|
Repost: Read original and complete article here
Three New False Myths the SEO Industry is Busy Birthing
Search engine optimization is simple, really. You try something and if it doesn't make you happy you try something else. That is search engine optimization in a nutshell. When I started writing SEO Theory years ago I spoke about The SEO Method: Experiment. Evaluate. Adjust. But really it can be said more simply: You try something and if you don't like the results you try something else.
It's all a bit gut-level and unscientific when you look at it that way but there is darned little science being applied to search engine optimization today. Here are three examples where the failure to apply real science has created some new cautionary tales in our community repetoire of SEO tricks and pseudo wisdom.
Myth No. 1: Short Content Is Okay
Barry Schwartz occasionally highlights comments from Google employees that he feels provide useful insight or debunkery. Barry's judgment in these matters is (in my humble opinion) exceptional and that is why I have been a faithful reader of Search Engine Roundtable for years.
One of Barry's latest highlights points out that Google is telling people "Short Content Can Be Useful & Rank Well". Okay, we have known this for years but somewhere in the wake of numerous Google crackdowns on schlocky pseudo search engine optimization over the past two years people have begun to mutter that "Google does not like short articles".
As someone who has argued that size doesn't matter with Google (in more than one way) I can well understand Barry's concern over these misconceptions that arise (apparently) out of little more than fearful guesswork.
Someone with short articles got burned and therefore they concluded that the short articles may/must have contributed to the problem. That happens every day of the week. It's a never-ending process. We can't stop it. People who know no better — or who get all their SEO information from SEO blogs and forums — will leap to wrong conclusions. Heck, we all leap to wrong conclusions — there is nothing wrong with that so long as we learn to recognize our mistakes and learn from them.
But here's the rub: SIZE DOES MATTER. Yes. I said it. Size matters.
Size matters in ways that most SEOs don't even think to measure. Size of article matters because the shorter the article the fewer queries it will potentially rank for, the fewer topics it will potentially apply to, the fewer people who are likely to find value in it.
I still have friends who to this day are afraid to write 2500-word articles because "the best articles (meaning 'the articles people are most likely to read') are less than 1,000 words long". To which I say: BALDERDASH. A good 2500-word article is more likely to be read all the way than a bad 600-word article. I rarely wrote a 500-word theme/essay when I was in college because my professors never bothered to tell me that I didn't know how to count that short. And yet, many of my classmates complained that their professors stopped reading their essays after the first paragraph.
It's not that I am a great writer — it's that I spend a great deal of time writing, and therefore I have more practice at the art than most people. Or so I tell myself. But when the time came for me to learn how to write a 500-word theme so I could pass a standardized university system test I struggled with the concept and went into the exam really doubting my ability to write something that short which could say anything interesting.
"Don't worry about being interesting," my favorite English professor Don Fay said to me. "I have graded thousands of these exam essays and believe me, the last thing I need to read is an interesting 500-word theme."
In search engine optimization you just have to decide on one thing: Do you want someone to appreciate what you have to say or not? It doesn't matter what you're writing. It could be a 50-word product description or a 10-word product image caption — you may have to make the sale in that little amount of space. And some people are better at writing those short, cryptic phrases than others. I hate writing that kind of content.
Give me 5,000 words or go find someone else to write your 300-word article. A good 5,000-word article should bring in traffic for years to come. A bad one will go unnoticed just as much 2 years from now as yesterday.
The problem is, if you're writing bad LONG copy then cutting it down to bad SHORT copy really won't change much. You'll still suck at writing. That's not to say that everyone who writes great 5,000-word articles can write great 400-word articles. It's an entirely different style. I have, in fact, used word counters to tell me when I've passed a limit and gone back to "tighten up" the writing and get down to the agreed-upon limit.
As Mark Twain, Voltaire, and so many other great writers of the past are supposed to have said, "I apologize for writing such a long article but I didn't have time to write a short one." Frankly, that is one of the few axioms of good writing: that it takes time to write good short copy. So if you're getting paid by the word and you're writing "GOOD" short copy you'd better be charging a lot more than the guy next to you.
The length of your copy matters in ways that the search engines have not yet figured out how to quantify but they manage to punish short copy in a lot of ways. Think of all the Websites that oh-so-cleverly paginated their long articles so they could plant more advertising on their copy. A few of those sites are still getting by with pictorial slide shows but a lot of paginated content has vanished. Why? Google didn't like it.
Was it the pagination or the execution that Google objected to? For many people the pagination and the execution were one and the same thing. They lacked the vision and intuition to make any difference. Hence, their badly executed pagination proved that short copy wasn't a good thing.
The takeaway here is that we should not be concerned with how many words we put on the page but rather with how well we say whatever it is we want to say, even if we're only listing 10-in-1 screwdrivers. I'll grant you that not many people would hang around to read 5,000 words about 10-in-1 screwdrivers but if anyone ever asks me to write that kind of copy you can be damned sure it will be the best 5,000 words I can write about 10-in-1 screwdrivers. Give me enough time, pay me well enough, and I'll whittle that "best 5,000 words" down to 50 — but you may not be able to afford that.
