Friday, February 7, 2014

Driving Dramatic Improvements in Outbound Telesales with Lean/Six Sigma




Six Sigma leaders often struggle to maintain management's attention on the Quality effort.  There is a lot of management enthusiasm at the start, but if they aren’t being flogged by the CEO, interest can quickly wane.  There are many reasons for this but sometimes it's due to the fact that the projects are not focused on what the leadership team is most concerned about.
However, you would be hard pressed to find something that matters more to management than increased revenue.  And if you can increase revenue, while decreasing the costs associated with securing that revenue, you’re going to have a lot of interest and support.

A great revenue process to apply Lean/Six Sigma tools to is outbound telemarketing.  Every company does not have outbound telesales activities, but for those that do, it is a “target-rich” process full of waste, poor design, and variation that is ripe for the application of Lean and Six Sigma tools.

The Problems:  Muda, Muri, and Agent Variation
Value-added analysis is a bread and butter Lean tool.  With telesales, the thing going through the process is a potential sale to an existing or prospective customer.  One of the keys to deciding if something is "value-added" is whether the thing going through the process is changed.  Well the only changes to a potential sale are either a sale or a client who says they are "not interested."  "Call back later" or "My Mom's not here" or a wrong number or an answer left on an answering machine do not change the status of the potential sale.
The implication here is that most of the activity that outbound telesales agents engage in is non value-added (NVA).  For example, as good as outbound dialing technology has become a lot of "wrong calls" still come through....calls still get made to "Joe's Pizza" or you get the nanny's number by mistake.  When this happens the agent has to spend valuable seconds dispensing with that call.

Another example of “NVA But Necessary” work is reaching an answering machine, waiting for the message to play, and leaving a message.  While there can be real utility to leaving a message that motivates the prospect to call back, many agents just hang up because from their standpoint it is more valuable to move on to try to convert the next call.
Wrong numbers and answering machines are two examples of muda or waste in the telesales process, but they are the tip of the iceberg.  Table 1 presents a full assessment of waste we have identified in the telesales process.

Types of Muda
Telesales Examples
Process Defect
* Skipped sentences, phrases, steps, etc deemed critical-to-quality
* Incorrect path/response selected based on customer response
* Incorrect or no information entered into CRM
Extra Motion
* Wrong numbers (could be a process defect but adds to talk time)
* Reaching answering machine and leaving voice mail messages
* Talking more than necessary to make conversion
* Added/Excessive pauses in conversation
* Excessive time to complete a step
Over Processing
* Gathering data that is never used
* Call Monitoring (because it is not improving center-wide output measures and therefore over processing)
Waiting
* Customer placed on hold
* Customer waits for agent to complete task
* Waiting for systems to complete processes
* Agents perform after-call summary work
* Agents off the phone being coached
Table 1:  Examples of Muda in Outbound Telesales Processes
For those familiar with call center operations, two examples in Table 1 may be surprising:  Call Monitoring and Agents off the phone being coached.  Despite its ubiquity, monitoring is clearly an inspection step that does not improve quality and only finds the lack of quality (#3 of Deming’s 14 Points:  Stop Depending on Inspection (Deming, 1982)).  And while coaching has the potential to reduce defects and other forms of waste, if you graph a center’s agent output metrics over time, you do not see continuous improvement in agent output metrics despite a huge investment in monitoring and off phone coaching time.  Therefore, empirically, coaching is not adding value (Adsit, 2010 and 2013).
While many lean practioners are familiar with muda in all its various forms, some are less familiar with muriMuri is a type of waste that occurs when people and processes are overburdened and unreasonable demands are placed on them.  A long list of factors can contribute to muri.  It can be production demands, poor work environments or process designs, lack of training, clutter, noise, poor tools/equipment, and on and on.
Along with multiple causes there are multiple indications of the presence of muri, but one is the amount of voluntary turnover.  Low turnover certainly doesn’t mean the process is free of muri. People get desperate and will hang on to brutal jobs when there are few alternatives.  However, high turnover almost guarantees the presence of muri (pay increases at competitive companies can drive turnover, but that can’t go on forever).  And the turnover numbers in call centers in general and telesales in particular will drop your teeth. 
In US-based call centers, the average turnover is in the mid 30’s percent.  This is higher than any other position in the corporation.  Moreover, mid 30’s is the average.  Many centers are much higher than that, especially overseas.  I spoke to the leader of a call center in Michigan at the height of the recession and her turnover was 100% annually.  Another leader I recently spoke with is running a 500 seat center that is mostly inbound sales.  Not outbound telemarketing, but similar.  His turnover is 350% annually.  One-hundred fifty to two hundred percent turnover in overseas call centers is prosaic.
Why is call center turnover so outrageously high?  In part because of muri…the unreasonable demands of the job.  Cold calling, fund-raising, and telesales work is physically, mentally and emotionally draining.  I don't care if you have the skin of a rhinoceros, people hanging-up on you, yelling at you to leave them alone, and just saying No to you and the cause you believe in, day in and day out, is draining.  You can sit in a Tony Robbins Sales Training class for four days out of five and it would still be hard to not "take it personally" after a day of "No's,” curses, and hang-ups. 
That, in and of itself, is tough.  However when you combine caller belligerence with the endless repetition of the job, all the administrative tasks, and the wear and tear of talking all day, you end up with an unreasonably stressful job and people quit…in droves.  And this turnover is enormously expensive (HR costs for recruiting, hiring, terminating, extra coaching costs, etc) and it completely precludes the ability to drive center-wide improvements in agent output metrics.  The turnover eats the agents’ gains and you start over with new, lower performing agents.

