Search Archives

January 31, 2009

Do URL shorterners pass page authority?

This is something I've got to experiment with: do URL shorteners like TinyURL and hurt a targeted page's authority? And if they do intercept the authority, is the added traffic they drive worth the loss?

Services like TinyURL are extremely useful for sending pages with long URLs to people over e-mail or Twitter, where you only have 140 characters. But bloggers use them, too -- because shorter URLs are just easier to deal with.

But how do those services redirect the traffic? When search engines find TinyURL and URLs on the Web, where do they assign the authority: to the TinyURL URL or the underlying page? Because I don't recall seeing bit.lys or TinyURLs in search results -- and I look at a lot of search results -- I suspect that they pass the authority just fine. But it would be a big deal if they didn't; a few good backlinks can be difference between a non-existent search position and an excellent one.

February 2, 2009

Nice Algorithm You Got There

Google would like you to believe that it's all automatic, that there is this army of search spiders that digs out every last page and image on the Web and decides which is "better" for any given search term. It's true, as far as it goes, but the company tends to carefully elide the human element that goes into its search result. Until something goes horribly wrong, as it did Saturday morning. For an hour, Google said every site on the Net was dangerous -- itself included.

Continue reading "Nice Algorithm You Got There" »

February 9, 2009

Newspapers cut 9 percent of staff in 2008

Ow. Ow. Ow.

From AdAge:

The U.S. advertising and media industry slashed 18,700 jobs in December, bringing industry job losses in this recession to 65,100...

Continue reading "Newspapers cut 9 percent of staff in 2008" »

June 12, 2009

SEO is where marketers should start

If you want to sell something, tweaking your organic search is a great place to start. That's the conclusion of a new study (PDF) published by Forbes.

The study found that:

  • The tools seen as most effective for generating conversions were SEO (48 percent) email and e-newsletter marketing (46 percent), and pay-per-click/search marketing (32 percent).
  • In the coming six months, respondents expect that ad networks will see the biggest declines in allocation of marketing spend; viral marketing and SEO will likely see the biggest increases. Behavioral Targeting is the category that is least likely to see any changes in spend.
Note that "effectiveness" is defined here as generating conversions, not mere page views -- and that SEO is half again more effective than PPC. Notice also that SEO campaigns exceeded expectations of 45 percent of respondents; the next most satisfying tactic -- PPC -- exceeded expectations of only 25 percent of those surveyed.

And one more encouraging thing: the companies Forbes surveys understood that the important thing about search isn't traffic (37 percent) or SERP position (34 percent.) It's conversions: 70 percent.

Organic search = money. Remember that.

June 15, 2009

Dan on SEO

Rob Kelly, as best as we can put it together, is a guy I didn't quite get to work with CMP. He was a senior adviser to the SEO out on the West Coast while I was editing NetGuide in New York. Now, he's off building building a bunch of interesting sounding new companies.

Rob interviewed me the other day about SEO, and it came out pretty well, although I was talking a little faster than he could type. Click through for the whole deal, but here are a couple of pullquotes:

Ranking in the SERP (Search Results Page) is meaningless. Anyone can get to the first page for something. What I always watch for is traffic, and changes in traffic. I care about the conversion of what happens once someone hits my page…clicking the buy button or the ad. I can rank #1 on a search of “cellphone”…but if they come to my page and don’t convert, all I’ve done is cost my company money. If I can generate meaningful traffic to my reader, to my customer…that’s the win.
SEO isn’t an event, it’s a process... Too often, employers aren’t emotionally equipped to understand what SEO really is — it’s a quality process… that involves the entire company. When Toyota decided they were going to out-quality Detroit, they didn’t hire a quality guy and stick him in a cube. They hired someone who would come in and look at the operations of the entire company and build a process that baked quality in. And the best companies that do SEO, bake SEO in.

July 1, 2009

What's new about Bing?

Not long ago, I gave an interview to Betanews, which (who?) wanted to know what Microsoft's new Bing search engine was up to.

A white paper from Microsoft didn't provide much detail about Bing's algorithm, but was forthcoming about why and how its user interface got that way. Interesting stuff.

But what's even more important is that Bing (like Google) is presenting ever more of a web site's content before users go to a web site. If you're a site owner who tries to monetize eyeballs, you should start getting the message that you're less and less in control of the presentation of your information. That's not a good thing.