Myth No. 2: Anchor Text No Longer Matters
I was dumbfounded the first time I read this on an SEO Website. I have long since stopped going numb when I see this kind of erroneous "fact".
The search engines have not replaced anchor text in their algorithms. They may not look at it the way they used to but it's obvious to anyone who knows how to use a Google or Bing search box that there are plenty of sites ranking on anchor text.
As Bill Slawski pointed out recently, not all anchor text is the same. Let me be more blunt about it: Just because your sorry excuses for "linking strategies" were filtered by Google doesn't mean Google stopped caring about links.
The Google crusade against "Paid links" and "Low quality links" wasn't meant to signal the end of anchor text — quite the contrary it signals the start of a whole new era in anchor text influence in the SERPs.
Search engine optimization has never been about anchor text anyway. Faux SEO emphasized manipulating SERPs with anchor text but everyone knew (wink, wink) that it was only a matter of time before Google devalued some of the links. Now that Google is better at filtering out manipulative linking than ever before, people have concluded (wrongly) that links no longer matter.
Links still matter — it's just that Google's linky world looks more like the way Google wants it to. Eventually people will figure this out and stop coming up with crap theories to explain the obvious: that as bad links fall away good links are better able to influence the SERPs the way Google wants them to.
The takeaway from all this: Look for a new wave of manipulative linking in 2013. It's already begun but it hasn't quite reached the SEO blogosphere just yet. Once you start seeing the new linking styles you'll tell yourselves that these are great "Google-proof" methods for obtaining "quality links".
Remember those two expressions: "google-proof" and "quality links". People have used them before and they will be resurrected like a bad PageRank sculpting nightmare from "correlation studies".
Myth No. 3: There Is No Long Tail Distribution in Referral Traffic
Okay, Rand Fishkin stunned everyone with his latest analysis of the Web's traffic. Rand "surmised that only ~20% of the referrals that the average website receives comes from the tail of the distribution curve." Structurally this is a sound hypothesis in that it can be tested.
It's the manner of the test that fell flat on its face; but given that Rand did not publish enough detail about the parameters of his test the only flaw I can point to is that Rand did not publish enough detail about the parameters of his test. Once again an unscientific test takes to the airwaves and creates a buzz where no buzz should be heard.
But lest I make it seem like I'm beating up on Rand too much, let me point out that other people "confirmed" Rand's findings by looking at their own non-search referral data. Some folks found a similarly small sampling of traffic-sending referrals.
Now, before I allow myself to be dragged into a pointless unscientific non-search referral graph pissing contest, let me point out that the general unspoken assumption here (both in Rand's post and some of the reactions to it which I read) is that you can read a long tail in non-search referrals much the same way that you can read a long tail in query strings.
In other words, if 1,000 queries send traffic to my Website over a given 30-day period, I can usually expect to see that the majority of those queries place my site at or near the top of the SERP. And I may even get most of my search referral traffic from those low-volume queries.
But the long tail of search is usually defined as the largest portion of active queries, which tend to be longer and lower in search volume than the head and body of the search query corpus. This is due to a power law distribution and regardless of how many non-search sites are sending traffic to your site there is a power law distribution at work there — it just may be a very short distribution.
On the other hand, we measure search query corpi very differently from the way we measure backlink corpi — hence, if you're analyzing non-search referral traffic you have to be more inclusive.
In both data sets there are secondary factors or aspects that change the data sets over time. But in search query analysis we know that once a query dies it rarely comes back to life. That is, many queries reflect current consumer interests and concerns which may be driven by product models, availability, pricing, seasonality, etc. Even in the non-commercial search world there are queries that die off drastically (such as searches fro stories about Michael Jackson's death, the 2000 US Presidential election, and so forth).
Hence, there is an inferred finite limit to the viability of search query data. Most people in the SEO industry never look at 2-year-old query data. If you're pitching keywords at clients or vice presidents you'd better be focused on what people are searching for today and may be searching for in six months.
Every 30 days the value of a set of queries diminishes (in fact, this rate of decay in search query relevance/value is governed by a power law as well). You can pretty much ignore many queries that are more than a year old unless your Website doesn't change its content very much.
Backlinks, on the other hand, don't lose significant value on the 31st day after they last sent traffic to your site. In fact, backlinks can experience Renaissancial awakenings, whereby their framing content suddenly takes off and becomes hyper-relevant again. This is not a guaranteed process (wouldn't that be nice?) — it's just a phenomenon that happens. Old links wake up and send traffic again.
There are all sorts of case studies about this phenomenon and they cover many different time frames. A good case study (that covers relatively short windows of opportunity) is Danny Sullivan's 'Second Chance Tweets' article from September 2011. The estimated lifespan of a Tweet is very short — but you can revive that Tweet with periodic reTweets (Ralph Tegtmeier seems to live on this process — so do some other aggressive marketers I follow).
What is true for Tweets is true for blog posts, directory pages, news stories, and every other kind of content that provides links to your Website. So if you want to test Rand Fishkin's (perfectly valid) hypothesis, how should you do that?
