Big Data is the new black. No, let me rephrase: Big Data is the new 2000’s website: over-hyped, technology-focused, visually-oriented (of course, data visualization is the new web design) and a marketing protectorate.
So yes, if you ask me I predict lots of wasted effort, in the form of excellent IT sales (hardware, storage and software licences), result-independent bonuses, more and longer meetings, fancier and bulkier Powerpoints, management bickering, eventual backfitting data to a self-serving narrative of reality and quarter after quarter of Business As Usual waiting for The Next Big Thing.
But in the current rush to Big Data there’s something that’s escaping many.
Allow me to illustrate. There was a time when being an “editor” was like this:
Today, the world’s largest editorial project relies on this other kind of editor:
If the task is defined enough, bots will outperform humans.
Yes, I know. Human editors are still needed. Actually there job is improved by having this many bots taking care of menial tasks such as typos, missing cross-references, flagging disputable or offensive content, enforcing editorial guidelines, etc. . That’s not the point. That humans are no longer needed for low-level editing is the point.
Skeptic? OK. But do you really think your favourite human-made newspaper or news-site would not improve greatly by employing edit bots? Would you bet good money? We are talking competent, non-unionised workforce, working 24/7 at the assigned task for practically nothing at all with what can only be described as bloody-minded stubbornness and determination. It’s an employer’s dream, let me tell you.
But of course we are used to machines taking over the hardest and most menial jobs, leaving us our rightful tasks: deciding and directing in the command room. If only the machines were not taking our place in the command room as well.
Do you remember those stock-exchange movies showing a floor of sweated, yelling, overweight (mostly white) males on the brink of a nervous breakdown? That is also a rapidly disappearing job:
This is not about evil employers seeking underpaid workforce. This is better business performance. No human can trade in 30 milliseconds (the advance high-frequency traders get on the next fixing). Bots can. Bots do. Trading bots raise more money than humans. Follow the money, baby.
It gets better.
The show business (“there’s no business like show business”), this sanctuary of physical humanity, is going algorithmic, too. There used to be a time when ego-inflated, temperamental alpha males with a tendency to go ballistic at the slightest hint of a problem ruled the world:
Today, strong leadership is still the best way to meet production deadlines, and humans still yell, cajole and otherwise boss other humans around like no other species. But when it comes to decision-making, humans increasingly find they only have to ratify an algorithmic decision:
Think it’s hard to program better acting than your favourite M$-a-film star?
OK, I hear you say, but even if some formula may eventually decide which artist to invest on, artists are still humans, with their creativity, their flare, their unique personalities. You are convinced of this, right? Do you realize you live in a culture where One Direction, Take That, Spice Girls and other lab-created performing ensembles routinely top the charts?
It took almost three-years to find out that on-stage Milli Vanilli was only providing “marketable image” to real vocalists. Since then, recording companies have side-stepped the problem by doing to vocals what magazine covers do model bodies. Both are real as a unicorn, and both sell like hell. But you still cling to the “unique personality” illusion, eh? OK.
Enter Hatsune Miku:
- 1000000+ songs
- 170.000 YouTube videos
- 1.400.000+ Facebook fans
- booked out in LA, Taipei, Hong Kong, Singapore, Tokyo
- 13-15 December, 2013: at Opéra Châtelet, Paris with Luis Vuitton & Marc Jacobs
From a business point of view, Hatsune Miku:
is just as artificial
- has comparable market potential
- and a tiny fraction of the cost. Bots don’t get royalties.
I am convinced that only thing keeping synthetic artists (actors included) from entering the mainstream is the agent cartel. And that will last until they find a way to represent synthetic personalities. Less than five years, in my opinion. As usual, porn will lead the way.
A task must be performed by the lowest-cost agent capable of performing it. Anything else is waste.
So? So work is evolving. Many things we are used to call work today do not necessarily require a human to perform them. They algorithmically determined, hence best handled by an automated assistant.
Look around you: call centers are the way of the past, chatbots are what we’ll increasingly deal with. And I think we will be more and more happy with it. Why? Because the marginal cost of employing software is next to zero, while humans insist on being paid. So the choice is between trying to reduce the fixed cost of manpower at near-slave level by employing unskilled people in exotic locations, or to invest in something that can produce a level of service equal to or better than a human at a near-zero marginal cost.
It gets even better.
As you read these lines, business automation is already at middle-management even in marketing. I don’t know how else to call Google’s keyword planner. Other business functions will be taken over soon. The data-enabled flattening of hierarchy makes no distinction.
So, yes, you should seriously loo into Big Data, with one advice:
The current selling proposition is that Big Data is data but, you know, it’s BIG, so you shold buy my wonderful platform. That’s bullshit.
Everybody will tell you Big Data is about Volume, Velocity and Variety. True, but we’re missing an important step: data quality.
Big Data is process. And it starts before you collect data. As a matter of fact, most of your very large company databases is just scrap from a data-driven business perspective. If you had the time to make some investigation, chances are you’d find your client, accounting, purchase, production, logistics databases are not aligned. Yes, even if you are using the famous three-letter industry standard ERP.
Data-driven business is not a software issue. It is a systemic strategy to acquire, maintain and leverage high-quality data. You either do it or you don’t, but it doesn’t happen by itself. It’s a process you must want, decide upon, put into place and govern.
The first step is building into your company’s culture that data belong to the company, not to the originating function. Easy? Look your function owners in the eye as you say it, and see what they really think. It’s full of function owners out there who simply refuse to hand over their raw data. Reports? No problem. Statistics? With pleasure. Raw data? No. Fucking. Way.
This insures the company knows exactly what the function owner wants it to know. It’s Cover-Your-Ass Management, and it’s widespread. Just ask for raw data and see.
Enabling data-driven business is more than having databases and software. It requires strong governance and continuous-cycle attitude. Daily reports are better than weekly reports, which are better than monthly ones. And having KPIs is better than having reports, no matter how nice the PowerPoint is. Yes, I am saying you should have a dashboard.
But the problem is not the dashboard: it’s what you put on it. Unless you have a data-driven culture, your dashboard will only be a son et lumière fiction, not a decision tool.
Now, the big question:
How do I promote a data-driven culture?
My answer is: with the Noble Eightfold Way to Data-driven Business
- Right Identification (what constitutes meaningful data?)
- Right Collection (where and how do data come from?)
- Right Validation (data falls into two categories: quality data and junk)
- Right Governance (who reads and modifies what, when, and how)
- Right Interrogation (42 is the answer; the problem is the question)
- Right Interpretation (what data mean as opposed to what you want to hear)
- Right Communication (right people get right data at the right moment)
- Right Purging (too much data is noise)
Obvious? Maybe. Action item: try and see how your company scores (no cheating!) and we’ll start with step 1 on Monday. See you soon!