I have just finished reading “The World Is Flat” by Thomas Friedman. Firstly a health warning – it’s a REALLY long book. Even skimming some sections it took me a long while to read it. Overall it is a good if somewhat long winded read. As someone working in technology I found it a little patronizing in places but that could just be a function of its target audience not working day to day with some of the technologies he’s discussing. The book lays out a series of trends and technologies that have, in his phrase, flattened the world by making it more interconnected than ever before. He goes on to discuss how this fits with globalization, how companies are reinventing themselves in the face of these changes, some of the problems and risks and what kinds of political and public policy impacts it might all have.
I was reading this in the context of Enterprise Decision Management, EDM, and several concepts introduced in the book resonated with me.
The first is the idea that deciding where to source work is becoming more complex. There are more options with advantages and disadvantages than ever thanks to the overall increase in interconnectedness. For instance, Thomas discussed how JetBlue reservations use “homesourcing” and are 30% more productive in terms of bookings made and how other companies are outsourcing call centers, for example:
“There are currently about 245,000 Indians answering phones from all over the world or dialing out to solicit people for credit cards or cell phone bargains or overdue bills”
Thomas points out that
“Homesourcing to Salt Lake City and outsourcing to Bangalore were just flip sides of the same coin – sourcing.”
or as Thomas Koulopoulos called it when I heard him speak recently, Smartsourcing. Thomas K. also gave a presentation called The road to Agra that touched on these same topics. Thomas F. explains that the work that will go where it can be done most effectively and that increasingly only “creative, complex strategies” will be done in developed world if it is possible to say “I am getting the grunt work done efficiently far away. ” Now this last phrase made me think about EDM in this context. Why would I have the “grunt work” done far away if I could automate it and control it locally? Much of what EDM delivers is the automation of grunt work, decisions in workaday transactions that do not really require intelligence to make – just the application of rules and analytic insight. So when considering sourcing the various pieces of your process you should consider if you need a person at all – perhaps you can use an EDM approach and automate a step rather than outsourcing it. Even if you decide that a piece of the process should be outsourced or homesourced or moonsourced or what ever then you still have to think about how you can control this sourced process. Will you just rely on policy manuals and training? Will you assume that the folks making decisions on your behalf can interpret data correctly from their reports and apply your business strategy to what that data tells them? Perhaps you should automate those decisions so that you can control the logic in them even though they are sourced and so your unique data can be used to go beyond BI and actually inform how they work. In the case of the homesourced booking agents, wouldn’t you want to make sure they offered your best travelers upgrades when they could and knew how to prioritize customers that needed re-routing as well as what upsell to make to whom? What about the 245,000 phone operators? Would it help if they had an automated system for approving credit or for telling what kid of collections strategy would work? Of course it would. And think about the legal issues here – who’s on the hook for the legality of the behavior of these folks? Not the Indian outsourcer but you. Can you show that the decisions they took were legal, compliant, unbiased etc? Not if the decision is manual. Let’s make this concrete using one of Mr. Friedman’s own examples. Here’s what he says:
“In the coming phase of work flow, here is how you will make a dentist appointment: First, there will be a common standard for making dental appointments with any dentist. You will instruct your computer by voice to make an appointment. Your computer will automatically translate your voice into a digital instruction. It will automatically check your calendar against the available dates on your dentist’s calendar and offer you three choices. you will click on the preferred date and hour. The week before your appointment, your dentist’s calendar will automatically send you an e-mail reminding you of the appointment. The night before, you will get a computer-generated voice message by phone, also reminding you of your appointment”.
Now leaving aside Thomas’ belief in the growth of standards to cover everything, let’s think about this scenario with EDM:
- You will instruct your computer by voice to make an appointment.
- Your computer will automatically translate your voice into a digital instruction.
- It will automatically check your calendar against the available dates on your dentist’s calendar and offer you three choices. you will click on the preferred date and hour.