You cannot just look at the last 30 days or even the last 180 days' worth of non-search referral traffic. That's not going to tell you a complete story about the links that send your site traffic. You have to look at ALL the data across all time; and then if you want to get scientific about it you need to start looking at the lifespans of the various links, and whether they were resurrected for any particular reason.
In other words, your long tail of non-search referral traffic cannot be measured in the same way as your long tail of search referral keywords. In fact, what happens with many Websites (both commercial and non-commercial) is that the leading non-search referral sources change over time.
These transitions in non-search referral leaders happen for a variety of reasons. Marketers tend to favor a small number of Websites for "exposure'. A few years ago StumbleUpon was all the rage. Now hardly anyone talks about how much traffic they get from it. I actually still see occasional good Stumbles in my non-search referral data.
Twitter, Google, Facebook, AHrefs, AllTop, and other sites have all been good sources of traffic for SEO Theory this year. They weren't always the leading sources of traffic. There were times when we loved being DUGG, SPHUNN, bookmarked on Delicious, and listed in Dave Harry's weekly SEO article roundup.
The leading sources of non-search referral traffic change over time much more than vanity keywords. If, like me, you're almost 100% passive in your link acquisition you have no control over the ever-changing flow of new traffic sources. Some Websites stay active for years. Some Websites stay active for just a few days or weeks. And then, POOF! The sites vanish, never to return.
And sometimes a site you had forgotten about suddenly comes back to life. It starts sending you traffic again.
Put all this data together and you do see a long tail in your referral traffic. This long tail doesn't operate by the same rules as the long tail of search so if your tests are modeled on that type of analysis they will fail.
Search engine optimization is a very different thing from true "Web marketing". You're comparing apples to oranges and such comparisons invariably lead people to draw wrong conclusions.
You have to use the right parameters for your test — and you have to understand WHY each test requires its own parameters. You have to understand that the environment of the phenomenon you're studying dictates the parameters, not what you already know (or think you know).
UPDATE: See AJ Kohn's comment below and my reply. Non-search referral data requires a different type of analysis from search referral data.
Real Science Asks Hard Questions But …
Once you get past asking the question you have to adhere to rigid methodology for seeking answers to those questions. The scientific method cannot bring insight to every question. For example, you cannot reverse-engineer Google's algorithm no matter how badly you want to. All the attempts to list the "factors" that contribute to Google's SERPs are just silly and foolish — if they are intended or used for real-life search engine optimization strategy.
It's one thing to say, "Yes, Google looks at a lot of factors and here are some examples of factors that have been shared"; it's quite another thing to say, "Well, Factor 117 is assigned 50% of the weight of the algorithm." There is no science in either of these efforts but documenting disclosed or proposed factors at least helps to explain the general process. You don't have to be an algorithm wizard to be good at search engine optimization.
In fact, I have argued for years that focusing on the guidelines will make you a better SEO than guessing at the algorithmic factors. The guidelines are the roadmap to safe search optimization; the algorithmic guesswork is just a painful way to mislead a lot of people with fakery and snake oil.
And there is only so much that search algorithms can do anyway. They still have to scour the Web for content to include in their indexes. So all your favorite SEO tools that do their own data collection mislead you because they have no way to calibrate their data with the search engines' data — leaving you with more incomplete guesswork and faith-based SEO.
Add to that the frequency with which the Web changes (not to mention the search engines' knowledge of the Web) and there is darned little room for science in the execution of the Searchable Web Ecosystem. So all these "studies" that people are sharing as proofs that this or that works better — they're just nonsense. Nothing is proven because a scientific proof requires sufficient facts to trace a logical path from point of origin to conclusion.
Search engine optimization bloggery is mostly about assumptions, undisclosed methodologies, and pretty graphs — smoke and mirrors analysis. Real scientific analysis explains how the data was gathered in such a way that other people can replicate the experiment(s); real scientific analysis explains how the data was organized so that other people can organize their own data in the same way; real scientific analysis shares the data that was captured so that other people can do their own analysis.
It's tedious to do this stuff. And you can't say, "Oh, yeah, it has to be peer-reviewed". One of the dirty little secrets of modern science is that peer-review misses a lot of crap science. It's mostly the reaction to the post-review disclosures that help shape our scientific opinion. Once you publish your experiments and the results you obtained other scientists go out and try to replicate those results — and if they fail to poke your experiment and look for holes in it then they are just as guilty of bad science as you when the truth finally emerges.
So even convergence of opinion and replication of results don't guarantee that something is "right". We may never really know if Einstein's theory of general relativity was right. So far it has withstood all challenges, so it seems pretty darned right — but science continues to probe and test the limits of Einstein's work because science simply cannot accept that that is all there is.
It's the same way with search engine optimization and web marketing. We make many mistakes in our pronouncements but the worst mistake of all is our failure to push back and question the claims that are being made, even if they are only made in agreement with earlier claims. Of such failure are the simplest (and wrongest) of myths born.
|
