1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Question for those who have coded their own link/conversion tracker...

Discussion in 'General Programming Chat' started by savvypro, Jun 1, 2013.

  1. savvypro

    savvypro Regular Member

    Joined:
    Jan 14, 2010
    Messages:
    211
    Likes Received:
    133
    I know there are existing solutions (e.g: Prosper202, CPVLab etc), and I can look at the source code of some of them.

    I'm coding my own link/conversion tracker (it's one module for a much bigger platform I'm putting together), and just want to pick the brains of those who have rolled their own.

    Basically I'm looking to make sure that I'm not missing anything in the concept I have in my head. By starting this thread, it should allow me to get clear in the ideas, and have something that I can reference when stuck.

    I have a couple of ideas to ensure that near 100% of the actions being tracked are actually being tracked. For example if cookies/images/javascript etc are disabled, then it falls back on to other means, by degrading gracefully. Actually I should say upgrading gracefully.

    Assuming the average prospect is someone using a browser with default settings. I need to keep my secret tracking sauce, secret for the time being.

    Assuming a multi offer funnel: each stage of a funnel is link by tracking links. Each landing page is tracked by pixel and by page request.

    New never seen before prospect is recorded, cookied and counted. If they can't be cookied then they go down the funnel via an alternate tracking method.

    Each link request results in a couple of data points - if you will. As well as serving of each pixel.

    If a cookied visitor requests a page with a tracking pixel then data is recorded and tallied.

    If it can't track and tally, then the data is incomplete. The other problems is load handling, the more the tracker has to do (or database limits are hit), the greater the chance of data points being missed. The plan I have will avoid some of the database issues but will result is others due to the way the tracker is to be operated, however the issues can be handled. Making it possible to handle large amounts of hits per second with out issue and without a single point of failure. The tracking is the hard part.

    So am I missing anything?
     
    Last edited: Jun 1, 2013
  2. moijeishaianu

    moijeishaianu Newbie

    Joined:
    Jun 28, 2013
    Messages:
    15
    Likes Received:
    0
    oh, this is huge project, having doubt it possible to develop it alone, although you may code some simple implementation that will count some major data cases