Additionally, the number of complex processes increases as a customer care organization handles more lines of business and volume.  The result is operations managers within those firms find it hard to be successful at what the industry calls Customer Interaction Management ("CIM").

Adding to the problem is that consumer expectations and sophistication are rising, budgets remain under severe downward pressure and technological complexity continues to increase. It always seems to boil down to three seemingly competing tasks:

• Delivering higher quality customer service experiences
• Deriving more revenue from existing clients
• Reducing the cost of customer service

The good news is these do not need to be mutually exclusive. All three results can be achieved through effective “orchestration” of the interactions and complex processes within a customer care operation.  Technology solutions are available and proven to enable orchestration.

The good news is that many customers have some fundamental expectations in common, such as a requirement for informative, accurate, friendly, available and responsive service.  By creating a customer survey that asks customers how the contact center measures up to each of these areas of service, a contact center manager can better understand, in general, where improvements need to be made in the center.  Measurement at an agent level can be even more effective.  Finally, if the survey results are tracked by customer in an IT system (e.g. CRM application), then the next time a customer calls in an agent can see what part of the call needs to be handled better.  This process will steadily increase the way a company manages its customers' expectations.

This company wanted to ensure companies who did not pay for maintenance were not able to get free support.  So they front-ended every call into their support center with an entitlement check.  This typically took about five minutes, since the database of entitlements was unwieldy.  Basically, a low-cost agent had to ask who was calling, get some codes, find out the serial number of the product being called about (many customers have mixed maintenance fee situations, depending on the product).  If the customer passed the entitlement check, they were "allowed" through the gate to talk to an engineer.

Think about this from the customer's perspective.  You are a senior engineer trying to get a million dollar system online, probably on a tight deadline.  You are extremely knowledgeable.  You need help, and you know exactly what kind of help you need.  It will take several calls.  Each time, you have to waste time talking to someone who knows nothing and whose only purpose is to verify that you are worthy of being connected.  Sounds nice, doesn't it?

There is a really simple way to fix this, that saves a ton of money and generates revenue as well.  My customer did it in less than a month.  Here is the idea:

Let everyone through, first off.  Front end calls, if needed, with a triage person to figure out who the right engineer is to take the service call.  Let the engineer handle the call, no questions asked.
AFTER the call is done, check entitlements off line for all calls received. When you find someone who got "free support" (one call only!), you do two things.  First, you have your person who used to act as a (depressed) gatekeeper instead act as a sales person.  Have the person call the "offending" customer and thank them for calling you.  Tell them you were to happy to take their first call, at no charge.  But then tell them that future calls will be charged, and ask for credit card information so you can allow them to get the help they need without delays.  And of course try to sell them a long-term maintenance contract.
If they buy a maintenance contract or give you credit card data, you follow up with the necessary paperwork and meanwhile continue letting them through (but you are capturing usage data for credit card billing).  If they decline, then their name, company, phone number and product codes go onto a short list that can be checked automatically on the way in to the triage person.

This company did that, and did generate revenue.  More importantly, in the 90% of cases where the customer was authorized to begin with, the customer experience was much improved (and costs were reduced).

This is a good example of something that is technically simple (not even evolutionary -- they had all the tools they needed except a lightweight database app for checking against the hot list).  But it required a change in the mindset of the business leadership.  It wasn't even hard -- more of a "why didn't I think of that?!" situation.

The issue is that many of these needless, customer-denigrating steps arise from good motives and too little attention to the experience of the customer.  And too little clear thinking about costs and benefits -- sometimes all you need is a quick, back of the envelope calculation to show the "ROI"!

Again, are you proactively planning on how you will manage trust relations with your customers, or are you fulfilling their likely expectation that they will be treated like children or, worse, cattle?

I think many companies have done similar things, so this isn't really any sort of a breakthrough idea, but more of another good example of how thinking flexibly about customer service, and considering trust as a ket variable to take into account – can lead to surprisingly good results with little or no investment.

I find that oftentimes major changes can be made in business results without major IT investment projects. While IT projects are important (and it's what we do at Aria, after all!), it is equally important for enterprises to become agile, and to learn quickly and adapt quickly.

Hopefully this example helps.