August 13, 2009

Optimizing press releases

I was quoted yesterday in a BNet article about using press releases to boost backlink profiles. It's a good article with lots of useful information. Inevitably, as part of the process, lots of better stuff from the underlying e-mail interview got cut. Here's a fuller version of what I told the writer, the estimable Drew Kerr. (Questions are paraphrased; answers are verbatim.)

Q: What does it mean for a press release to be "search engine ready"?

Search engines put a premium on relevance, so the press release services work (to varying degrees) to make the release as relevant to the release's subject as possible. That means using the targeted keyword repeatedly, and linking to the site being promoted with keyword-y anchor text.

It sounds simple. It isn't. First of all, press releases are usually written with more care than United Nations resolutions. They're the result of endless rounds of writing, re-writing, negotiating and re-negotiatiating -- and it's worse if there's more than one company involved. By the time a press release gets to a distribution service, the competing interests have been balanced like a Wallenda. The distribution service isn't going to have any leeway to change words just for SEO purposes. The best they can do is stick a bunch of links onto the most promising text.

That's a shame, because most releases have been so acutely lawyered that they don't use words that most constituencies would want to search for. Press releases, of course, have multiple constituencies -- the media, the public, investors, competitors, government agencies, suppliers, consumers, the closely interested, the casually interested, and now search engines. They frequently will use different terminology to discuss the same thing; that's poison for SEO, which is most effective when the copy uses language that people are actually searching for.

For all of those issues, distribution services are a fine way for a site to to gain authority by building backlinks. Unless it is hopelessly ham-handed, a release about the latest frammistat from XYZ Corp. will inevitably use the words "frammistat" and "XYZ Corp." repeatedly enough to appear relevant for those terms in search engines. By including a few links back to the XYZ Corp. Web site -- particularly using the word "frammistat" as the anchor text (and, even better, linking to the frammistat landing page on -- the release distribution service has cast a "vote" for the XYZ site. The more votes for a site on any given search term, the more relevant the search engines see that site as being, and position on search results pages go up.

But wait -- it gets better. Many web sites republish press releases put out by the distribution services. So if publication sites that search engines regard highly take the bait -- sites like,, whatever -- now *they* are linking to the XYZ Corp. Web page. The technical SEO term for this is "jackpot."

Q: How do people go wrong making documents SEO-ready?

Over-optimizing. It's true that repetition is your friend when you're writing copy that you want to be especially SEO friendly. But even though repetition is your friend, it's possible to overdo it. Repetition is your friend, but keep in mind that search engines understand that the Web is for people and not machines, and if you insist too hard that repetition is your friend you will write copy that no one will read -- and although repetition is your friend, search engines will understand that they are being pandered to. And repetition will no longer be your friend. Neither will anyone else.

Q: Does this work?

As noted earlier, it's very effective and it's an excellent way to start building a good backlink profile. But like anything that works, it's possible to overdo it to the point that search engine algorithms catch on and start devaluing those links. If the only sites that are linking back to you are news distribution sites, you need to rethink the way you're addressing the public. For one thing, if your only backlinks are from distribution services, it's a good indication that no one cares about the news you're putting out and search engines will understand that. At their core, press releases are a not-especially-pernicious form of paid links, which search engines -- Google in particular -- rail against. Like anything that can give a sugary rush, press releases should be a delicious part of a well-balanced SEO breakfast -- not the whole thing.

Q: Why are backlinks important to search?

Backlinks are important because Google sees them as votes of relevance. If one page has 1,000 backlinks from relevant pages and another page has 10,000 backlinks from relevant pages, Google sees the second page as being more relevant for a particular search term -- and will place it higher on the SERP (search engine result page) for that search.

Counting backlinks and judging their relevance is what Google's revolutionary PageRank algorithm is all about. But it's important to remember that PageRank is only one of 200 or 250 "signals" that Google uses to rank pages.

And it's also important to remember that only Google has ever said anything about how it ranks pages. Yahoo and Microsoft have never said how they build their SERPs. It may be some PageRank knockoff, or time-based randomization, or chicken entrails, or 5 million hamsters running on 5 million hamster wheels with roulette numbers randomly attached to them.

