Slightly Shady SEO | Blackhat is Back, Baby.

Ok. So longtail traffic is something frequently undervalued. And not without reason. Take a look at any adwords tutorial you’d like; they all mention the fact that you should avoid “broad” match as much as you possibly can. This is because longtail searches are much much harder to convert into sales(for random longtails/blackhat longtails). Making a landing page or site relevant to every single search someone could possibly come up with is nearly impossible. Or is it?

Today we will explore dynamic pages/dynamic redirection, and how to use it to convert clicks into sales by understanding the purpose of the user’s query as Google was apparently unable to do. It’s something I’ve used on my blackhat sites for quite a long time(in a slightly more limited form that will be discussed here) and has elevated my conversion ratio on these blackhat “longtail” searches to nearly that of it’s short tailed, whitehat brethren. Note: If the tactics discussed here are not applied delicately, it could qualify as cloaking. For whitehat sites, be wary of this and design accordingly.

The Concept
We are going to use “web hosting” as an example for the rest of this entry. We will be acting like we run a web hosting review/selector website, that takes in the user’s data to find the best web hosting plan. Web hosting is a lovely longtail niche, because it has high payouts, and lots of variety to the searches. Our goal is to make it so the page that they hit, regardless of which page it is, will show them the optimal results.
We’re going to cover a basic and advanced version of this. Advanced is first, basic is at the end. If programming and data structures are not your forte, just read the basic version

But XMCP, isn’t this Cloaking?
In a way, yes. Since the search engine doesn’t come in from a search page, and hence has no referrer, they will be seeing a different page. This could lead to problems. However, the whole point behind their ban on cloaking is to ensure the user gets to the most relevant page possible. Ideally, we will be SHOWING the user the most relevant page possible. So complaints should be low. Heh. Cloaking to make better search results? Amazingggg.

The Problem With the Longtail
While the longtails are a great traffic source, the user is normally looking for somethig super specific. Like overly specific. So the chances of them hitting the right page and buying are quite small. By trying to turn their queries into a computer parsable search, we can remedy this situation.

Grouping your Offer Modifiers
Ok. So when someone searches for webhosting, what do they search for in the longtail? Hard drive space, bandwidth, options they want, and adjectives like “cheap”, “free”,”discount”,”inexpensive”, etc. So we’re going to have to group these up into categories that will dictate how we organize our potential offers.
The way I’m going to format this is {} signifies a group of words, and “,” indicates a different option. So
{Hello, Hi} {How are you, Fellow} would match “Hello How are you”, “Hello Fellow, “Hi How are you”, and “Hi Fellow”.
# means any number matches. {“”,” “} means there may or may not be a space. I put quotes in if the presence of a space matters in the keyword.
Alright. So here are the categories as I see them. These will later become our table columns.
Memory: {#}
Hard Drive Space: {#}
Cheap_Modifiers: {cheap, discount, free}
Type_Modifiers: {dedicated, vps, virtual, shared, web}
Bandwidth: {#, unlimited}
Packages: {mysql, catch-all, catch all, email, e-mail, php, postgres, sql, frontpage, ssl, mod_rewrite, curl}

Ok. So classify each host/package according to that criteria in a mysql table. We’re going to have to search it later, and the more you have the better, so ensure it’s properly organized.

Picking Apart The Query
Ok. So load the google query from the referrer when the user comes in, so you have their keywords stored in a variable. This is where the fun begins. Remember those patterns we set up there? There’s 2 different classes for this.
The first class is the standalone type. Words like “cheap”, “discount”, etc that mean what they mean, and have nothing else. But we have multiple things that match {#} (hard drive space, memory, etc). So we need yet more mapping.
Memory: {#}{“”,” “}{gb, gigs, gigabytes} {ram, memory}
Hard Drive Space: {#}{“”,” “}{gb, gigs, gigabytes}{” of “,”",” “}{hard drive, hd, storage, space}
Bandwidth: {#,unlimited}{ {“”, ” “},{gb, gig, gigabytes,”"},{” of “,”",” “} }{bandwidth}
Using more advanced queries like that we can figure out which category the number is supposed to be relevant to. In addition, we can use logic. Chances are if the number is >5gb, it’s storage. Less than 5gb, it’s memory. Little tricks like that.

Querying the Database
So by searching for our patterns in the text of the query, we can greatly reduce what it is they’re looking for. What was once a mess of every VPS/Dedicated Server/Shared Hosting plan we could find, can be greatly limited.
Like for example. Let’s say they searched for “where can I find cheap shared hosting with unlimited bandwidth and 50gb of hard drive space” – A super specific, longtail query. We can now only display for them exactly what they want.

1. Cheap Hosting (SELECT * from plans WHERE monthly_price<5… //matched the keyword “cheap”
2. Unlimited Bandwidth …AND bandwidth>=1000 //matched our bandwidth pattern. Plus most people who search including “bandwidth” want a lot of it. So it’s safe to assume there’s a lot needed in our query.
   
3. 50 GB Hard Drive …AND harddrive>=50 //matched our hard drive pattern
4. Shared Hosting …AND type=’shared’) //in due to the presence of the keyword “shared”
   

And that should knock it down to a few relevant results.
To Handle any Logical Errors or Ordering Errors, make sure to show them forms to refine/change the percieved search

The Basic Version:
Since I tend to stick to multiple tiny niches, I find myself doing this simple version much more frequently than the above complicated one.
It’s pretty simple. You load up a product catalog(yay scraping) of the company, eliminate all stop words so all that’s left are the product related keywords.

So Acme Co. Sells blue widgets, red widgets, green widgets, and old man daily pill cases. Their website is www.acme.com. And they give product numbers for their navigation.

We’d create a text file like the following:
blue::http://www.acme.com/products?id=1&aff=9945
red::http://www.acme.com/products?id=2&aff=9945
green::http://www.acme.com/products?id=3&aff=9945
pill::http://www.acme.com/products?id=4&aff=9945

Then we search through the query for each of the keywords that appear before the “::”. When it’s found, we redirect them over to the proper page(after the ::)

Livin Off The Longtail,
-XMCP

Posted on Monday, March 17th, 2008 at 11:21 am in the category:Keywords, affiliate marketing, automation, cloaking. Comment RSS 2.0 feed. Comment it , or trackback this post.