Statistics are only as good as your interpretation is
Posted on 02. Dec, 2009 by Franz in Cultures
Enough said about the title above – Any more words into the title could make it sound offensive already. There are certain things marketers, researchers, analysts and optimizers can do to help themselves & the company – Get your minds right before facts.
The accuracy and credibility of statistics has long been debated hotly around the world: On one side, some would say, “80% of all statistics are made up” and on the other side, “Educational programs & courses are using statistics for base learning”.
So are they actually teaching the concept, idea and methodology of educational courses or; research companies ate crap from its respondents?
Which leads to today’s topic: Statistics is only as good as your interpretation is.
Statistics are measures
Whenever you come across bars, charts, numbers or anything statistical next time, put this in your mind – “If I’m not prepared to digest these numbers in an understandable & positive manner, then I might as well look elsewhere until I fully understand the situation/need a base assumption point/looking for a more tangible supporting character.”
Statistics are measures. Measures of which define values of a particular object/subject’s property. For example (not real figures):
- Subject: Population volume in Malaysia
- Subject’s Value: 28,200,000 (Property 1 – Amount)
- Date of Update/Since last refresh: June 2009 (Property 2 – Date)
You can argue and say that these numbers are dead – Well of course, to a certain extent. But if you were to make that argument, then all publishing materials should be considered also dead: Including novels, educational books, training documents, user manuals and so forth.
What makes publishings like statistics valuable is [how is the story told to its audience] and [what benefit it brings them]. To users like me & you, it’s how we interpret them.
You can buy a RM10,000 research paper from the best research company in the world and not gain any value out of it – Or you can scourge around the Internet and look for topic-specifics, then make base assumptions to support your experiments.
Statistics still need actionable proof
No matter how much statistics on Malaysia we compile and I, retweeting them on Twitter, supported by tons of market research companies, they are all there to only give you certain standings. You still need additional proof on your subject of speech.
Because for all you know:
- Reporting on statistical numbers can be inaccurate: Rounded up, close figures, missed entries, etc.
- Data gatherings can differ: Tool A for data and Tool B for data may not necessarily tally in results although subject of test is the same.
- Data could be inaccurate: A good example – Surveys. How many times have you answered a survey question and know that’s not your true answer? If it doesn’t apply to you, it applies to many people out there.
I’ve said this once in the blog to implore and remind experienced marketers: “Don’t do guesswork, get your answers“. I was speaking about search engines, optimization techniques and technical digestions for search in that post.
There are still so many questions in my mind until this very day about search – Internet applicators keep their secrets and algorithms so well and all I can do is sometimes try out ’strategies’ or ‘tactics’ that MAY work out.
Also: What works today, may not work the next hour; or tomorrow.
Statistics need good interpretation
Let’s take for example a keyword research mechanism provided by Google Adwords for PPC campaigns.
Google's Keyword Tool - "Diamond Earrings" for Malaysia
Now if you look at the picture you can see that I’ve done a few things – Filter out unrelated results here and there, target it specifically to Malaysia, enable results for mobile search, placed my maximum CPC bid for the day and per click, and sort the table out by estimated average CPC.
Am I doing the right thing here? Probably. Here are a few questions you can ask yourself (example answers in green):
- What am I trying to achieve now?
Find out search volumes for possible keywords to buy; find out max CPC; get an idea of local search trends; get estimated CTR values. - What data can I get from this tool?
Values like est. CPC/CTR; est. competition levels; keyword ideas & expansions; local search trends. - What for?
Know the Google’s search market on “diamond rings” and other related keywords; find out how much it costs to buy these keywords from Google on their PPC campaigns. - How do I use this data to improve my marketing strategies & campaigns?
Now I know what keywords to push for natural search & buy for paid search; now I can pretty much gauge what are the search terms used in Malaysia; now I can see different 1-year trends for keywords in Malaysia. - Do I need a second opinion/data table on this matter or this tool is totally sufficient?
Better be safe – Let’s get a second opinion; let’s play safe – better get a customized report from comScore; we can afford to take the risk – let’s just do it. - Can I use this data to make base assumptions for my marketing report?
Maybe. Let’s try experimenting on Google now to get our answers!
Well that’s one side of it. Imagine if someone were to come along your way and not bother thinking three times or more before deciding on keywords to purchase, target for natural search or targeting for different hierarchical pages on their website.