A final contributor to telesales sub optimization is between-agent variation.  I have discussed the between-agent variation problem in depth in this publication (Adsit, 2010) and only want to make a couple more points here.  First, it is really between and within agent variation as Sally making calls at the beginning of the day is not the same Sally that is making calls the end of the day when she is tired.  Nor is the Sally making a call after she has been cussed at the same Sally making a call after a sale.
Variation reduction has been more the domain of Six Sigma than Lean.  What is interesting about between/within-agent variation is that if you analyze the root causes of the variation on process adherence, disclosure compliance, AHT, etc, the bars in the Pareto are always the same though their relative influence changes across time and call types.  The drivers of agent variation usually boil down to:
  1. Relatively Stable Individual Differences (IQ, personality, accent, motivation, etc)
  2. More Temporal Individual Differences (mood, focus, fatigue, illness, etc)
  3. Knowledge & Skill Differences(training, coaching, experience)
Faced with these drivers of unwanted variation, most call centers try a combination of selection improvements, training, coaching, incentive compensation, and a host of efforts to gussy up the work environment (Adsit, 2013) and then sit back and hope they work.  There are a lot of tools in the Six Sigma tool box, but I have never seen hope listed as one of them.
The larger point here is this:  in call centers, the agents are the process.  And when the agents are the process, you have no process.  You are left with trying to improve each agent one at a time.  And that is the losing game call centers of all stripes have been playing for 40 years.
The Fix is In

The first thing that needs to happen is to get the scoreboard up.  If they don’t already exist,  start tracking the key output measures over time (run charts or, even better, control charts).  We need this “over time” scoreboard up because this is how we are going to track the effects of our improvement efforts.
On the revenue side, conversion rate is perfectly fine to start with and eventually it might make sense to add order size if that measure matters and if there is a wide variation in performance across agents.

You need a cost measure as well.  "No we don't.  My managers don't care about the cost of sales...they just want more revenue."  Really?  Well then why don't you staff your call center with people like Og Mendino, Zig Ziglar, Tony Robbins and Ron Popeil?  The fact that you are not willing to pay to staff your organization with high caliber sales people means that costs do matter.  So find a cost metric...Cost per Call, Cost per Conversion, Cost per Dollar of Revenue, etc.  Just find one that management or the finance team can get excited about.
With the scoreboard in place, start attacking the waste in its various forms.  The way to address NVA activities is to 1) eliminate them, 2) reduce the steps/time it takes to complete them or 3) find a way to get them done less expensively. 
The key technology for accomplishing all of these things is agent-assisted automation.  According to Wikipedia, Agent-assisted automation is a type of call center technology that automates elements of what the call center agent 1) does with their desktop tools and/or 2) says to customers during the call using pre-recorded audio. It is a relatively new category of call center technology that shows promise in improving call center productivity and compliance.
In telesales, agent-assisted automation can address two huge sources of Extra Motion muda:  wrong numbers and answering machines.  Pre-recorded audio and pre-programmed system actions allow the agent to politely end the call with a wrong number, update their system and signal to the dialer they are ready...all with a couple key strokes.  Agent-assisted automation can do the same with an answering machine...leave a message (instead of just hanging up), update the internal system and case notes to plan the next call to that prospect, and signal the dialer the agent is ready for the next call...again all in parallel.  Shaving 15 to 30 and even more seconds on every call like this adds up to more time spent on value-added activities...an immediate productivity lift.
The next two big areas of muda are process defects and waiting.  Telesales calls are fairly scripted, meaning they don’t branch in dozens of different directions.  There is some kind of greeting, perhaps some kind of verification, a pitch of some kind, a few ways to handle No’s, a close and a way to end the call.  That whole process including updating systems and case notes, but with the exception of “closing” an interested qualified customer, can be completely executed with pre-recorded audio. 
With the agents executing outbound calls with pre-recorded audio you’ve provided, the process defects are almost completely eliminated.  There is of course some training/coaching needed at the beginning to make sure the agents are making the right choices, but once the choice is right, the process is defect free.  Also eliminated is the bulk of the between and within agent variation (Madrigal, 2013).
So we did Process Improvement 101 (eliminating and streamlining non value-added work and driving out agent variation) and just made your outbound group 10-15% more productive. 
We have barely scratched the surface.

