by Andrew White | April 4, 2018 | Comments Off on Changing the Time-Series at GM and What it Means, and Of Honey-Pot Data
News yesterday in the US printed edition of the Wall Street Journal that GM will stop it’s traditional monthly reporting of car sales. See “GM scraps a Standard in Sales Reporting.” Instead the firm will continue reporting car sales quarterly. The reason given is that 30 days is too small a period in which to make a fair conclusion on the trends in the business. I head just the other day a CEO of another firm say just the same thing on CNBC’s Mad Money show. It maybe a valid reason. But there are implications and unintended consequences to this action.
First, by increasing the interval of the reporting period, it is quite possible that investors, looking to understand the business, may now react in more significant ways due to increased volatility that comes about from grouping the sales data and using larger time periods. If you review three numbers in a series, and you see small fluctuations, the average may move slowly – weekly as in the current method. If you now look at three numbers but they change more markedly in quarterly buckets, as in they are more representative of the business, changes may lead to over reaction.
So the time period in question will impact the importance investors put on the information thus shared. The very fact that the market was able to consume small, incremental even noisy data led to gradual market shifts. The only time a market movement would be significant is if monthly data did imply a one-directional trend! So what is GM really playing at here?
I really don’t know. But the logical of going from monthly to quarterly data seems like a return to the past. We really ought to be looking at more than weekly data, even smaller periods, like daily data. With more data for smaller periods we could turn on our algorithms and be able to look away! They can do the math and look for hidden trends in large sets of small numbers. With quarterly numbers, even AI could react more nervously.
There was another interesting story in the same newspaper- see “You’re not 113 years old? Why people lie to Facebook.” This article reports on folks who deliberately lie to Facebook and enter false information about themselves. One chap notes with glee that he smiles when he hears of data breaches since all his information, other than his name, is made up. This is a kind of fake date, or another variant of honey-pot data: fake data set up deliberately to ensnare an erroneous action or action. In this case these Facebook-falseys are seeking upset marketers and data analysis (I didn’t say data analytics; there is no such thing).
Can we do this on a large scale? Is the value one gets from Facebook undermined if we falsify everything? Even our likes? Who we are friends with? I figure they such honey-pot data or false data can only go so far. Eventually the value you get from a service will decline below the safety and additional pleasure you get from the loss of the exploiters of your data. So it seems fake data can only go so far.
Sent from my iPhone. Excuse the typos.
Comments or opinions expressed on this blog are those of the individual contributors only, and do not necessarily represent the views of Gartner, Inc. or its management. Readers may copy and redistribute blog postings on other blogs, or otherwise for private, non-commercial or journalistic purposes, with attribution to Gartner. This content may not be used for any other purposes in any other formats or media. The content on this blog is provided on an "as-is" basis. Gartner shall not be liable for any damages whatsoever arising out of the content or use of this blog.