Huge Project need help with analysis

12 replies
Working on a content marketing project.

Changing 5 variables across 1000 urls
  1. Modifying h1's etc
  2. Content quality improvement
  3. Add image
  4. Add video
  5. URL Changes
  6. Publish date changes

Goal is optimize content that already exists for an organic search gain which provides a majority of the sites traffic.

Only 200 of the pages have been analyzed. Fellow person used some crazy complex statistical analysis and machine learning. The person who did this is now gone. Me and one other person are the only meaningful contributions to this project.

If I can do the analysis and provide some insight I think I'll be slated for leadership role. I want to take initiative on this project but don't know where to start.

I don't know where to start but have access to the site analytics. I want to keep this close to my chest and just email him a report (results better than talking about doing it).

TLDR: Want to measure change in organic growth after cleaning up old landing pages but don't know the first thing about analysis
#analysis #huge #project
Avatar of Unregistered
  • Profile picture of the author GordonJ
    Originally Posted by longfacebear View Post

    Working on a content marketing project.

    Changing 5 variables across 1000 urls
    1. Modifying h1's etc
    2. Content quality improvement
    3. Add image
    4. Add video
    5. URL Changes
    6. Publish date changes

    Goal is optimize content that already exists for an organic search gain which provides a majority of the sites traffic.

    Only 200 of the pages have been analyzed. Fellow person used some crazy complex statistical analysis and machine learning. The person who did this is now gone. Me and one other person are the only meaningful contributions to this project.

    If I can do the analysis and provide some insight I think I'll be slated for leadership role. I want to take initiative on this project but don't know where to start.

    I don't know where to start but have access to the site analytics. I want to keep this close to my chest and just email him a report (results better than talking about doing it).

    TLDR: Want to measure change in organic growth after cleaning up old landing pages but don't know the first thing about analysis
    Try google, sites like semrush. You'll see 10 alternatives, but the point is, when you visit these sites, you see what ANALYTICS look like, so you at least have a starting point.

    Good luck.

    GordonJ
    {{ DiscussionBoard.errors[11475627].message }}
    • Profile picture of the author longfacebear
      I will do this thanks for your advice. Do you think advanced statistical analysis like a regression model is even helpful or just a fancy waste of time?
      {{ DiscussionBoard.errors[11475679].message }}
      • Profile picture of the author GordonJ
        Originally Posted by longfacebear View Post

        I will do this thanks for your advice. Do you think advanced statistical analysis like a regression model is even helpful or just a fancy waste of time?
        Sounds like a project where a lot of online info is out there, and someone wants it either updated, or changed. Whether any of it is a waste of time, or not, depends on what the intent of the project is.

        What do they want from this effort? That might give you insight into the kind of data they need analyzed, or changed.

        Gordonj
        {{ DiscussionBoard.errors[11475683].message }}
        • Profile picture of the author longfacebear
          Objective of updating all of the content is to increase the amount of organic traffic.

          So basically I want to Measure the increase in organic traffic across the project then point to why our traffic went up so that we can repeat it.
          {{ DiscussionBoard.errors[11475720].message }}
  • Profile picture of the author akshay7
    I'm not sure what you are asking here. Changing content, analytics, regression models etc. If you provide some details, I'll try to help you out.
    Signature
    For global monitoring, online reputation management and reactive marketing strategies, message me or
    Visit - www.redswanlabs.com
    {{ DiscussionBoard.errors[11475685].message }}
    • Profile picture of the author longfacebear
      Trying to measure change in organic traffic for web pages that have been "SEO optimized".

      That part is the main issue the secondary issue is pinpointing what caused the increase or decrease.
      {{ DiscussionBoard.errors[11475716].message }}
      • Profile picture of the author akshay7
        Originally Posted by longfacebear View Post

        Trying to measure change in organic traffic for web pages that have been "SEO optimized".

        That part is the main issue the secondary issue is pinpointing what caused the increase or decrease.
        I am fairly confident that your predecessor did not use machine learning or regression for this. The most he would have used for this would be simple stats like confidence intervals. It sounds like an overkill unless you are in a research role.

        Anyways, the solution would comprise of the following steps -
        • Connect to Google analytics or any other analytics tool.
        • Use the api to get results of before optimisation and after optimisation.
        • Find the change and if it is irregular.
        • Form your fancy reports and impress your boss.
        Signature
        For global monitoring, online reputation management and reactive marketing strategies, message me or
        Visit - www.redswanlabs.com
        {{ DiscussionBoard.errors[11475718].message }}
        • Profile picture of the author longfacebear
          I definitely agree on the overkill but he was a Data engineer or something from Wharton. but who cares ;p

          This definitely seems like a great place to start. Will jump into google analytics. Any thoughts on getting to the root cause behind the changes? Because theres like 900+ threads so big data set
          {{ DiscussionBoard.errors[11475723].message }}
          • Profile picture of the author akshay7
            Originally Posted by longfacebear View Post

            I definitely agree on the overkill but he was a Data engineer or something from Wharton. but who cares ;p

            This definitely seems like a great place to start. Will jump into google analytics. Any thoughts on getting to the root cause behind the changes? Because theres like 900+ threads so big data set
            Once you point out the pages that had an increase in the visits, try to find the difference between them and the ones that didn't have an increase. This can probably be done manually or semi-automatic.
            Signature
            For global monitoring, online reputation management and reactive marketing strategies, message me or
            Visit - www.redswanlabs.com
            {{ DiscussionBoard.errors[11475882].message }}
            • Profile picture of the author Jason Kanigan
              Originally Posted by akshay7 View Post

              Once you point out the pages that had an increase in the visits, try to find the difference between them and the ones that didn't have an increase. This can probably be done manually or semi-automatic.
              This seems like a valid approach to me. You have the original page code and the new page code, and the difference in results for comparison.

              I'd dump the data into an excel sheet and filter for increases of a certain % or higher. Then look for correlations in changes made.

              Next question is: What is statistically significant? The math will tell you most of the 1000 urls are required, like over 700 of them. So I think while you'll get at correlating effects, causation will be difficult to come by. But if you see the same changes resulting in similar effects over, say, 50 pages, that's enough to make a decision on.
              {{ DiscussionBoard.errors[11475936].message }}
              • Profile picture of the author longfacebear
                Okay, so Ill start in Google Analytics and decipher the finding in a spreadsheet (using 700+ urls). I'm thinking conditional formatting and sorting by percentage change for starters. Might post findings here later on.
                {{ DiscussionBoard.errors[11476026].message }}
  • Profile picture of the author Jason Kanigan
    Hmm I'm a statistical process control guy and I'm having a hard time seeing how regression analysis would be helpful for your situation.

    Sounds like the guy who is gone now was 'being the wizard.' Did his analysis focus on what changes to make would have the most impact on SEO results, so you'd know which steps to do first? If so, while the idea is interesting, I doubt there's any way you could be certain about the predicted impact. The analysis might look cool, but the assumptions supporting it would likely not be valid. Of course, if the buyer is thinking the person giving the analysis really is 'the wizard', that doesn't matter, does it ;-)
    {{ DiscussionBoard.errors[11475934].message }}
Avatar of Unregistered

Trending Topics