All About Content

On Monday, while doing client research, I discovered that Google is now displaying a site search box underneath Google Sitelinks in results for some general queries.

Sitelinks for navigational searches are becoming more common and you’ve long had the option to see “more results from [domain]” if you had an indented result — the “more results” link would show you results from the query you had just conducted but limit them to just pages from that domain — but a search box within the results is much more interactive. A search box basically prompts you to refine your search, but do so only within the most authoritative site.

Even more fascinating to me, however, is that this internal site search is displayed not just on results where the search was clearly navigational — like a search on “Intel” might have been.

Doing searches on bible-related phrases (don’t ask, it was for a client), I saw the same thing for queries that seemed pretty general.

I ran queries on [online bible], [bible search], [bible passages] and even queries using advanced operators and still got not only Sitelinks but also a separate query box for searching that specific site.

Imagine how powerful this type of display would be if you’re an e-commerce retailer! It immediately gives your site an appearance of even more authority and people can search your site without even needing to click away from the SERP.

Queries conducted with this Sitelinks search box also get saved to the user’s Google Web History if they’re logged in. Something you wouldn’t get if they used the search box inside your site.

Neither of the two sites given in the examples use the Google Search Appliance - so whether you use the Google enterprise system for your internal search doesn’t appear to be a factor in getting this to show up for your results. (The Google Search Appliance is a horrible choice for site search, btw. Not at all suited for site owners looking to provide on-site search functionality.)

I did a search on [Washington DC SEO] in Google and found a local listing in the #1 spot. Not surprising with Universal Search and the inclusion of a city name clearly indicating that there’s a geographical component to the query.

(Actually, I used to have a first page ranking in Google for Washington DC SEO and a #1 spot for the same phrase in Yahoo and MSN.)

So if you’re a small business looking for SEO services, you might think, “wow, this guy must be good; he’s #1 in Google after all!”

Here’s the funny thing: There’s no there there. The site has no content on it. This is what you get when you click on the link:

Luckily for him, the OneBox listing includes a phone number, so the guy might still be getting clients (yeah, right!)

Is that all it took to be #1 for “Washington DC SEO” - registering an address with Google Local Search? I’d be all over that, but I don’t think I want to make it that easy for any old stalker to know my address. What a shame.

p.s. I’d upload a screen shot, but for the life of me I can’t figure out how to do screen grabs on my new Mac.

Does an over-optimization penalty exist? And what counts as over-optimization? I came across this discussion on on the “950 penalty” on Webmaster World. They posit that too many internal links with optimized anchor text is what triggers an over-optimization penalty. And maybe they are onto something.

This week I was testing the effect of increased cross-linking on a Wikipedia page — by finding related articles and linking back to the target, by adding the target to more relevant categories — and I probably doubled the number of internal pages linking to the one I wanted to boost.

Instead of boosting the page in the SERPs, it appears this effort might have torpedoed the page. This morning it was nowhere to be found in the top 100 results anywhere. It’s not a competitive term, and none of the other results appear to have changed dramatically, so it seems reasonable to me to assume it was something about this page, not the results in general, and that there was a causal relationship between the links and the drop.

I’m not entirely surprised that “excessive” interlinking could hurt; I’m just surprised that (what I’d consider to be) a moderate amount of interlinking could get a Wikipedia page penalized so quickly. I truly believed pages on that domain were more robust and could stand up to that strategy. (After all, there is a ton of internal linking and all internal links on Wikipedia use optimized anchor text.)

Well, time to undo some of those links and see if it comes back.

Update: March 2
I removed most, but not all, the internal links and the listing came back rather quickly, though in a lower position than before.

On an upbeat note … I managed to get a Wikipedia page erased from Google, temporarily? Damn, how awesome is that?

ComScore’s Year in Review press release seems to underscore the principle that the rich keep getting richer: Google, of course, saw more gains, as did Wikipedia (which some conspiracy theorists seem to think is in cahoots with Google somehow) and Craigslist.

