The money has been spent, what now? 5 steps to iterate content development for results: Research, Community, Media and Latent Dirichlet Allocation (LDA) … and interns

Most Search Engine Marketers (SEMs) will have to step in on already running projects and deliver results.

I have built a process to take a page of content that already exists and improve it for users and for search engines. This process was created for working with interns, so that they will have a framework to follow to improve website content.

The high level process for iterating content

  1. Understand our constraints
  2. Undertake Quantitative Research via tools
  3. Undertake Qualitative Research via direct user interaction
  4. Add Rich media to make it an educational and engaging experience
  5. Use SEOmoz’s Latent Dirichlet Allocation (LDA) tool to help us write measurably more optimised content

I am going to us a case study of a client with luxury hotels in Croatia to show how this can be applied to a real life example.

The clients only mentioned one of their big selling points, scuba diving holidays, in passing on their website. The Hotel Dubrovnik Palace has a full PADI diving school attached to it and a crystal clear diving spot right in front of the hotel. Their website currently focuses on the end of the buying cycle; getting people to book a hotel room right away. They don’t use informational or inspirational content to target people earlier in the buying cycle, when people are looking for information on where they want to go on holiday.

If you, like me, are stuck behind a desk in a cold, rainy part of the world, I apologise for the images you are about to see.

Figure 1 – The shark is this big and it eats SEOs

THE ORIGINAL

The page below was their original activity page to promote a diving holiday in Dubrovnik. Diving in Dubrovnik is amazing but this short bit of text is not going to convince an American diver to swap their Cozumel dive holiday for one in Dubrovnik, Croatia.

Figure 2 – Original short page of text called Scuba diving
  • It doesn’t drive traffic for the competitive terms they need and misses usability and conversion elements
  • There are few of the specific details that will matter to divers, things like price of tank refills or substrate type
  • It doesn’t keep the conversation going by actively driving people to Facebook or Twitter
  • It doesn’t inspire or educate, with images and video. These are major influencers of getting people to share and link to you

So let’s dive into how I could give this problem to an intern to solve.

Step 1 – Understand our constraints

The client approached us after they had already had their website built and spent their budget. This will be familiar to SEMs; you have to create value with few resources. In this case, we have stuck to their original page structure to minimise the work their developers need to do to implement improvements.

When working with interns, clearly setting these boundaries early in the process means I am more likely to get appropriate content back from them.

Step 2 – Undertake Quantitative Research via tools

These are the tools I find useful and examples of how I get the most out of them. There are a number of alternatives to each tool, but I stick to a core set that I understand and am familiar with so that I can explain them to interns and get a consistent output.

Google keyword tool

Using “Dive Croatia” and “Diving Dubrovnik” as my starting points, I set the Google’s keyword tool to use language:English and Locations:All

Figure 3 – Google keyword tool results for diving keywords

Very quickly I can pick up new keywords to expand my content with, like “scuba” and “snorkelling” as well as conversion related words like “dive shop”, “dive center” and “diving equipment”. Luckily the hotel has a PADI dive centre on its lower floor so it’s a perfect match to search queries. I hope to be able to rank for all these “halo” diving words quite easily.

You can see from the competition column that there is not a lot of PPC activity around these terms, meaning I can cheaply drive Paid Search towards this page.

Google Sets

Figure 4 – Google sets result for diving keywords

Google Sets automatically creates sets of items from a few examples.

I use this tool as a proxy for understanding what term and keywords Google thinks are related. For my input of dive, diving, scuba and scuba diving I get back the following:

As I want to add more text to the page, I also want to make sure the text I add is strongly related to my main keywords.

From this output will make sure to include words like “fish” and “wreck” and “island”, to attempt to capitalise on what Google is telling me is its contextual understanding of my main diving terms.

Step 3 – Undertake Qualitative research via direct user interaction

Next we turn to divers themselves to find out what they want in a page. I searched Google for diving forums and found one that seemed active. I signed up and asked them what information they would like to see in a diving page.

It turns out real divers wanted more and very specific information. Some examples out of a long list:

  • Substrate
  • Recommended certification
  • Perhaps recommendations for the photographer, e.g macro, wide, ambient etc
  • Nitrox availability
  • Trimix availability
  • Rebreather support (i.e. bottles, sofnalime etc)
  • Any diving regulations specific to the local area (e.g. max depth, no solo diving, Open Water to be accompanied by Instructor etc)

Although I grew up on the coast, there is no way I would have known this specific diving data without asking the experts.

I only thought of doing this after the page went live, but I will be doing this as part of the initial process going forward. This is one of those things that seem so obvious when you do it, and you kick yourself that you didn’t do it before. This is the reason for creating an iterative process; we will improve this page over time, based on data.

Step 4 – Add Rich media to make it an educational, engaging and linkworthy experience

There are two main reasons for using multiple types of media on the page.

