AI, Big data, Automation,
|
I admit to mostly enjoy looking forward. Looking back can be cringe-worthy - parachute pants and Vanilla Ice and so forth. Like a lot of people I got into tech because I prefer to think about what can be done versus what was done. So it was a total accident when I stumbled upon this Advertising Age article that used a few quotes from an interview that I did in 2003. At the time I was pretty focused on using data analytics to do a better job of online marketing. A lot of what is second nature and obvious today was still new or even future tech then. Stone Age Analytics - 2003 AD The Early 2000s Was a Simpler Time for TV but not analytics. The interview came from work I was doing looking at advertising optimization. We were doing early work into slicing our analysis by the viewer's time. Advertising and commercial analytics tool sets were primitive. The company I was working with provided a broad range of outsourced marketing and advertising services from ecommerce (we had just launched the company's first platform), to direct mail - yes, postal mail, to outbound and inbound customer service and sales calls and marketing creative services. I had a million square feet of fulfillment and 600 or so call center seats. The systems were on different hardware platforms and my office was in a raised floor data center with server racks as my only view. Mid-range and PC platforms were everywhere. UNIX boxes, IBM mid-ranges and so on. Integration involved a lot of transfers of files, especially CSV and even some email and FTP. Mostly it involved us creating our first data warehouse by hand writing ETL and then using shockingly complex SQL statements to get at data we wanted. There was no meta data for most of the sources and we dealt with very different data standards used by our clients, their vendors and partners and the laundry list of agencies and data service companies we worked with to enrich or validate our data. Insights From The Big Data Dark Ages The point of this was mostly looking for usable patterns in the data. Once we found data we wanted to use we would then move some logic to code. The more we looked the more interesting stuff we found. It was real analytics and it certainly was Big Data although nobody had coined that term yet. One thing is wasn't was Artificial Intelligence or Machine Learning. We iterated our way through the analytics, usually coding and scripting until the data looked usable and then trying to figure out if we had found what we were looking for. Half the time it was difficult to know what to do with the insights and I never quite knew how much to trust it. Fast forward to today and the tools and, especially the code, have changed a lot. It certainly makes what we were doing look primitive, inefficient and superficial. Harnessing and accessing Big Data has improved dramatically, analytics tool sets can automatically generate insights it took us days or weeks to get at and code learns through iteration the way. We Have Learned But We Haven't Evolved Recently MIT Sloan published a story illustrating how AI will likely widen the growing gap between our ability to leverage analytics and Big Data to generate insights on the one hand and the ability of managers to do something useful with them on the other. We need solutions to address the fact that accelerating developments of all sorts are going to increasingly make people the weak link in an ever stronger chain. I think the skillsets of the manager of the future need to look a lot different or we risk dependency on systems that will lead to a lot of lost opportunity.
0 Comments
Leave a Reply. |
AuthorMichael Zammuto is the CEO of Completed.com and Cloud Commerce and a strategic adviser to several startups. Mike's background is in SaaS services, B2C sites and B2B firms and has worked extensively in online reputation, digital marketing and branding. Archives
October 2019
Categories |