@mitch @aum007 I like to do integraiton and quality decision-first. Find the decisions that matter, work out what a good one looks like, work back from there to see what data and what data quality is required to improve it.
@aum007 yes resource to build predictive analytics still fairly severe. It's getting better with more cloud-based and automated modeling tools available but for SMBs you are really looking for third party services/software that embed and use predictive analytics rather than building your own I think. At least for now.
@Mitch @django1 I am not sure I would disagree with the characterization so much as regard it as a maturity curve. Each piece of data added to the EDW does not need to go through that sequence but most organizations will! Gradually companies become more and more decision-centric (white paper on this here)
@James-Yes thanks for your input! But are present automation solutions accurate enough(to pick up various idiosycranacies between different cultures) and also to take care of hundreds of maybe overlapping Compliance mandates?
@aum007 actually the more diverse and spread the call center the more valuable the automation of decisions. Ensuring compliance, effectiveness and learning across disparate call centers is almost impossible but a decision management system can do it.
@Kasi Big data works best when applied in operations; improving operations means improving operational decisions; operational decisions are high volume, increasingly real-time so must be automated; applying predictive analytics to big data let's you apply the data to improve these operational decisions; best kind of systems for doing this are agile, analytic and adaptive; more in my book.
@Mitch well the reality is that the world keeps changing so no decision ever stays perfect long! one of the reasons an adaptive system is so important is that it continually challenges your decisons to see if better ones are possible or if previously "perfect" ones have degraded.
Of course, you also run the risk of repeating the mistakes that "efficiency experts" made more than a century ago -- thinking that if someone succeeds by turning a screw with their left hand, then everybody should do it that way.
@Mitch while I know how many in a group fairly accurately, I am not making decisions about a group. I may know that 70 of 1,000 applications are fraudulent but I need to know how likely it is that the NEXT application is fraudulent. That's never going to be black and white, I am going to get a probability or score that I can use with other data to make a good decision. Uncertainty about THIS transaction turned into a probability about THIS transaciton.
@DEWacht great point. All too often organizations are ignoring what they learn from prior decisions to improve the next one. With operational decisons you make so many that the value of this is hard to underestimate. Capturing data baout decisions made and the outcomes that result and feeding this back into the system is an important step in the maturity curve.
@James:" Most are not yet using decision management systems to improve interactions nor are most capturing call center notes in a usable way but it is beginning to change." << Do you have a prediction as from when this will change making it more common?
James, you said predictive analytics turns uncertainty into probability. But it seems to me that predictive analytics turns uncertainty into certainty. We know that out of every 1,000 customers, N will do X. Am I wrong about that?
@Kim Davis trust is an issue for sure. Couple of thoughts on that. First thing is that it depends on the way the decision is currently made. If people in the call center are frankly guessing today then if I can give you a 60/40 probability you will be ahead! Second many of the decision management systems I work with explain their reasoning, including the basis for their predictions, so that the consumer of the decision can see how and why it was made. That tends to help adoption too. Finallly and most importantly if the decisions are being made in a way that makes the user more likely to hit their personal objectives they will rapidly come to trust it - if it works they will trust it :)
Mitch - not a specialized term. Just mine for how does the organization factor in a decision made last week in a different part of the business for a similar decision point. Are we re-inventing the wheel each time a decision is made?
@Mary Jander I see more and more organizations trying to make the call center a profit center not just a cost. Those folks see gathering, capturing and using the data from the call center as valuable. Others, as you note, tend to focus on minimzing call length. Most are not yet using decision management systems to improve interactions nor are most capturing call center notes in a usable way but it is beginning to change.
@James: Do you think that garnering analytics from call centers is a tactic that is catching on with enterprises? If so, do those enterprises need to make sure they're not cutting off the callers too soon in order to meet call quotas?
Slide 35 - the end result is to build system, decision management systems, that are agile (so they can be readily changed by the business without delay), analytic (embedding predictive analytics to improve decision-making) and adaptive (so they can support test and learn and other adaptive approaches to improving decision-making over time).
@JeffT: If it doesn't work, switch browser, Chrome works well. If Blog Talk Radio is blocke by your firewall, wait until you get home and download the audio from the archive from this same link or from the class schedule. :)
@Mitch: Good point. There's also a need, though, for call centers to stop demanding that help desk personnel not run through calls to terminate them too quickly. If there is no quota of calls to complete, more input can be obtained for analytics.
Second Wind - Try refreshing the page. If that doesn't work, try a different browser. If that doesn't work, your network might have blocked streaming audio, in whcih case we can't help you, but the audio will be archived here immedaitely after the presentation.
One of the most intriguing points I've heard recently is taht companies should treat their call centers as a research opportunity rather than a cost center.All those people calling with problems are helping you improve your product.
@Everyone: yes, it's a live audio Webinar. No video today. Live chat later with the speaker. Hold your questions. Slide deck you can find above this chat window: Special Educational Materials, slide deck- If you can listen to audio, refresh your browser. If it doesn't work, switch browser, Chrome works well. If Blog Talk Radio is blocke by your firewall, wait until you get home and download the audio from the archive from this same link or from the class schedule. Any questions? ask. :)
Stages of data warehousing as I learned in my Teradata classes: "reporting (what happened), analyzing (what is happening), predicting (what will happen), operationalizing (what could happen),active data warehousing (make it happen)"
@MSCrosby, sorry for that. :( But stay here for the chat and later questions with the speaker. Get the audio at home from the archive, it will be availabel as soon as the live presentation finishes. Find it from this same link or from the class schedule from the front page