How to Use SEOMoz’s LDA and 10 Things to Use it For
If you don’t know who Ben Hendrickson is yet, you will now. He’s the genius who essentially reverse engineered the search engine ranking algorithm as it applies to content relevancy. Now in English: He created the LDA tool to show us how relevant our content is to our chosen keyword. This is huge news in SEO, especially since LDA can pretty effectively predict the rank of a site for a specific keyword based solely on the content. Just let that soak in for a second.
Briefly, how to use it: open it up in your favorite browser by going to http://www.seomoz.org/labs/lda (bookmark this one, trust me!). Put your desired keyword into the “Query” box. If you have content you’d like to test, copy and paste that bad boy into the “Document” field. Otherwise, put the URL of the site you’re interested in into the “URL” field, and hit compute relevance. Note that this checks out the whole site, so if you want to look at a specific page only, you should paste the content directly into the “Document” field.
Once it’s done thinking, you’ll get a lovely number like 65%, which is much nicer than yesterday’s random cosine value. This basically means the content on the site or the specific content you put in is 65% relevant to the keyword you chose. You can get a lot more out of that number by comparing it to your largest competitor to see how their content stacks up. Drawing conclusions from there is pretty easy.
The LDA has hundreds, probably thousands of applications for anyone who is interested in how the web works, but I picked just 10 to share with you today to get those gears going.
10. Your Site’s Relevance
If your site has better relevance but your competitor is still performing better, consider looking at other known ranking factors, such as back links, site speed, or multimedia.
9. A/B Testing
Yes, I know, we’re not supposed to get lost in A/B testing, so do not go overboard, but hear me out. If you’re copywriting for a site and feel ambitious enough to come up with two copies, pop them into the “Document” field and test their relevancy against the keyword you’re trying to hit.
Version 1:
Version 2:
The tool strips the excerpts down to a set of vocabulary lists. If I had written both excerpts and was wondering which was more likely to rank higher, I’m clearly going to choose Version 2.
Don’t do this for every single page, but a few will give you a good idea if you’re on track or not. It’s a quick and dirty way to get an idea of where your site might rank with the content you’re writing.
8. Vocabulary Lists
If you’re not ambitious enough to write two copies of a page (and really, I can’t blame you there) for A/B testing, you can use the LDA to help you build a vocabulary list. Put your favorite competitor’s site into the LDA for a keyword you like and see what it comes up with. Use that list of words to help guide you as you write the copy for your site.
Using eyes.com:
7. Relevant Keyword Lists
You can also use that list to help you find other relevant keywords and get a feel for what words are considered related to your query. Use a competitor’s site to help you find the words search engines will group together as synonymous or related.
6. Compare Keyword Lists
This one builds on number 7. You can compare the keywords your competitor’s site is targeting and likely ranking for with the ones that you’re targeting. Do you have any that he doesn’t? Does that scoundrel have a market cornered that you should be a part of? Use this to help you figure out how you can improve your site.
5. Find Content Holes
Test your site’s relevancy against the different keywords that you want to rank for. Is your site relevant for “plastic surgery,” “breast implants,” and “liposuction,” but not “mommy makeovers?” Looks like you should add some content about mommy makeovers!
Since 34% leaves a lot of room for improvement, a personal injury lawyer may find it worthwhile to add some dog bite content!
4. Content Ideas
If you’re just generally looking for ideas on what to write about, you can use the LDA to help you with that, too! You can use your site or a competitor’s to get that giant word list. Skim through and see what inspires you!
3. Find Where to Link to Your New Content
You’ve done the content research, A/B tested it and now are ready to add it to your site. You can quickly see which pages you should link to it from by looking through the word list. Or, better yet, you can compare how relevant two individual keywords are by putting the second keyword into the “Document” field.
If you’re a bariatric surgeon who wants to add a page on heart burn surgery, you can see that your pages on weight loss surgery are relevant, but those on lap band surgery are not. Use the LDA as your guide to making sure your site seems tied together properly.
Option 1: Your page about weight loss surgery
Option 2: Your page about LAP-BAND surgery
The relatedness of heart burn surgery to weight loss surgery is much higher than to LAP-BAND, even though it’s a type of weight loss surgery itself! The weight loss surgery page is the clear winner.
2. Localization Relevance
How well are you targeting your pet localization? Put it into the “Query” field and compare your relevance to that of your favorite competitor’s. If they’re doing better than you are, investigate their site and see how you can target your favorite place better.
1. Compare your keywords head to head
Using the same method as number 3, above, you can see how relevant your keywords are to your keywords. Making sure your keywords are related keeps your site focused and relevant, which the search engines love.
For now, the LDA is graciously offered for free through SEOMoz labs, but I don’t expect that to last for very long. We have a pro account here, and it is absolutely worth it. The SEO tools available through SEOMoz are incredibly helpful and allow us to do better work for our clients.
