Keyword Research #3 – Introducing Thirst™
4 votes, 3.50 avg. tacos (69% full)

Update (Jan 15, 2015): After you’re done reading this post, be sure to check out Han and Nate’s new keyword tool that makes it easy to scrape your competitor’s keywords and do bulk keyword difficulty analysis.

Update: Although PBNs still work, they now have a history of being targeted by Google and therefore may not be the safest option. This is why we now focus on creating online businesses that are independent of SEO traffic.

 

This is a remake of the last of 3 keyword research videos I did for interns that started in the summer of 2013.   I remade it as the way I’m calculating the # of definite wins is different since we’re working with new PMDs, and the way I calculate the number of links required is now based on PA, which is much stronger than LRDs.

 

Notes:

Estimated Earnings based on Oracle – this is based on selling to PPC advertisers and assumes a 20% SERP CTR and 30% of the AW CPC. If the landing page is basically a direct copy of an existing Adwords landing page and you are selling based on conversion, you essentially have the equivalent of a 100% Page CTR. If you compare this to Adsense, then in most cases you’re looking at a higher CPC (depends, you’d look at the contextual CPC), but a page CTR that is in the 3-30% range (depends on the nature of the kw).  I’d say if you want to estimate for Adsense, take 1/5 of the amount, or if you’re really conservative you could take 1/10.  If you want to be exact then go record the contextual CPC, take 68% of it, and assume a page CTR in the 3-30% range based on your experience with the keyword.

 

Thirst = Value/mo per backlink.  It is a RELATIVE(!!!) metric that is used to compare 1 keyword against another when prioritizing either your keyword targeting for potential new sites, or your link building efforts for your existing sites.

I estimate this based on the number of links from a network of decent domains (PA30+) it takes to get to PA30, PA40, etc.

 

Assumptions:

All the following numbers assume that 1 link = 1 link from 1 root domain (the same root domain does not link to a target root domain more than once).

They also assume that links are within posts that are stickied to the homepage and showed in full.  The homepage would vary in quality from the PA30/DA20 range to the PA40/DA30 range (with ZERO spam).  If all your sites are at the lower or higher end of this spectrum, you can adjust accordingly.

 

To get to PA30:

There are many variables, but essentially assuming your landing pages is being link built directly, it would take approximately 5-15 links to get into the mid PA30 range.   Now this changes a lot if you’re link building your homepage as well as several money pages, all of which are in the global navigation.  From experience, if you were to send ~10-30 links split between your home pages and say 5 different subpages, each of those pages will have a PA in the 30s.  This is why it’s now better to target multiple pages on the same domain, you have better use of link juice. Information from veli.services .

 

To get to PA40:

PA is logarithmic, and going up in PA requires exponentially more link juice.  Here you will require more like 30 links to a single target landing page to get into the mid PA40 range.  If link building 5 targets equally (all are in global navigation), you’re likely going to require around 90 links total.

 

The Chart:

The chart is based on link building brand-new PMDs from a PBN that has sites varying from PA30/DA20 to PA50/DA40.

 

Unknowns to Look for:

  • Multiple results from the same root domain
  • PA of 50+
  • Obvious TLD leaning
  • DA >75 (doesn’t count for subdomain)
  • EMD homepage with PA above 40
  • Obvious SERP style lean that you can’t match (shopping usually).

 

You only want to target SERPs that have less than 5 unknowns.  Usually less than 3.   If it’s right on the border (5 unknowns), check the 2nd page and make sure there are at least 8 wins there.  If there are, go with it, if not, pass.

 

Once you have passed the definite win filter, you calculate thirst based on the average PA of the definite wins * the number of links needed to hit that PA.  I am working on reverse-engineering PA to automate this and will have the results posted soon as well as a sweet calculator.  For now though use the estimates I showed above.

 

This is obviously quite a bit of work to do for every single possible target, but I feel it’s important you try it out just so you understand it.  Oracle will eventually calculate thirst automatically, including working with a rank checker to take into account your existing rank and the competition of the sites above you.

 

As usual if you have a question just leave a comment!

 

 

 

 

 

Keyword Research #3 – Introducing Thirst™
4 votes, 3.50 avg. tacos (69% full)
  1. maxime sincerny says:

    Do you add links to get a “legit” links profile?
    Or you just add your 20 High PageRank links?
    What kind of anchor text do you use on your links?

      • I am a little confused by that statement… If 90% of your anchors are generic/brand, how are you targeting your desired keyword? Is keyword targeting now a factor of on-page factors only?

        • Yes, keyword targeting is pretty much all on-page at this point. Make sure that the keyword is in your title and heading tags while also making sure your content is relevant. The anchor should also be relevant to the keyword as much as possible, just not exact match.

  2. As ever Hayden, awesome stuff… I think Oracle is pretty cool.

    I wasn’t aware the change to PA over LRDs, what made you adjust? is it based on the intern site build activities?

    cheers
    Stu

  3. Great stuff Hayden.

    Was wondering whether you have a video tutorial or blog post that takes this a further step to demonstrate your procedure to target PPC advertisers to sell to them directly once you are ranking for your kw, rather than the Adsense model.

    Q How do you decide whether to target affiliate deals or PPC advertisers?
    Q How much do you charge PPC advertisers, is it an agreed CPA?
    Q What payment method / software to monitor this?

    cheers

    • maxime sincerny says:

      I know you asked hayden, but there is my version.

      1- I always prefer affiliate over adsense, more work at first, but after some testing different affiliate you get multiple times the amount of adsense each month
      2- I hate to sell advertising to advertisers myself it is a total mess and you will probably make less then with adsense.

      Only from my experience here.

  4. Thanks for the post Hayden.

    In the past you mentioned you do not put links on the homepages of your PBN sites.

    Have you eventually found it more useful to switch to displaying the posts (with the links) on the homepage after all? Would you recommend building the PBN this way?

    Thanks!

  5. Thanks for the post Hayden.

    In the past you mentioned you do not put links on the homepages of your PBN sites.

    Have you eventually found it more useful to switch to displaying the posts (with the links) on the homepage after all?

    Are you using excerpts or the entire posts on the homepage?

    Thanks!

  6. Hi Hayden,

    Great stuff!

    I think your Thirst metric makes a lot of sense ($ / month / backlink) as long as your PBN can accommodate that number of links. Otherwise, I would suspect your earnings vs #backlinks-made would be very non-linear.

    Some questions:

    1) After you get your page ranking in the top 3ish, do you periodically make more posts to ensure a #1 rank? Does this factor into your metric?

    I guess you could just wait till the ranking drops and iterate the process again…

    2) Just to clarify, would your most conservative earnings estimate for adsense be = (20%*30%*10%)*#Searches*CPC

    or essentially 0.6% * Search*CPC?

    Cheers,

    Nathan

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>