- In an EDM enabled process it would use your rules and predictions of when you are likely to want an appointment to make the three selections
- In theory the dentist might have rules constraining appointments (new patients, cleaning only etc) and these would be included in the decision-making
- A prediction for the length of time you would be at the dentist, given the kind of appointment and previous experience with you and patients like you, and the likelihood of a follow-up might constrain these choices further and even, perhaps, suggest pre-booking of the follow-up appointment
- Information about your choice would be used to improve the model of your preferred slot
- The week before your appointment, your dentist’s calendar will automatically send you an e-mail reminding you of the appointment.
- You would have set rules both for when you wanted to be reminded and how so that this decision was personalized
- A model predicting the likelihood of you being late or missing the appointment might cause additional activities such as a live call if you are a high risk for missing it
- The night before, you will get a computer-generated voice message by phone, also reminding you of your appointment.
- Similarly this would be customized to suit you
- The system that called you would give you various options (confirm attendance, say you might be late, cancel) and these options might reflect your particular coverage (yours might say “Cancel and pay a cancellation fee” for instance)
- If you cancel your session an automated conversation would be started to capture a new booking time and a decision would be taken as to who to call and offer the short-notice visit to (given the length of appointment etc).
- Staffing and scheduling of people and equipment for the actual visit might be dynamically altered based on the results of all this
Lots of decisioning making the process more personalized, more efficient and more agile.
Another area of interest highlighted in the book was that of global, dynamic supply chains. In particular the Walmart supply chain and its immediate responsiveness was discussed alot. The move to real-time or just-in-time manufacturing and delivery was highlighted in the phrase
“[coordinate] disruption-prone supply with hard-to-predict demand”
Thomas describes a number of scenarios where companies are making rules-based decisions to keep these automated supply chains moving. However, he also talks about sharing data as a critical aspect of these supply chains. I don’t have a problem with that, per se, but it seems to me what companies need in these circumstances is not data but insight from that data, Is it more useful for me to tell you I just sold one of your items or to tell you that I am predicting to run out of them next Thursday? As we add RFID and generate yet more data I believe the value of insight will exceed the value of raw data by an ever increasing margin and that automation of decisions that take advantage of that insight will be key. As Thomas quotes in the book:
“In this world a smart and fast global supply chain is becoming one of the most important ways for a company to distinguish itself”
Note the use of “smart” here. I might say “smart enough” – there’s no need to try and embed artificial intelligence or anything in them to make progress.
The need for business agility came up again and again. For instance supply chain problems were highlighted as being
“exacerbated by the short life cycle of product today… Innovation is happening much faster, and so products go in and out of fashion much faster”
and the example of Spanish retailer who works on the basis that it is more profitable to have shortages and then respond REALLY fast to them. This company is taking customer preferences and feeding them into a rapid turnaround system to meet new demand. Clearly customer preferences can be expressed as rules and used to do this but, again, I could not help feeling that predictive analytics might both improve the decision and act as an early warning that a decision is needed.
There were also a couple of nice examples of what I would consider EDM applications. There was a story about UPS developing a system that allows US Customs to specify rules for inspection. This shows what I mean by outsourcers having to allow customer some control over the rules in their system. But what about prediction? As data on contraband and other issues is gathered it should be possible to have the system predict the risk of certain packages being problematic and routing them for inspection even though they don’t fail any of the specific rules. The combination of explicit rules and data-driven analytics has proven enormously successful in fraud detection, it would work here too. Similarly in the story about embedding intelligence into Rolls Royce engines to allow for remote diagnostics to see, for example, what to do about a lightning strike, there are clearly rules but there could also usefully be analytics.
A few final thoughts:
- “But first you need your own customers – your own distinctive competency for your company”
and if you are going to run a distributed and largely automated company, you had better be able to embed that distinctive competency into your systems
- “digital, mobile, virtual and personal”
Carly Fiorina’s comment on the future still stands and EDM matters because personalization across mobile channels requires the kind of deep personalization only an EDM approach can deliver
- there is social pressure on the global supply chain
Not all compliance issues are about external regulations, some of them are about ethical compliance and self-regulation. Are your automated systems behaving ethically?
Originally published on the EDM blog.