To searchers and readers, all that matters is that they get highly relevant results quickly. If Microsoft ever decides to reveal anything about how they rank the Web, you could expect SEOs to start optimizing more for Bing, which will increase the utility of Bing's index, which will draw more users, which will increase the advertising value of Bing's index.

August 14, 2009

Is Google shrinking?

There's a remarkably clueless post on ZDNet today, wherein someone who should know better says that Google's index is 25 percent smaller than it used to be. His evidence: when he checks his own name, he found only 102,000 instances, down from 135,000 in March. (, he says, has 157,000 results.)

The technological Onanism of this aside, it's just a dumb observation. The idea of a search engine isn't to provide the most results. It's to provide the most relevant results. Maybe those "missing" 33,000 results weren't especially interesting; it's hard to imagine that 100,000 results on anything, let alone a blogger, would be worth investigating. (And I mean, really -- who keeps track of the number of results for their own name?)

Does this have anything to do with Google's "Caffeine" update (about which I'll write more shortly)? Possibly, but unlikely; Caffeine seems to be about speed and resources, not index size.

August 26, 2009

Google and Bing as a threat to content

Bing has now been around long enough for people to start looking for referrers in their server logs. Most people aren't seeing a ton of traffic from Bing so they think it's not a big deal.

It's a dangerous and possibly self-deluding conclusion for any content provider. Remember: you -- the content provider -- are not the search engines' market. You are, in fact, the product that they're selling.

Bing is not innovating in search, as far as anyone can tell. It gives some very different results than Google; in many cases, it presents much deeper results than Google's while missing other stuff.

Where Bing (and Yahoo) are innovating are in user experience. The goal of all these search engines is to give their users as much information as possible without their leaving the SERP. You want them to click on your URL to see your content and ads. But the search engines would just as soon that their page be the final word. That's why Google is relying less on Description tags and more on page scrapes and microformats for its snippets. It's why Bing has page excerpts pop up next to the URLs on the SERPs.

Bing has looked at the heat maps of what people look at in SERPs and is innovating around that upper left quadrant of the window. Google is making more options more easily available to searchers. But what kind of business model would your site have if it only existed to send people away? Right: none. The search engines are ever more in the business of helping people stick around, showing your information on their pages, and building environments where they let people leave only if they really want to -- but would rather have them stick around, thanks.

December 7, 2009

Google Real-Time Search: some questions

Google today announced its inevitable reach into real-time search, instantly adding results from Twitter, FriendFeed and MySpace. As cool and useful as this may be, I've got a couple of questions about it.

How do they assign relevance to tweets? Google's search algorithms are based, in large measure, on the number and quality of backlinks -- relevant links to a given page. How is Google assigning relevance to tweets? By hashmarks? By the number of followers on an account? That's obviously problematic, given the triviality of manipulating it. By the number of relevant followers? Then what determines relevance?

Is more weight given to official pronouncements from known or trusted entities? How do the entities get to be known or trusted? And if "official" sites do get preference, isn't that contrary to the whole idea of Twitter?

What about a common situation where a reporter for a publication tweets about a story that's about to break on their site or in print? How is that differentiated from an unsubstantiated rumor that some celebrity has died? It's not Google's place to be judging the truth value of tweets or posts, but I worry (as often) that this will only accelerate a race to the bottom.

And one more: Tweets are 140 characters long. Google snippets are 156. By presenting tweets on SERPs, Google is preempting one possible means for Twitter to monetize: ads. Why click through to a Twitter page when you get all you need on Google's?

March 26, 2010

New Bing Coming

Microsoft's Bing search engine will be rolling out UI changes starting in the next few days. Since its launch about a year ago, Bing has been innovating mostly on its interface, and these changes continue that mission. The emphasis for the update will be on providing more context -- including real-time feeds -- and visual information.

What's worrisome about Bing, from a content provider's perspective, is that the search engine will provide so much context that a click-through to the originating site becomes unnecessary. (This, of course, allows Bing and not the publisher to monetize the publisher's content.) Microsoft doesn't see that as much of an issue. From Redmond Channel Partner:

One attendee of the keynote asked [Yusef] Mehdi [Microsoft's SVP of online audience business] if Microsoft’s efforts to render more contextual data within Bing would result in fewer click-throughs to sites. Mehdi responded that he doesn't see cause for concern. "What we have found is there are more click-throughs when you add richer captions," he said.

