Question

Topic: Research/Metrics

Data Interpretation & Analysis

Posted by supria on 9/18/2006 at 5:28 AM ET 250 Points
Im a marketing professional struggling with numbers and data interpretation. I'd like to know of tools that can help in fine tuning my ability to read numbers and analyze data.
I find it difficult to read the fine printand miss out on some critical analysis. I know i can do better than this....

Please Help.
To continue reading this question and the solution, sign up ... it's free!
SUBSCRIBE
Or, SIGN IN

RESPONSES

  • Posted by wnelson on 9/18/2006 at 10:07 AM Accepted
    Supriya,

    There's no "one size fits all" tool for analyzing data. It depends on the application/information need. For instance, if you are looking at data from surveys this is a different tool than if you are looking at revenue trends. In some cases, Excel works - using pivot tables and graphs - even regression analysis. I use this a lot. In other cases, you may need a more sophisticated tool. In ALL cases, a background in basic statistics for marketing professionals is very helpful. If you don't have this, then as Daniela suggests, get some help from someone who does have a statistical background. If you can tell me what exactly are you trying to analyze, I can suggest further.

    I hope this helps.

    Wayde
  • Posted on 9/18/2006 at 10:42 AM Accepted
    Hi Supria,

    One way to get some clarity is to talk to others in your industry and find out what benchmarks or key performance indicators they use. Getting that information will give you some guidance into the usefulness of what you already have and what you also need to collect.

    I would also add to what the others have said in terms of getting someone experienced to assist you - it can also help you to avoid some really costly and time-consuming mistakes.

    Good luck!

    Joy
  • Posted by steven.alker on 9/18/2006 at 12:16 PM Accepted
    It would be helpful if you said what type of data you were handling – is it time related or multi parameter. What kind of results are you looking for from it? Then it’s easier to point to a relevant source of information for regression or basic statistics, cluster analysis, neural techniques, Bayesian stats or advanced stuff like CHAID.

    I’d go a stage further and suggest that if you don’t understand the basics of statistical analysis, using a software package or tool will either further confuse you or result in you producing meaningless interpretations of your data. The figures might be rubbish but you’ll produce them with such confidence that you’ll never know it until someone tells you----!

    If on the other hand you are confident about your Stats basics but can’t handle the sums, there are an impressive number of packages there to help. See: https://www.lionhrtpub.com/orms/surveys/sa/sa1.html for a reasonably comprehensible survey.

    If you need to polish your basics and get a handle on the types of analysis that you can put to given types of data set, then without a doubt, interacting with someone who is a professional analyst is probably the next best thing to hiring a personal tutor. There are some very good courses offered by your Universities and basic statistics is available form many as a free resource .

    One site I can recommend is https://www.statistics.com/courses/basicconcepts/ which even if you don’t want to sign up to a formal course, give enough information in it’s agendas and application areas for you to work out the kind of fundamentals you need to assimilate before you can be comfortable with using high powered software or working from basics on an Excel spreadsheet. If you can’t understand the basics on the agenda sheets, you need either a course or tuition!

    Hope that helps


    Steve Alker
    Unimax Solutions
  • Posted by doreen on 9/19/2006 at 10:23 AM Accepted
    Start with the objectives of the study. What are the key questions that need to be answered? You may not have the time or resources to mine everything out of the data at the moment. It's best to prioritize the marketing questions that initiated the need for the study.

    You may find that many of the answers come straight from the crosstabs and do not require much further statistical analysis.

    Good luck,
    Doreen
  • Posted by steven.alker on 9/21/2006 at 11:23 AM Member
    I rather agree with Matt’s (mmaule) point about the need to be able to handle this at great depth and the ability not to be bamboozled by “Black Box Maths”

    The latter is a problem in most corporations. I have encountered this as an employee and as a consultant working on sales forecasting, projections and sales and marketing activity analysis.

    Many managers, both senior and middle ranking like to have their own magic tools which allow them to see the world the way they want to. If your magic tool is actually valid, it can be very impressive when demonstrating things, events or projections which no-one else had even thought of. If your warnings or action-plans are ignored, and the worst comes to pass, you have your proposals on record to vindicate your stance and make of you a hero. If good things happen, you have a record of having said that they would. If the opposite happens, you will have your margins of error and some platinum plated excuse appropriately handy!

    One of the advantages of having a reasonable base of statistical knowledge is to allow you to ask appropriate questions of the black box merchants, especially as Matt puts it, “When you are being blinded with science” or pseudo science. Mind you, it takes a strong and brave person to ask the FD or CFO, “Can you explain to me the variables, algorithms and assumptions you used when you came up with this projection?” Like good freemasons, they like to keep their secrets!

    And that brings me to my latest tool – Neural Networking. I think that it’s a terrific tool, used in the right hands. But can you imagine the scope for trouble making with a software package which doesn’t even require you to assume a relationship between the measured data and the calculated result. It doesn’t even tell you what algorithm it decided to use in the end. It is just data in and results out. Now that’s what I call a magic black box.

    Steve Alker
    Unimax Solutions
  • Posted by Chris Blackman on 9/26/2006 at 2:54 AM Accepted
    Supria

    I think you need to exclude every piece of data from your analysis unless you can provide a very clear justification for its inclusion.

    Don't get swamped in data. Look for the key pieces of information you need to make decisions.

    Sounds like someone is confusing metrics with KPIs - there is an important difference.

    Care to share some examples of the difficult cases and get some applied insight from the group here?

    Cheers

    ChrisB
BACK TO TOP

Post a Comment

Related Discussions

Community Info

Top 25 Experts
(Research/Metrics)