Beware Of SEO Truthiness

When SEO started, many people routinely used black-box testing to try any figure out what pages the search engines rewarded.

Black box testing is terminology used in IT. It’s a style of testing that doesn’t assume knowledge of the internal workings of a machine or computer program. Rather, you can only test how the system responds to inputs.

So, for many years, SEO was about trying things out and watching how the search engine responded. If rankings went up, SEOs assumed correlation meant causation, so they did a lot more of whatever it was they thought was responsible for the boost. If the trick was repeatable, they could draw some firmer conclusions about causation, at least until the search engine introduced some new algorithmic code and sent everyone back to their black-box testing again.

Well, it sent some people back to testing. Some SEO’s don’t do much, if any, testing of their own, and so rely on the strategies articulated by other people. As a result, the SEO echo chamber can be a pretty misleading place as “truthiness” - and a lot of false information - gets repeated far and wide, until it’s considered gospel. One example of truthiness is that paid placement will hurt you. Well, it may do, but not having it may hurt you more, because it all really…..depends.

Another problem is that SEO testing can seldom be conclusive, because you can’t be sure of the state of the thing you’re testing. The thing you're testing may not be constant. For example, you throw up some more links, and your rankings rise, but the rise could be due to other factors, such as a new engagement algorithm that Google implemented in the middle of your testing, you just didn’t know about it.

It used to be a lot easier to conduct this testing. Updates were periodic. Up until that point, you could reasonably assume the algorithms were static, so cause and effect were more obvious than they are today. Danny Sullivan gave a good overview of search history at Moz earlier in the year:

That history shows why SEO testing is getting harder. There are a lot more variables to isolate that there used to be. The search engines have also been clever. A good way to thwart SEO black box testing is to keep moving the target. Continuously roll out code changes and don’t tell people you’re doing it. Or send people on a wild goose chase by arm-waving about a subtle code change made over here, when the real change has been made over there.

That’s the state of play in 2013.

However….(Ranting Time :)

Some SEO punditry is bordering on the ridiculous!

I’m not going to link to one particular article I’ve seen recently, as, ironically, that would mean rewarding them for spreading FUD. Also, calling out people isn't really the point. Suffice to say, the advice was about specifics, such as how many links you can “safely” get from one type of site, that sort of thing....

The problem comes when we can easily find evidence to the contrary. In this case, a quick look through the SERPs and you'll find evidence of top ranking sites that have more than X links from Site Type Y, so this suggests….what? Perhaps these sites are being “unsafe”, whatever that means. A lot of SEO punditry is well meaning, and often a rewording of Google's official recommendations, but can lead people up the garden path if evidence in the wild suggests otherwise.

If one term defined SEO in 2013, it is surely “link paranoia”.

What's Happening In The Wild

When it comes to what actually works, there are few hard and fast rules regarding links. Look at the backlink profiles for top ranked sites across various categories and you’ll see one thing that is constant....

Nothing is constant.

Some sites have links coming from obviously automated campaigns, and it seemingly doesn’t affect their rankings. Other sites have credible link patterns, and rank nowhere. What counts? What doesn’t? What other factors are in play? We can only really get a better picture by asking questions.

Google allegedly took out a few major link networks over the weekend. Anglo Rank came in for special mention from Matt Cutts.

So, why are Google making a point of taking out link networks if link networks don’t work? Well, it’s because link networks work. How do we know? Look at the back link profiles in any SERP area where there is a lot of money to be made, and the area isn’t overly corporate i.e. not dominated by major brands, and it won’t be long before you spot aggressive link networks, and few "legitimate" links, in the backlink profiles.

Sure, you wouldn't want aggressive link networks pointing at brand sites, as there are better approaches brand sites can take when it comes to digital marketing, but such evidence makes a mockery of the tips some people are freely handing out. Are such tips the result of conjecture, repeating Google's recommendations, or actual testing in the wild? Either the link networks work, or they don’t work but don’t affect rankings, or these sites shouldn't be ranking.

There’s a good reason some of those tips are free, I guess.

Risk Management

Really, it’s a question of risk.

Could these sites get hit eventually? Maybe. However, those using a “disposable domain” approach will do anything that works as far as linking goes, as their main risk is not being ranked. Being penalised is an occupational hazard, not game-over. These sites will continue so long as Google's algorithmic treatment rewards them with higher ranking.

If your domain is crucial to your brand, then you might choose to stay away from SEO entirely, depending on how you define “SEO”. A lot of digital marketing isn’t really SEO in the traditional sense i.e. optimizing hard against an algorithm in order to gain higher rankings, a lot of digital marketing is based on optimization for people, treating SEO as a side benefit. There’s nothing wrong with this, of course, and it’s a great approach for many sites, and something we advocate. Most sites end up somewhere along that continuum, but no matter where you are on that scale, there’s always a marketing risk to be managed, with perhaps "non-performance" being a risk that is often glossed over.

So, if there's a take-away, it's this: check out what actually happens in the wild, and then evaluate your risk before emulating it. When pundits suggest a rule, check to see if you can spot times it appears to work, and perhaps more interestingly, when it doesn't. It's in those areas of personal inquiry and testing where gems of SEO insight are found.

SEO has always been a mix of art and science. You can test, but only so far. The art part is dealing with the unknown past the testing point. Performing that art well is to know how to pick truthiness from reality.

And that takes experience.

But mainly a little fact checking :)