What other uses for the LDA have you come up with?
9/6/10 edit: For more detailed information about LDA and the LDA Tool, please read Rand Fishkin’s excellent post, Latent Dirichlet Allocation (LDA) and Google’s Rankings are Remarkably Well Correlated.
September 3rd, 2010 at 9:16 am
“He’s the genius who essentially reverse engineered the search engine ranking algorithm as it applies to content relevancy.”
The Latent Dirichlet Allocation model has been around since 2002. Wikipedia sites the first initial scientific paper: Blei, David M.; Ng, Andrew Y.; Jordan, Michael I (January 2003). “Latent Dirichlet allocation”. Journal of Machine Learning Research 3: pp. 993–1022.
I haven’t seen in IR a tool that applies LDA to the web, so Ben definitely deserves kudos for that. To be clear though, the algorithm, technique and process have been around. You can academically find a lot of scholarly information if you merely do a search for “Latent Dirichlet Allocation”.
September 3rd, 2010 at 9:37 am
Oh, definitely. The concept is discussed in some depth in The Art of SEO, at least in the vein of “search engines understand context”, and Google itself has said for years that webmasters should concentrate most on providing good, relevant content. I think the genius of this is that no one knew to what extent it was factoring into the algo. I’m not sure we can now nail down a number, but it’s clear from Ben’s research that it’s a much larger part than previously thought, and the fact that he managed to work it into a fairly accurate and actionable tool is rather awesome.
September 5th, 2010 at 12:15 am
Rebecca, enjoyed your summary! Wish we would have met in-person at the Mozinar.
My favorite take away from your post is what we need to remember:
“If your site has better relevance but your competitor is still performing better, consider looking at other known ranking factors, such as back links, site speed, or multimedia.”
The factors are surely more complex than we realize, and we can take these relevancy scores as one signal. As you allude, so many other factors must be considered.
Well done!
September 5th, 2010 at 7:48 pm
It’s Patti’s summary, not mine, I’m just pimping it
But you’re right – this is still just ONE ranking signal out of many. Eyes.com has a much smaller following and link profile than AllAboutVision.com right now, so even though the former is the runaway winner for relevancy on Lasik surgery, the latter continues to flog it in the rankings. Relevancy clearly can be overpowered by other signals. Thanks for stopping by!
September 5th, 2010 at 8:59 pm
I come from a more traditional, scientific IR background, and so I’m always a bit skeptical of a tool that’s marketed so heavily towards the SEO community that claims to be from the scientific IR world.
I’m honestly looking forward to an SEOMoz LDA Tool 2.0. I really like the concept of the tool, but I want to see an aggregator. If you enter in a web URL, the tool suggests what topics it believes your site to be related to, showing a % relevant for each. This would be great in identifying if your content-based SEO work was going anywhere, as well as identifying any long tail markets you may be able to position for.
PS: Thanks @LehmanPhoto for the @ reply, I actually didn’t notice it until yesterday! Sadly, I believe that linking strategies are still huge in terms of ranking, but these tools, as you have shown are able to provide an edge if you’re looking for that small 5% to push you above the rest. I’ll definitely be following you, and your blog posts in the future! Thank you very much for the legwork.
September 10th, 2010 at 5:04 am
[...] Links: It’s impossible for me not to talk about SEOmoz’s new LDA tool. Heck, we did our own post about it here after it was announced at the Mozinar in Seattle, and another one here after we realized what a [...]
October 7th, 2010 at 8:12 pm
I have been seeing the LDA idea kicked around for some time now and even witnesses the disaster that went on in the comments when the SEOmoz team was cleaning up their correlation data mistake. I still see value in this tool and I think that item #3 above is crazy SMART. When it comes to internal linking, it is very easy to just look for the page that has the most linkjuice and add an anchor text link there. Now this is a more analytics driven way to do the same thing that makes a lot more sense. Thanks for posting. I am bookmarking this for future reference.
November 30th, 2011 at 11:54 pm
Wonderful beat ! I wish to apprentice while you amend your site, how can i subscribe for a weblog website? The account aided me a appropriate deal. I were a little bit acquainted of this your broadcast offered vibrant clear concept
December 10th, 2011 at 2:07 pm
I am impressed, I have to say. Really rarely do I see a blog that’s both educational and entertaining, and let me tell you, you have hit the nail on the head. Your blog is outstanding; the issue is something that not a lot of people are speaking intelligently about. I am very happy that I stumbled across this in my search for something relating to it.
December 11th, 2011 at 8:16 pm
Ohhh, i’ve been looking for the perfect flatware (i’m really picky!) and i think these might be it! My top four are
December 12th, 2011 at 3:55 am
pharmacy one rx…
[...]White Hat, Black Belt » Blog Archive » How to Use SEOMoz’s LDA and 10 Things to Use it For[...]…