Extending the Productivity Gains
 
As with reducing inventory in a factory, the elimination and streamlining of the NVA has drained the swamp and revealed that there are really two calls or two process involved with telesales work.  There is a lot of pounding through numbers and admin work and a basic pitch to identify qualified and interested buyers and there is “the closing.”  The former requires virtually no real skill, while the latter requires the agent to be enthusiastic, well-spoken, and accent-free.  This suggests that there is a natural and potentially productive division of labor and in fact there is.
There is a British expression I have always liked, "horses for courses," which means different people are suited for different things and you have to get people doing work that is aligned with their skill set.  By splitting the job into a Tier I admin group that is pounding through phone numbers to get prospects on the line and a Tier II group of closers you create two different hiring profiles. 
The first group, using the pre-recorded audio provided by the agent-assisted automation, do all the dialing, waiting, leaving messages, dealing with wrong numbers etc.  When they get a prospect who expresses some interest, they transfer them to Tier II with a few button presses.  Tier I agents almost never even speaks with a customer.  Some of our clients even turn their microphones off. 
The Tier II team is your closer group.  This is job with a higher skill set requirement.  As mentioned, the agents need to be enthusiastic, have the ability to connect with people, be well spoken, know how to deal with rejection, etc.  They are harder to find and you have to pay more for them.  But the good news is you don't need as many of them as they are only taking the qualified calls being handed off by Tier I agents.
The news gets better.  Though we streamlined the Tier I work, there is still a lot of waiting time (dialer delays, phones ringing, etc).  As a result, most Tier I agents find they can easily handle two calls at once and some even three!  Remember the calls are all in different phases, the basic interchange with the customer is often very straight forward, and even though it is a cacophony of voices in their headsets, the agents don’t seem to have any trouble handling it.  In fact, if you are deploying a kind of piece rate compensation system the agents are super motivated and grateful to be able to handle multiple calls.

The productivity gains from this Tier I/Tier II change are incredibly multifaceted.  First, obviously, one agent handling three conversations is a big productivity hit (See Figure 1)!  Second, the Tier I agents are easy to source.  It is a low cost, entry level job and since the agents microphones are often muted, it can be off-shored for additional labor savings.  Third, training is also reduced since most of what needs to be done is baked into the automation.  Fourth, the Tier I monitoring costs and off-phone coaching time are dramatically reduced (or can be redirected to Tier II) because the agents aren't speaking, the process doesn't vary that much, and the automation is always correct.
 
Figure 1:  Revenue and Cost Improvements from the use of Multiple PCs


Finally, though it is too soon to have collected data on this, this redesign of the process and use of automation shows signs of addressing the muri (unreasonable demands) associated with telesales work.  It thus has the potential to reduce turnover and all the costs and corrosive effects on performance from that turnover.  This requires a bit more explanation.

Outbound telesales agents love agent-assisted automation.  The Tier I agents using pre-recorded audio for qualifying the prospect don't feel bad when the hang-ups and curses inevitably come.  They tell us that it doesn't feel personal to them.  The way they describe it, the customer is not saying No to them, he/she is saying No to the software.  Further, they don’t get yelled at because of their accents; they don’t get as fatigued because they don’t have to talk at all; and they make more money because of the piece rate incentives.  Agent satisfaction measures soar.

Also, your higher-priced Tier II closers are hearing fewer No's and virtually no cussing because they are only being handed calls where the potential customer has shown some interest.  You have also virtually eliminated non value-added activities for this important group.  They are selling on every call and their conversion rates go up (and their pay goes up too if you use variable compensation).  Satisfaction is way up in this group as well.
We hope to have data soon on the long terms effects on turnover reduction.  No one is ever going to make a career out of outbound telesales, but even small extensions in agent tenure can improve center wide performance metrics and reduce the HR costs associated with turnover.

