Search Engine Land column last week, I describe a powerful SEO tactic that we at Netconcepts call “thin slicing”. The term originally comes from Malcolm Gladwell (as used in his best seller Blink) and has no origins in the online world.

Gladwell uses the term in the context of “rapid cognition”; where one makes snap judgments in their field of expertise. Surprisingly, those snap judgments are often times more accurate than considered opinion, i.e. assessments that have been labored over. The important caveat: it only holds true for experts, not for amateurs.

We’ve co-opted the term and applied it to SEO. In that context, thin slicing is a tactic referring to mass optimization across a large number of pages, done quickly, and confined to just one or more high value elements (such as title tags). It relies on the gut-level instinct of the search engine marketer. Spare the in-depth keyword research and analysis and just take a guess, then move on. When you have a daunting number of pages to get through, deciding on synonyms, verb tenses and word order should rely on your intuition. Trying to optimize every element on every page perfectly is not scalable and will only sap your energy. “Thin slicing” could be done on title tags, keyword URLs, H1 headings, or meta descriptions. You’d monitor for impact, and then refine based on those results.

There are two approaches to thin slicing, and which one you use depends very much on your web site’s infrastructure and what it supports.

  • One is through your a forms-based web interface in your admin. We refer to this as “mass edit” capability. WordPress supports mass editing of title tags and URLs (”post slugs”, more accurately) - IF you have our free SEO Title Tag plugin installed. Through its mass edit screen, you can optimize all title tags across your blog - all your posts, category pages, tag pages etc., without having to go to each post’s Edit screen individually.

    One feature we found invaluable when using web forms for thin slicing was to make the number of rows displayed per page user-configurable. Some users will want to display hundreds of records per screen, others will want much fewer, as too big of a web page will cause their web browser to crash or time out.

  • The other approach is “bulk uploading”, where you import an updated list of title tags (or H1s or whatever) into your website’s underlying database. You start with a database export in CSV (comma separated values) format of your current title tags — along with the corresponding item ID numbers for each record, of course. Load the CSV file into Microsoft Excel and do your title tag optimization in the spreadsheet. Then upload the optimized title tags back into the database.

    Note that if your database does not have a field for the title tag, you’ll have to create it and re-code your site to override the programmatic title with the contents of this new field when it is populated with data.

    Rather than having to maneuver through phpMyAdmin or rely on your database administrator, have a CSV file upload function built into the admin interface of your content management system (CMS).

When we added the “bulk upload” capability to our GravityStream proxy admin, our optimizers and those at our clients and partner resellers experienced a nice boost in productivity. So we can attest to the fact that “thin slicing” works.

Whether you prefer working in Excel or within a “mass edit” view in your CMS’ admin interface, “thin slicing” is a great tactic to add to your SEO toolchest.