Jerry called back and told them "the fix was in" (couldn't resist a little Reno humor), but that it didn't work.  They had him open it up and describe the color of a little dot near the battery (a moisture detector to catch situations where consumers drop their cell phones in water).  He told them it was white, but with a little streak of pink on it.  They told him that his problem was not a warranty problem but a paid repair problem, since his phone had suffered water damage.  He reminded them that this streak of pink was from the cotton swabs when he followed their directions to clean the contacts.  You can imagine the series of escalations, arguments, pleadings, explanations and re-explanations that followed -- not to mention the hours spent on the phone by Jerry.

In the end Jerry was charged $115 for a new phone (skipping lots of details; you get the idea).  He disputed the charge, and continued paying his regular monthly contract fees while the dispute was handled.  On midnight of January 31st, without warning, his phone was cut off (while he was on a run with a customer).  Given that his business depends on his wireless phone, and given that no one else provides reasonable coverage in the mountainous area where Jerry does his business, he paid the $115 and the $38 reconnect fee, vowing to himself to move as soon as he could to ANYONE ELSE.  Happily for me, he had his mobile phone ready to go to synch up with me when I arrived, tired, at the Reno airport!

Let's stop for a moment and consider the economics of what just happened.  His carrier, by acting as it did (at least as perceived by Jerry), has put at risk a reliable current and profitable revenue stream averaging $1200 per year, with an overall past run rate of $12,000.  If they had served Jerry well, they could look forward to a reliable revenue stream of $1-3K per year indefinitely.  Plus, they invested at least 2 hours of people time on the phone with Jerry on this issue (Jerry's estimate); let's assume these are inexpensive agents and supervisors, and set this cost at $80, fully loaded (this is probably low).  Let's assume his carrier charges its loaded cost for the replacement phone, and let's assume the $38 reconnect fee is a profit-neutral item (a charitable assumption).  Then the net result of this is that his carrier invested $80 and forced Jerry to invest several hours of phone time (about 3 in an IVR, and 2 talking to people), in order to avoid taking a $115 charge for sending Jerry a new phone, no questions asked.  And in so doing it virtually guaranteed a reduction of several thousand dollars in future revenues AND got Jerry to advertise his story to others (fortunately including me).

Does this make sense?

Nobody would claim that it does.

So why did they do it?

Because of trust.  Large companies set up rules for contact center people to follow that are intended to guard against consumers' taking advantage.  They don't trust their customers.  Any of them.

Now, many consumers indeed do not deserve to be trusted.  If the customer is an unfortunate person who lives hand-to-mouth, actually knows where to find paycheck cashing stores, and churns cell phones as fast as they can, the suspicious approach would make sense.  But if the customer is a long-time, high-revenue, reliable and profitable customer, this DOES NOT MAKE SENSE.  Furthermore, it would be a simple matter to advise customers of the risk of warranty voiding when asking them to clean contacts -- proactively working WITH the customer to avoid putting them in a compromised situation.

I think this situation speaks for itself.  We all need to think carefully about how trust -- and lack of it -- is handled in our management of customer experiences.  The knee-jerk distrust exhibited in this case is HIGHLY UNPROFITABLE and should not be allowed by companies.  But it may be necessary to plan for trust by examining cases like these and thinking about how to establish conditions of mutual respect and trust.

I find it ironic that I wrote the last sentence, since I tend to be a quantitative analysis type, but the fact is that what always gets me going is when people do what they think is right even when it so clearly isn't -- if only people would step back and see the big picture!

Are you taking trust rules into account in your customer experience planning?

As is often the case when things are so demanding that we do our best just to keep our heads above water, there are some common implicit assumptions that tend to push us back down under, struggling for each breath.  One of these is the implied notion that the right way to understand what is really going with center performance is to ask structured questions that are driven by the design of the center.  This is so deeply embedded a notion that it probably does not even make sense on first reading (of course, that could also be due to my writing!).  It seems self-evident – of COURSE we would want to ask structured questions to measure whether our design is achieving what we intended.  But it turns out that self-evident, common knowledge can often lead us astray, and this is one example.  In fact, there are many important questions we should like to ask about our contact centers but typically don't.