The Icing on the Cake:  The Path to Higher Conversion Rates


There are two conversion rates we are interested in:  the conversion of someone who answers the phone into an interested prospect (the Tier I agent pitch and counters to the customers’ initial hesitance) and the conversion of a prospect to a customer (the Tier II pitch and counters).  Analyzing and managing existing variation and doing controlled experiments (DOEs) are the key Six Sigma tools for improving conversion rates. 
The first step at improving either of the conversion rate starts with graphing the agents on a p-chart.  At this point, you are looking to identify agents that are statistically better than the rest for best pitch/counter practices.
Now for the Tier 1 agents who are using a predefined process and pre-recorded audio, we have not seen as many statistical differences among the agents.  We have found agents that are converting less due to burning through too many numbers in an attempt to raise their piece rate, but even this is rare.

No between-agent Tier I differences is the bad news.  The good news is the fact that the Tier 1 pitch is completely automated which means that experimenting with different pitches is a snap.  You can run as many different pitch variations as 1) you have hypotheses about variables that affect the success of the pitch and 2) you have enough agents to run sound experimental designs.  And not only is the pitch something to experiment with, but the recorded voice becomes a variable that you can test as well.  Some voices convert better than others.  One company found that a women’s voice with a slight Hispanic accent converted better than all the others and they went with that for all agents.

Moreover, each agent running multiple machines enables you to run better experimental designs with more power.  Rather than run a randomized block design with half the agents running one pitch and half running the other, you can run a completely randomized design with each agent running the two pitches on his/her two machines.  It’s a better design with an extra degree of freedom (no blocking factor), which enables you to detect smaller differences.

As for improving the Tier II conversion rate, since they are not using automation, graphing them on a p-chart will likely yield agents that are statistically better (the Tier II group is smaller so you may not have enough of them).  Why is Sally converting at a higher rate than Jane?  Is it Sally's voice, her empathy, is there something in how she pitches and rebuts the initial "No", or is it her tenacity?  Just make sure you are using statistics to identify real differences and not just eyeballing the fact that Sally is #1 and Jane is towards the bottom of the pack.  Treating noise as a signal is a surefire way to develop superstitious behavior and make interventions that actually make things worse.

The p-charts of agent performance will also enable you to get better at hiring, training and performance management as well.  What is it about Sally’s approach?  Can we hire for that attribute?  Can we train and coach other agents to follow Sally’s approach?  Jane is worse, but is she statistically different than the rest?  We of course want to manage up or out those that are statistically worse, but it is essential that we get away from this capricious and inefficient “fire the bottom 20%" approach (See Figure 2).



 Figure 2:  p-Chart of Sales Conversations

Becoming better at 1) identifying the right attributes and selecting new hires based on those differentiating attributes, 2) training and cross pollinating best practices, and 3) managing out those that are statistically worse are great approaches for continuously lifting your center wide conversion rates.   (For more, please see these two articles in the reference section:  Taking the Guesswork and Gamble out of Hiring Call Center Employees and Call Center Coaching Remains a Labor-in-Vain.)

Companies know how to cut costs, but all companies are racking their brains for ways to drive revenue growth.  If you pick Six Sigma projects targeted at increasing revenue you will have management's undivided attention.  Using some of the recommendations here, when your projects actually deliver revenue increases while simultaneously driving down the cost of securing that revenue, well, I hope you won’t mind getting carried around the building in a sedan chair.


References

1        1.  Adsit, D.  (2010) The futility of call center coaching.  iSixSigma.com
 
2        2.  Adsit, D (2010)  Fixing between-agent variation can make all the difference.  iSixSigma.com,
3        3.  Adsit, D. (2013) Coaching Remains a Labor in Vain. It doesn’t have to be this way:  call center blog
4        4.  Adsit, D. (2013)  Why your turnover reduction efforts are not working. It doesn’t have to be this way:  call center blog
5        5.     Adsit, D.  & Bobrow, W.  (2007) Take the guesswork and gamble out of hiring call center employees.  Originally published in Call Center Magazine, now available at http://ifyouwanttoscream.blogspot.com/2014/01/taking-guesswork-and-gamble-out-of.html
6        6.     Deming, W. E.  (1982) Out of the crisis. The MIT Press.
7        7.     Madrigal, A. (2013). "Almost Human: The Surreal Cyborg Future of Telemarketing". theatlantic.com.