Facebook traffic jumped 81% year over year to 34.7 million visitors, now that registration is open to non-students (including quite a few pets, if Stewie’s ever growing circle of Facebook friends is any indication).

The release goes on to say that “the top-gaining site categories in 2007 reflected trends in both the online and offline worlds. The politics category grabbed the top position, gaining 35%, as the 2008 presidential election and primary season kicked into high gear.” Not to be outdone by current events of any gravity, sites devoted to celebrity entertainment news, “from Britney Spears’ meltdowns to Anna Nicole Smiths death,” kept up with an equally impressive 32% increase in visitors.

In total — including all searches for Britney, Anna Nicole and even “poop porn” — more than 113 billion core searches were conducted in the U.S. last year, with Google representing a 56% share of the market.

Breaking news: there’s more FUD about showing up in Google results!!! SSDD, as they say.

Google has announced they’re getting rid of the supplemental index, but I honestly don’t see what the big deal is. If your pages were crappy (or irrelevant or obscure or esoteric, or whatever gets them banished into the supplemental index), why would placing them into the main index make them any more likely to rank well? In other words, if your page is no good, what’s the difference between being in the supplemental index and being ranked #568 on some long-tail query?

“Oh, but supplemental results aren’t recrawled as often.” Right, but so what? No guarantee you’ll get crawled more often just because they combined the two indexes/indices now. “Oh, but my site didn’t deserve to be in the supplemental index in the first place!” But the point is that Google thought it did; and Google is still not going to think you deserve to rank well.

All the things you needed to do to get out of Supplemental Hell are the same things you’re still going to have to do to get found and showing up high in the results… and now you’re competing with that much more garbage in the main index.

The only stuff of mine I ever saw in Google’s Supplemental Hell was a lot of duplicate content stuff that I never wanted crawled in the first place. If anything, this change will make it harder for me to get Google to spend its resources on the pages I want it to be indexing. As far as I’m concerned, Google is just dumping garbage back into the main index.

Not an improvement for webmasters or searchers.

Original Google Webmaster Central post here:

News from PubCon: Google is looking to reduce the occurance of subdomains in SERPs.

That doesn’t sound nearly as exciting as “Google is going to start treating subdomains like directory folders when it comes to deciding how many results to show per page” as is implied in a recent Search Engine Roundtable post, but that’s still pretty interesting news out of Vegas this week.

According to Matt Cutts, “in some circumstances, Google may move closer to treating subdomains as we do with subdirectories [sic].”

Either way, the characterization of this is a “filter” seems apt. Sites have long used and abused subdomains as a way to dominate the SERPs - rather than getting just 2 of the 10 spots on a results page, savvy webmasters have known that 5 subdomains on the same site can help you take up all 10 results on a page.

Time to start rethinking some strategy.

More on this in a bit…

Google has been warning the SEO community it would put the smack down on paid links. The message started coming in loud and clear earlier this year with the Google guys and gals talking at conferences about how the GOOG don’t like no bought links, and that sparked a lot of debate on all the forums: Just how would search engines distinguish paid links from “natural” links?

Then, in case you weren’t listening, Google started a de-PageRankifying campaign that was sure to get the community talking: first it dropped the Toolbar PageRank of paid directories (presumably those that had been created for SEO purposes) and then, more dramatically, it stripped major general-interest sites like the WashingtonPost.com of their precious green Toolbar pixels.

“You listen’ now, punks?” asked Google.

That, all the savvy SEOs knew, was just a shot across the bough bow, because does WaPo really care about how many green pixels it gets in the Toolbar? Does that change the economics of their business? No. But…with the holidays coming and Google having a reputation for making big, disruptive algo changes right before the big shopping season, I for one warned that sites utilizing paid linking as part of their link strategy would see the effects soon.