To clearly see why this is so, consider call flows.  In most contact centers, considerable time is spent designing call flows to handle all the differing customer needs in an efficient fashion.  Often very large call flow diagrams are developed, and then these designs are coded/configured into routing strategies, ACD vectors, voice/IVR applications and so forth.  Many a contact center war room has walls festooned with detailed call flow diagrams...  Then, in order to measure performance, detailed reports are designed that measure the performance of each of the call flows.  These reports are built up from very structured data that enables the counting of calls that took each flow, as well as the time spent in each leg of the flow (and often the business outcome from the call).

So far, so good.  But, this approach does not measure things that you did not plan for, except when we are lucky.  Unfortunately, unplanned flows are commonplace in contact centers – how it could it be otherwise?  Customers do not care, and should not have to care, how we define our call flows.  They do what they want to do, and the more freedom we give them, the more they will like the experience we provide.  The situation is even  worse on web sites, where users are quite proficient in using navigation tools (back keys, favorites, history lists, etc.) to traverse web sites as they see fit.

These realities call for a different, complementary approach to analytics.  This approach is based on discovering and measuring all of the patterns that occur, rather than measuring only the patterns that are designed in.  In web terms, this means studying the paths people actually take as they traverse our web site, rather than analyzing the paths we designed in.  This is somewhat familiar in the web world, but it is far less so in the "call flow" world.  But the difference in results can be startling.  In one real world contact center where I did some consulting, there were two dozen call flows set up.  Reports were set up for each of them, and these were reviewed by many people.  But, when I asked the question "What are all the call flows that actually occurred, regardless of whether you designed them in, and how often did each occur?", no one could answer it.

In fact, new tools and a new approach were needed to discover all the flows that occurred.  Again, for emphasis, this is quite distinct from measuring all of the flows that were designed.  In the real world contact center situation, there were over 500 distinct call flows that actually occurred.  Not surprisingly, most of the calls were accounted for by counting the most-used flows (the 80/20 rule held, more or less).  However, one of the top ten flows was a completely undesigned – and unmeasured – one.  A very meaningful fraction of callers to this company were ending up in the "Hotel California" queue -- you could check out any time (by abandoning), but you could never leave (that is, get an agent)!  Because the reporting was designed around planned flows, this unplanned flow remained undetected, and these customers remained unserved (if they remained customers or not could not be detected...).

A similar situation likely exists with agent work patterns.  Many centers rely heavily on hard scripting or rigid rules about how to handle calls, and these centers usually also maintain significant investments in highly structured reporting to measure compliance with these designed work patterns.  Analyzing performance using only these "as designed" reports is doomed to miss many important features of agent work patterns.  Some of these might be bad ones – ways that agents might mistakenly or otherwise fail to accomplish what is desired – but many will also likely be good patterns that will not be detected and amplified (by teaching them to other agents).  We too often rely on chance to detect these unplanned flows -- someone following up on a complaint (or a kudo) discovers that things are not, after all, going as planned.

The good news?  There are techniques for performing unstructured data analysis.  Many techniques are well-established today, and we are adding some new ones right now at Aria, in our Strategic Consulting Group (ping me and we can discuss them!). 

But even without these cool tools, it is all about attitude and the approach to data: business analysts should approach their task, and their data, in the spirit of discovery and not in a legalistic search for compliance and deviations.

But what is not so obvious is just how green it all is!  As it becomes more common for agents to work from home, companies will inevitably consider other options as well.  In a day of widely available broadband and global networks that have pushed most of the intelligence to the edge, all it takes is a broadband connection for individuals and small groups to take part in the global economy from wherever they are.  This does not only apply to India and other popular offshore locations (although clearly it does apply there!).  It also applies to places like my little home town of Seabeck, Washington.  Sporting fewer than 2000 people and located in the woods of western Washington state, with an espresso stand and a general store for downtown, it is not a likely site for a new contact center.

