SEO folks are trend victims. Someone presents a new acronym, an idea, or a tool and we all pile on. Then we chase those nibbles of knowledge, good and bad alike, without thinking. And we land in deep pits lined with sharp, nasty sticks.
These ideas must go:
1: TFIDF Score Predicts Rankings
A higher TFIDF score means you’re more relevant, right? Wrong.
I remember the first time I heard about TFIDF. I latched onto it like a tick at a blood bank and started running around the office yelling “TFIDF ALL THE THINGS IT’S THE BEST THING EVER OMG WE’LL GET ALL THE RANKINGS.”
But TFIDF doesn’t measure relevance across the internet. It measures importance across a collection of documents. If you spell it out, it’s easier to figure out:
TFIDF = Term Frequency, Inverse Document Frequency
Term Frequency means “the number of times a phrase appears in this document.”
Inverse Document Frequency means “the uniqueness of this word among the document collection.”
If the target phrase occurs more often on your site, then the TFIDF score for that word on a single page may drop. TFIDF shows the importance of the phrase on a particular page. More sitewide mentions may reduce TFIDF score. As document frequency goes up, inverse document frequency goes down, and TFIDF score drops.
It does not show relevance of a page for a phrase. Use it alongside other algorithms, and it helps you filter out words you use a lot, like “crap.”
Again: It does not show relevance. A higher TFIDF score doesn’t mean you’re going to rank higher. Don’t fall into that pit.
Use TFIDF to measure the importance of a phrase on one page on your site. Use it as one signal, and use it very carefully.
2: Google Penguin is Gone
Barry Schwartz pointed out in this article that Google Penguin no longer penalizes for bad links. Instead, it devalues them.
Yaaayyy!!! Penguin is gone! Salvation! Start buying footer links on www.valuablelinks-for-seo-stuff-and-poo.com again!
Not so fast. Compare the effects:
Before Penguin 4.0: Get crap links. Move up. Get in trouble. Your rankings tank. You pull the pin in your head and explode.
After Penguin 4.0: Get crap links. Move up. Get links devalued. Your rankings tank. You pull the pin in your head and explode.
Link abuse still has the same effect: Lots of wasted effort. An SEO strategy that relies on over-optimized link spam. Lost rankings. Increased dependence on chocolate and whiskey.
If I catch you telling a client they can start buying links because Penguin is dead, I’ll bury a hatchet in your medulla. It’s irresponsible advice.
3: “Faster than my competitors” is Enough
How many more times does Google have to scream and yell about site speed before someone understands it’s about faster, not fast enough? Please, someone, tell me. Please.
If I have to trim my fingernails by the time your site loads, that’s bad, no matter how slow your competitors are.
If your site loads so fast my pupils dilate, it can still get faster.
Stop asking “how fast should it be.”
It should be faster because page speed matters.
Then the web team loads the pages in Google Cache and says “See, all the content’s there. We’re fine!”
If you want to test, pick three or four phrases from your site. Put them in quotes and do a Google site search. Does your site appear as expected? Then you might be OK.
A high authority site has this product but doesn’t rank for it:
The ideal solution, though, is to render content server-side. Use frameworks to drive interaction, not basic content delivery.
If you must use a framework to improve speed (doubtful), provide device-dependent interactivity (OK, maybe), or deliver single-page experiences (OK, but why?) then understand DOM rendering. Indexing depends on when content appears in the critical rendering path.
Google/Bing have a love/hate relationship with frameworks. Use with caution, learn the nuances, and test.
5: More Blog Posts = Better
I have no words.
OK, maybe I have a few.
Broccoli is good for you. But if you eat 500 pounds of rotten broccoli, you’ll still float away, and then perish in a flaming ball of methane when struck by a passing satellite.
Blog posts are like broccoli.
Get it? No? OK. Then I have no words.
6: Rel=Canonical is Canonicalization
Canonicalization is what you need.
Rel=canonical is what you do when you can’t achieve perfect canonicalization.
Canonicalization means ensuring that all things have a single, standard, canonical form. In SEO terms, that means no duplicate content.
Rel=canonical is a spit and bailing wire fix for impossible duplication problems. Rel canonical is a mixed bag at best.
Fix duplication first. Use rel=canonical to reduce duplication only as a last resort.
7: I’m Going to Tone It Down
Some folks think that, now that Portent has a parent company, I’m going to tone down my blog posts.