And in three… two … one:

Here’s are the Google rankings of one site for its top keywords and keyword phrases — check out what happened to that site’s rankings since Friday (I was out of the office on Monday, so it might have happened as recently as yesterday). [ed. I should note that none of the affected pages themselves lost any Toolbar PR, because the TBPR thing is only symbolic. It's the rankings (and therefore traffic) that actually matter.]

Just eyeballing this chart, it looks like paid links were stripped of their link juice, but not that the buyers of those links were heavily penalized. In the example above, most of the rankings dropped only a page — that’s significant, don’t get me wrong, but it’s not as bad as the dread Minus 30 penalty, or an outright ban. Unlike the Minus 30 penalty, there isn’t going to be an opportunity to beg forgiveness. The links that were devalued can be removed, but that won’t help get your rankings back up. In fact, it’s entirely possible that Google isn’t done with this update yet and that further drops in these sites’ positions will occur.

But not everyone loses out in this latest rankings shuffle, obviously. Some sites — those that didn’t participate in detectable paid linking — get to move up a couple of spots. (Well la-dee-da, lil’ Goody Two Shoes.)

But, you knew the “buy yourself top organic rankings” strategy wouldn’t last in the long run.

Or maybe I speak to soon… also possible that this was a one time index clean-up and not a permanent change to the algorithm, in which case you could continue to discreetly buy links on other sites that haven’t been outted yet. hmmm…. I’m not saying that paid links are the best value for your money right now, but people tend to overreact to changes in the search engine landscape. I really wouldn’t advise you to pull the plug on every paid link you have out there; knee-jerk reactions to an algo shift aren’t going to help. There’s no way that Google figured out every single link that’s been paid for on the entire Internet … point of the hyperbole: some of those paid links might still be worth quite a bit. The trick is knowing which ones.

I also want to know how this is going to effect business of the big text link brokers. Are people going to stop buying text links now? I for one still think there’s plenty of opportunity there, as long as sites selling links stop being so frickin obvious about it. Like the WashingtonPost - they put the same block of paid link on ALL of their tens of thousands of pages. Of course that was going to be detected!

How ’bout you? Are you changing your strategy? And how do you hope to make up for the lost rankings now that the paid links aren’t boosting your position (assuming you paid for links, that is)?

Updated December 7: May I Have This Google Dance?
Less than 2 weeks after I published the post above, a funny thing happened. All the rankings came back to what they were before and have been holding steady since. I think they listened to me when I said “good things come to those who don’t panic.” (Paid links are dead! Long live paid links?)

It’s the end of the month and I realized I’ve barely posted at all in the last few weeks. There are at least half a dozen posts started but never finished just sitting in the queue, but a heck of a lot of good that does anyone.

So, yeah, I’ve been pretty busy. Lotsa stuff going on at work. I can’t really talk about it (shhhh) but I’ll mention that I had to lay off two of my staff, which was really hard to do. I’m also spending more time on brand management (again) and product merchandising, the latter of which generates a greater sense of urgency going into the holiday season than SEO projects do. (Xmas SEO? Puh-leeze, we took care of that back in July. J/K)

Obviously the big SEO chatter this month was about the contentious paid links debate and the related issue of several major sites (like the WashingtonPost.com, for example) losing a lot of PageRank. Bruce Clay does a nice job of rounding up relevant posts on the subject here and here. If you haven’t been following the debate, you might want to start with Rand’s roundup of the links session at SES San Jose.

Other news some of you may find interesting is that Google seems to be changing the way it displays sitelinks (at least intermittently). Here’s an example of the new 8-link layout I’m seeing a lot in Google’s sitelinks:

I’m hopeful that I’ll be able to start blogging again more regularly soon, once the work drama subsides (assuming it does).

Over many months, I’ve noticed that Google Sitelinks (the list of internal or secondary links that display under the first result of a search) can differ dramatically. Sitelinks seem like one of the most inconsistent result types being displayed.

Following are some examples.