However, it is likely that within a few years it will be common for small groups of people to act as job shops that participate in global service markets.   These are already starting to form in various ways; for example, many large outsourcers subcontract some of their work to smaller, specialized outsourcers, and these may in turn have contractors who do some of the work as freelancers.  Coming back to my town of Seabeck, it turns out that there is a fairly high concentration of techies there.  There is no reason they could not form a small niche outsourcing team focused on, say, small business network help desk services.  Many of these teams could be aggregated by a large service provider to provide a global, always-on support network for small businesses.  These small teams (and freelancing individuals) would be "living locally, working globally" -- and they would escape the kinds of lifestyle pressures that come from commuting in to big cities.
Of course, many things will need to be in place before this becomes a widespread phenomenon.  But it will happen, I believe, and more quickly than we expect.  And I believe once it is proven out, it will accelerate.  The simple fact is, once the kinks are worked out, such arrangements will result in much more motivated workforces, much more flexible scheduling possibilities, and lower prices.  And, by the way, much happier customers as well!

But I mentioned that this trend was a green trend.  Here's how: consider if large numbers of individuals and small groups were able to participate meaningfully in the global economy without having to drive to work (or fly there).  The transportation time is one (lifestyle and productivity) benefit, but the reduction in traffic and CO2 emissions that can come from people living in a distributed way but working in a coordinated way will become socially significant as the numbers of such people grows.

Food for thought...





The problem comes from the reality that it becomes a habitual way of thinking.  It is another instance of the common mental habit that led to an overemphasis in management circles on cost elimination to the exclusion of growth in recent years.  The habit is that of taking a good thing too far.  You can improve profits up to a point by eliminating costs, but you ultimately lose if that is the sole source of profit growth.
In terms of customer service, I have seen companies state that their goal is "100% call avoidance" -- if they could fully automate their customer service so they would not have to have any agents, they would be heros.  That is truly the way they think  of it, but the implications of this choice are not always explicitly called out.  It's not that they say "I want to isolate myself from my customers", but rather that "I want to reduce costs, so I want to keep increasing my self service percentage if possible, especially because my bonus is tied to that".

I believe many customer service professionals realize that you generally don't want to eliminate contact with your customers.  There is a balance between service needs that are best met, both from the customers' point of view and the enterprise's, by self service capabilities, and service needs that should be handled by live people.  The use of assisted service could be driven by the complexity of the problem, but it could also be driven by the desire to establish human contact for its own sake.  After all, people buy from people much more readily than they do from machines.

More generally, think long and hard before contemplating giving up direct customer contact.  For instance, before outsourcing all of your customer service to save money, think about how you will hear from your customers without direct, regular contact.  And think about the whole question from the customer's point of view -- if you are pushing customers to self serve in order to avoid incurring costs, they will realize it and devalue your company (after all, you have just devalued the customer, haven't you?).  If you make the customer listen to an announcement -- which cannot be interrupted and which gets done every time they call -- that tells them they could be served more easily on the Web site, they will probably be fine the first time but then become more and more resentful.  Think how you would feel -- the first time you are just being notified of a new option -- no problem.  But when you have to listen to the same announcement every time, even after you obviously know about the web site and have chosen not to use it, then you feel manipulated, devalued, in the way.

Bottom line -- think carefully about the customer experience and about your level of human contact with your customers every time you contemplate adopting a "call avoidance" tactic.  After all, people with no customers never get phone calls!

• There is a LOT of data. When two proton beams are made to interact at CERN, there are LOTS of collisions per second, and each one of these interactions creates a shower of energetic particles whose trajectories constitute the primary data collected (note that each trajectory is a set of data: positions of a particle at successive time intervals).
• One of the first issues at CERN (or any other accelerator) is to have detectors capable of detecting the specific types of particles being generated by the collisions. CERN has several newly-designed detectors, and it is an area of ongoing research driven by the need for more data.
• When looking at data from CERN, there are many aggregate quantities that are useful to measure, such as the strengths of the very large magnetic fields used to steer proton beams, the intensity of the beams, the collimation of the particles in a beam, and so forth.
• Finding interesting data in a sea of numbers: most of the interactions that occur at CERN are not interesting, for any number of reasons. Only a tiny fraction of the particle collisions is useful for detailed study. Therefore, it is crucial to have filters that can separate the very small amount of interesting interactions from the huge amount of background noise.
• Once collision products are detected and interesting interactions are filtered out, they can be studied in careful detail. Getting to this point is already a triumph of innovation (which occurred steadily over the last 75 years or so since the first accelerators were invented) – it is sometimes hard to appreciate what an achievement it is to be able to carefully study a single complex interaction at extremely high energies!
• Data visualization is extremely important at CERN, as it helps humans to pick out patterns from very complex data sets, or to form intuitive understandings of what is really going on which would be difficult to form while looking at nothing but tables of data!
• Simulation is a valuable tool for particle physics, as it allows scientists to understand ahead of time what certain postulated interactions might look like. For example, the major goal of the LHC is to detect the predicted Higgs boson that is predicted by the current Standard Model of particle physics. In order to be able to build filters to look for Higgs interactions, it is necessary to know what to look for. Furthermore, before powering up the LHC, simulations were run to test different safety and scientific scenarios, as the cost of getting it wrong could be disastrous. In general, large complex processes are such that convenient equations describing what is going on do not exist, so simulation is needed to explore likely scenarios and to prevent unsafe conditions.