There's a interesting data point, if he'd be willing to share his numbers. It would also be interesting to discover if rich-content clickthroughs convert on a publisher's site better than clicks based on less-rich content. In other words, do people who click on high-context links bounce more or less than referrals from low-context links? Or are high-context links killing the geese with the golden eggs?

April 23, 2010

Auto-captioning YouTube

I somehow missed the news that YouTube is now automatically captioning all videos in English. That's awesome news for the accessibility crowd. It's a little problematic for the content industry.

Think about it. There's about 20 hours of video uploaded to YouTube every minute. For every one of them, Google's now generating text. This text is not created on the fly as the video is played back; it's presumably built as part of each video's preprocessing and is made part of the video's metatext. Which means it's being indexed, which means that the videos are now competing directly with more traditional textual information.

It's true that the Google voice-to-text technology is not exactly accurate. (My Google Voice transcripts are generally pretty incomprehensible.) But this is all the more reason that serious content competitors need to be looking more than ever at using YouTube to protect their brands.

May 23, 2011

Wuz LinkedIn Robbed?

There's a provocative column by Joe Nocera in today's NYTimes about LinkedIn's IPO last week. Nocera thinks that the investment banks Morgan Stanley and Merrill Lynch -- which LinkedIn hired to take it public -- essentially stole hundreds of millions of dollars that should have gone to LinkedIn's treasury.

Here's how it works. The investment banks gauge demand for the shares, which are sold by the company's treasury, and set what they believe is a fair-market price. Proceeds of those initial sales go to the company. In LinkedIn's case, Morgan and Merrill set a price for the shares three times, the last and highest being $45. LinkedIn took in $352 million for 7.5 million shares. The investment banks get 7 percent of that.

But once the shares are out, they can be -- and are -- traded freely. On its first day, LinkedIn traded as high as $122 per share, closing at $94.25, more than twice the initial price. LinkedIn got none of that money.

Who did? The people to whom Morgan and Merrill sold the initial shares. That's usually their best customers, people with connections, or other BFFs. A 100 percent gain in one day is a nice day's work.

Nocera says it's a scam. It's the investment banks' job to know what the market sentiment is, he says, and it's their fiduciary duty to price the coming-out shares as close to what they think the market will pay. And at the end of the day (literally), the market was apparently willing to pay $90. LinkedIn should have collected not $352 million but $700 million, Nocera argues. And their investment banks should have been paid $49 million, not a mere $25 million or so.

Why would the banks leave $25 million on the table by underpricing? After all, taking a company public is hard work. Maybe they traded some of their own stock and gained more than that. Maybe they were willing to chalk up the $25 million as a cost of doing business to appeal to their best customers. Promotional expense, you know.

Is this something that should be -- or can be -- fixed? I'm not so sure. I don't know that it's an investment bank's responsibility to take into account a market gone nuts. Maybe a best effort is all that's truly called for. though how you'd judge that is a mystery for smarter minds than mine. I know I don't want the SEC putting its thumb on the IPO scale.

Maybe a Dutch auction, which is how Google went public, is the right anwer. (The Street hated it, but it doesn't seem to have hurt Google's prospects any.) Maybe there should be a rule that a percentage of all first-day sales -- say, 7 percent, same as the banks' take -- be funneled back to the company issuing the shares. It might make for a more orderly coming out, or it might just move the madness to Day 2.

This may just be one of those things that Just Isn't Fair. One thing Nocera's inarguably correct about: it looks like hell at a time when the financial business ought not to be resorting to Old Tricks of the Internet Bubble.

May 24, 2011

Will Google Buy Sprint?

So I'm going to start a rumor here: I think, before the year is out, that Google is going to try to buy Sprint.

The ties between the companies...

Continue reading "Will Google Buy Sprint?" »

About Search

This page contains an archive of all entries posted to Over the Edge in the Search category. They are listed from oldest to newest.

Politics is the previous category.

Tech is the next category.

Many more can be found on the main index page or by looking through the archives.

Creative Commons License
This weblog is licensed under a Creative Commons License.