A search on “Sears” brings up the Sears homepage with Sitelinks that match the general navigation. Aside from the fact that no second page from the Sears.com domain shows up on the first page of results, this search engine result is displaying precisely as one would expect: Sitelink destinations match the main navigation and the anchor text for the links match the alt attributes for the navigation images (since the nav is not plain text).

In this next example, two results show up for a search on “Drs. Foster and Smith” (a pet supply web site). Second result is a duplicate of one of the Sitelinks, which makes sense. Interestingly, the anchor text within the Sitelinks portion of the results is inconsistent.

The link to “Cat Supplies” matches the destination page’s title tag, whereas the “Dog Supplies” link matches the anchor text of the navigation link. The “Fish” Sitelink matches neither the anchor text of the site’s navigation nor the page’s title tag, though the second result, which goes to the same fish page, does use the title tag. The final Sitelink, Clearance Outlet, is not a link found in the site’s main navigation, though it is on the main page of the site.

A search on “drsfostersmith”, which is the domain name of the previous company, brings up the same result and the same sitelinks, but with a different Title link for the main result! The second result following the Sitelinks was not from this domain.

A search on the brand name “T-Mobile” brings up two results from the http://www.t-mobile.com/ domain — neither of which uses the respective pages’ title tags in the results display. As would be expected, the Sitelinks go to pages that are located on t-mobile.com’s main navigation and uses text that is either anchor text or alt tag text for those links.

The second result, however, is NOT one of the Sitelinks, and is actually a 302 redirect to another domain. Neither the page titles, nor Sitelinks anchor text, are pulling from Page Title tag (in either the first or second result):

For a search on the brand name “Wirefly,” two results from the wirefly.com domain are displayed. The second result is also one of the Sitelinks, with all but one of the Sitelinks being ones found in the the site’s main navigation.

Both the Sitelink anchor text and the title of the second result appear to be pulling the page’s Meta Title, again as expected; however, in the case of the former, the Meta Title isn’t current and doesn’t match what was indexed in the general results (note use of Cingular versus AT&T):

Conclusions, Conjecture and SWAGs about Google Sitelinks:

• Google explains that Sitelinks are created by “analyz[ing] the link structure of your site to find shortcuts that will save users time.”
• However, Google Sitelinks are not necessarily the pages with the most internal links pointing to them (i.e., pages from the main navigation). It’s not entirely implausible that Google is analyzing user behavior (via the Google Toolbar) to determine which “shortcuts” are most likely to help users locate the information they are seeking. (Take a look, for example, at the Sitelinks that show up for a search on “Yellow Pages”; the pages that are displayed –pages about New York and LA — aren’t favored by the site’s structure, but are very likely heavily trafficked pages.)
• Anchor text of the Sitelinks results may or may not match the text used to link to those pages within the domain.
• Anchor text of Sitelinks may or may not be pulled from the pages’ meta title tags.
• The data for Sitelinks appears to be pulling from a different “index” than the main results, since title tag changes are updated in the main results more quickly than in Sitelinks.
• Your site does not need to have two of its pages showing up for a brand-name search in order to have Sitelinks displayed.
• Conversely, having two pages from a domain show up for a branded-keyword search does not guarantee display of Sitelinks.
• Short of re-architecting your site, there doesn’t seem to be a way to control which pages will show up in Sitelinks or what the anchor text for those links will be.
• Sitelinks tend to show up most often for searches on brand names, but not exclusively. I’ve seen examples where generic search phrases result in Sitelinks being displayed for the top result, but the keywords have tended to be used as part of the domain name.

More info about Google Sitelinks can be found on SEObytheSea.com, where Bill Slawski looks at Google’s patent application.

Updated: March 12, 2008
New post about additional search boxes now being displayed within the SERPS, inside the Google Sitelinks area.

Hey fellow Washingtonian Googlephiles, just saw a job on Yahoo HotJobs for a Google Product Marketing Manager. Go check it out if you’re interested or look for other Google jobs in DC (they’ve got stuff for librarians, lobbyists and a few things in between).