USERS

From a conversion point of view, users are hitting this page are early in the buying cycle; trying to make decisions rather than buying straight away. Adding data and utilising rich media allows me to educate and inspire users, rather than immediately driving a hard sell.

I am making sure I can link content to social sharing, to build audiences outside of search. This is one of the main drivers to adding Call to Action to drive people to Facebook and Twitter, so that we are not dependant on search for all our traffic.

Video 1 – Evening cave dive in Dubrovnik

I personally found this video on youtube to be inspirational and a little unusual, so we have run with it.

SEARCH ENGINES

With Google’s recent Panda updates negatively affecting low quality, generic content, I am writing content that provides a lot of specific, useful content, utilising multiple media types to make sure that content is more likely to be counted as unique by their algorithms.

Using multiple media types also allows us to rank in different search verticals – organic, image and video search.

Step 5 – Use SEOmoz’s Latent Dirichlet Allocation (LDA) tool to help write measurably more optimised content

Seomoz’s Latent Dirichlet Allocation (LDA) tool allows us to make measurably better content and gives us a focused process of content development. LDA is a form of topic modelling whereby search engines can pick up the main keyword focus of a page by evaluating the semantic connectedness of other keywords on the page.

We input our page text and then check for relevance against a specific keywords. These would be keywords we are interested in ranking for organically, using for PPC or believe are closely related to our main keyword topic, in order to add relevant, supporting contextual keywords.

Figure 5 – Actual LDA scores

You can see in my first pass I had a low score for the important term “Scuba diving”. I reworked the text and improved the score on the second pass. “Scuba diving” is an important term not just from a user point of view but from a search point of view, in setting the page context via keyword usage.

PS: Need a LDA review? Catch up on LDA with Rand Fishkin’s quick LDA overview with nice pictures

Results

So what the point of doing all this work if you don’t get results? Let’s have a look at the page we made and how it has performed for us so far.

Figure 6 – The current iteration of the scuba diving page

See the page here http://www.alh.hr/en/dubrovnik-scuba-diving/

In the last 20 days in Google.com, for the term “Diving Croatia”, we entered at 221 and have worked up to 129. For the term “diving Dubrovnik” we entered at 37 and have worked up to 30. We have now had visits for 25 different diving terms

Creating a plan for content iteration means that large amounts of improvement work can be carried out by non-specialists. We get a lot of great interns at ILN and developing this process with them means I get an outcome much closer to what I need than if I just gave them a topic and said “get on with it”

Using these 5 steps, I can focus their writing and create measurably better content

– Stephen Croome

This is a guest post, entered in the 2011 Unbounce Conversion Fest Blogging Contest. All opinions are those of the author.

Stephen Croome is interested in getting companies to re-engineer their products to market themselves. He works in-house as Search and Social Manager at Illustrated London News or mucks about with marketing strategy at firstconversion.com. You can follow his chaotic attempt to be a proper SEO on twitter @firstconversion at your own peril.

Comments

  1. Yomar Lopez says:

    MAJOR kudos for mentioning Google Sets!

    I was just mentioning earlier that the “fuzzy logic” search engines use sometimes is interesting, which would explain why some. ermmm.. suspect things show up in SERPs sometimes.

    LDA combined with Google sets can help you cement authority on specific subject matter without spamming. It keeps your content looking natural so you can focus more on humans than robots. It’s a good thing that keyword density is not as relevant now because we get to see more beautiful, user-friendly sites like the one you displayed here.

    Sadly, I think LDA will be something only hardcore SEO’s and Inbound Marketers will adopt, mainly because the name seems purposely coined to suit us data nerds. That’s why it’s refreshing to see you use the jargon but then break down what that means. We in the service industry, especially B2B, tend to assume that clients speak the same language. We mean well and assuming they are as intelligent as us, and they usually are, but that does not guarantee we talk the same talk.

    Now, in the research part of all this, before you tweak, (re)iterate, or repurpose the target content, I like to do about the same stuff you do as well. One thing I notice is that, often, it helps me to look at how many web sites match select keywords. This comes in more handy than the competition field in Google’s data. I mean, if there are tons of web sites and virtually no advertiser competition, you have to wonder why that is.. Maybe that market is saturated?

    On the flip side, you may have minimal competition and low saturation.. And that’s when you can go in and become an authority on a “niche” or specific topic.

    You also mentioned that you had launched the site without doing proper research. Sometimes, that’s the best way to do it. You launch a site that is professional-looking but may not be optimized in any way. Then you see what works and what doesn’t, check the data, and target your weak spots as you did in this case study.

    The first page was not terrible but it certainly was not compelling at a glance.. The end result definitely grabs ya. Good stuff, man!

    BTW, I LOLed at the SEO shark bit. ;o)

  2. stephen says:

    Thanks for the comment :) I think there is so much to discover within LDA by going outside its nerdy heritage and trying to find commercial uses for it.