These challenges have direct analogues in our business:

• Contact centers generate a lot of data as well (although nowhere near as much as at CERN). A large contact center can easily generate hundreds of gigabytes of raw data per day. Much of the raw data tends to be thrown away and only aggregate data is retained, but this is in general a mistake. Ideally, all the data would be retained at least to the extent needed to provide a "lossless system", that is, one where all the rich dynamics of customer interactions can be studied at a later time.
• The concept of detectors is very relevant in contact centers. Ideally, we would like to "detect" every nuance of agent and customer behaviors, including stress levels, buttons pushed in user interfaces, words spoken, and so forth. ASR is an example of a valuable class of detectors that are used widely in contact centers. In most customer care centers, there are usually many types of data that are theoretically collectible but are not yet collected, and one of the challenges of deep analytics is to figure out how to get more and more complete data sets to study. In general, in contact centers and other complex processes we detect events rather than particles, but the basic idea is the same. We figure out what is going on at the point of interaction by studying the large store of event-levle data thrown off by each interaction, just as CERN scientists study what goes on at the point of proton-proton interaction by studying the large store of particle trajectory data thrown off by each interaction.
• The current state of the art in contact center analytics is almost exclusively based on study of aggregate data. This is a good thing, but it is unfortunate that most analytics products and experts seem not to realize that this is only the tip of the iceberg!
• Filters are a very important concept. In a large contact center, there may be thousands of agents and millions of events captured. Most of these will be routine, and may best be studied through their contributions to the aggregate data. But there are interesting outliers in every contact center, and it is important to be able to pass many thousands of interactions through a filter to detect the few that require more in depth study.
• Finally, for deep analysis of customer care, it is usually necessary to be able to study individual interactions (selected by filters) in great detail. This is why lossless data collection systems are important; if all you have are reports and datamarts, you may not be able to reproduce the subtle dynamics of a certain group of interactions by studying each of them in detail.
• As with particle physics, data visualization is very important for understanding the dynamics of complex business processes. This is an area of particular weakness in our industry, along with the single-minded focus on aggregate statistics.
• Finally, simulation is an area where much work is needed in our business. There are in general no convenient methods (well, really, no methods at all) of mathematically calculating the likely effects of complex routing scenarios using queuing theory or any substitute. It's also much like the economy: there are so many moving parts that no simple equations can tell the real story. Simulation is the only method available to predict the utility of a complex routing strategy (or even a simple one, in a large contact center), or to predict the outcome from a certain list management strategy in outbound. And simulation as discussed here means realistic simulation, rather than simplified packaged simulations based on overly simplified assumptions.

This article is a lot to digest. The goal really, at this point, has been to introduce the metaphor of particle physics exploration as a suitable analogue for thinking about, and working on, the contact center deep analytics challenge. I will write much more about this, using this metaphor, and hopefully its utility will be apparent.

For now, it is helpful just to call attention to the strong parallels. And to return to the opening comment, it is easier to state more clearly what I mean by the term "deep analytics":

Deep analytics is the scientific analysis, in ways analogous to the analytical studies at CERN, of complex business processes, by using appropriate event detectors, aggregate statistics, event sequence filters, detailed analyses of individual interactions, visualization tools, and simulation methods, to develop a deep understanding of the real dynamics of the complex processes, and to determine ways to improve the effectiveness of those processes.