Jeff Jonas of IBM had a great post on Some Organizations Will Be Smarter-er Than Others recently in which he discussed his “obsession with real-time sensemaking systems”.
Now I like the phrase and, in the context of real-life, it is clear what it means. But for a business, what does it mean?
Well sometimes it means making sense out of the data you are collecting in a real-time setting, during a conversation in the call center say, and acting on it quickly enough to help the person in the call make a better decision about this customer. Sometimes it means completely automating a decision so it can be embedded in something automated like a website or a kiosk or in a transaction that happens too fast for human intervention.
But sometimes it just means making sense and acting the same minute, the same hour or the same day. If the data you are collecting is about how people who get your catalog go online once they receive it and the decision you want to make sense of is how to change the next batch of catalogs being sent out, then “real-time sensemaking” might not need to be all that fast. If you are working in a weekly or monthly schedule that is imposed on you, it might mean deciding before the next cycle. If you are making a decision that impacts compensation plans or legal contracts or service level agreements, you probably have quite a while because the implementation of your decision is not instant.
Regardless of the timeframe, you must do several things though:
- Capture, clean and integrate the data quickly enough that you don’t use up precious decision-making time
- Analyze and predict using the data so your decision is being made in the light of what you think will be true when the decision gets implemented, not what used to be true
- Define how you want to act based on those predictions in a way that lets those who understand the business participate, not just math or IT geeks
Jeff says, you need to decide
Fast enough to do something smart as the transaction is happening, not minutes, hours, or days later.
But not all transactions happen every second so sometimes you have a little but of wiggle room.
Comments on this entry are closed.
Seems like there’s a big difference between real-time analysis and real-time decision making. Each of these types can be analyzed by how much value is added and risk introduced by reducing latency.
For example, it might be very beneficial to make an instant decision about a customer when they make a purchase at a register. But it might not matter much whether that decision takes into account their activity as of today or just as of yesterday.