    The tool is certainly easy enough for people new to SEO to get to grips with. I wonder if anyone else uses this type of process? or what they use instead?

  3. Stephen, thanks a lot for sharing! I just finished a project on content optimization. I wish you could read this article sooner :). Great insights and I love the step by step instructions. Will definitely try this structured approach next time! BTW, thx for bringing up the Google Sets!

  4. Zoe says:

    As an intern myself I found the step by step guide very helpful and hope to put it into practice soon!

  5. Michael says:

    Absolutely quality piece thanks for writing this. Really illustrates both the art and the science of SEO

  6. stephen says:

    Cheers Alex.

    What did you use for your project?

    When I was trying to explain this for the first time I realized I didn’t actually have a process, I just did what I felt like and hoped experience would pay off

  7. frank says:

    Great article and full of useful tips. Do you do any freelance work ? :)

  8. stephen says:

    Rand just mentioned Virante’s tool http://www.virante.com/seo-tools/lda-content-optimizer Its based on SEOmoz’s LDA tool but with a few nice bells and whistles

    “This tool determines your SEOMoz LDA score for a particular URL and keyword. The tool then spiders your page and top ranking pages for that keyword to determine Colocated terms that you have not included in your page. The tool then re-checks your LDA score if you were to include those keywords on your page to determine the potential increase in relevancy.”

    Going to be building this in!

  9. Ryan Kelly says:

    I really like your approach here, Stephen. I’m wondering about the Google Sets tool. Wouldn’t it be better to use Google Suggest in this case to get a better feel for perhaps the combinations of words Google would like to see on a page? I’ve always theorized that if I could include some of the Google Suggest words on the same page as the main term(s), I might get a “leg up” on other pages since I’m including content and keywords Google already thinks are related.

  10. Gabriel Goldenberg says:

    A man who knows his stuff, sharing knowledge… though I’m not crazy about the LDA reference which I don’t feel strongly about.

    • stephen says:

      Cheers Gab :)

      I was hoping to get called on the LDA stuff. Its normally a divisive area within the SEO community!

      Its definitely not the main area of the process, but I think it can help non-SEOs write better content for SEO, as well as being a bit of help in the creative writing journey

  11. Jessica says:

    Great article! Good content is King and thanks for being systematic in getting good content. Interns appreciate clear instructions rather than, “Write up stuff for this site.”

  12. Great job with the new design! You really added value to the website for visitors. I would say you built a near perfect website for your customer. Now you need to do some link building to increase their rankings even more.
    Also… where is the blog? And the social media buttons?

    • stephen says:

      Hi Brian Thanks for the reply – you have nailed the one area that I wasn’t completely happy with on this page

      One of the main challenges with this piece, is that we had to fit it within the main website structure and design. There are Social Media buttons on the page, but not really in a useful way.

      I prefer to only use SM buttons when I have a) a call to action and b) incentivisation – otherwise they just take up space and get very little use

      I do add calls to action to follow on twitter and facebook in the actual text at the bottom of the post – this is because i feel if you’ve read the whole thing, you are more likely to take action with a direct request from the writer, than if you merely scanned the page and noted there was a facebook button

      Cheers
      S

  13. Ben Francis says:

    Absolutely loved this post Stephen, the page is now really interesting, I would imagine the conversion rates will soar compared to what it was before.
    So many times the content is seen as an afterthought in ecommerce, it is often a real job to persuade clients it is perhaps the most important part of the site.

  14. [...] guest post, The money has been spent, what now? 5 steps to iterate content development for results: Research, C…, offers a really nice process for doing some bang-up on-site SEO in the area of quality content.  [...]

  15. stephen says:

    Now up to position 9 in Google.com for key diving terms and starting to get a lot of halo term traffic :) A few links have really moved it along

  16. [...] is my guide on taking mediocre content and making it great content and an overview from our work on Panda on how Google might interpret content quality [...]

  17. [...] is my guide on taking mediocre content and making it great content and an overview from our work on Panda on how Google might interpret content quality [...]

  18. Raja Habib says:

    Great information! And very nicely and clearly presented as well. You have made it convenient for non-seo people too. But I have one question here. Is Google keyword tool good for initial research, for organic SEO?

  19. [...] is my guide on taking mediocre content and making it great content and an overview from our work on Panda on how Google might interpret content quality [...]

  20. [...] is my guide on taking mediocre content and making it great content and an overview from our work on Panda on how Google might interpret content quality [...]

  21. This is a very good article. Planning out content in such a way will definitely improve site rankings. And for every Panda update your site will improve. I will start planning my articles like this as well to improve the quality of my site. Thanks alot of sharing your experiences.

  22. [...] is my guide on taking mediocre content and making it great content and an overview from our work on Panda on how Google might interpret content quality [...]