Metrics: Sweet Spot

[RickB]

OK, here is the $64k question - (as asked by senior mgmt.): What is the SINGLE "magic metric" that reflects the perfect balance between customer satisfaction, associate satisfaction, staffing cost, teleco cost, customer hold time and ASA? Well, now that I write it down, maybe it is more of a $1M question.

I have some thoughts (a few of which are unprintable) but what are yours - as regards this metric : - ). The form would be something like:

((Occupancy * .5) + (ASA / 100) + (Sat Score) / Handled Calls)
(This isn't a serious attempt at the answer - just an example to show what I am getting at.)

Anyone up to the challenge? (Of course you won't really get any money - but you will experience a soft warm glow for any and all worthy contributions to the answer.)

Best Regards,

Rick

[gkern]

Cost per call.

Cost per call is the answer if you take everything into account. Part of that depends on what business you are in, but assuming that you have customers that have bought a product or paid for a service then cost per call is the answer.

Cost per call takes into account:
Agent cost
Retention
AHT
ASA
Customer Sat (as it relates to ASA, Surveys, and willingness to purchase again)
IT and telecom costs
Management costs

The customer sat. cost is important and probably hardest to pin down, but it influence the other metrics. For example, longer hold times and poor first call resolution may change buying habits or satisfaction with the service.

[rjpasto]

Rick -

Effective performance metrics balance value and cost.

The closest I've seen to a single, magic metric is this:

Average handle time (efficiency) over 6 months (washes out unusually short/long calls) coupled with caller satisfaction (business value/effectivness).

Effectiveness first, efficiency next.

Hope this helps.

Bob

[JoeB]

The challenges in defining a single magic metric stem from the fact that most call center metrics represent different and conflicting interests, the more obvious ones being handling time vs. speed of answer, handling time vs. customer satisfaction and the like.

Different organizations attribute different weight (level of importance) to these metrics and to the outcome (customer satisfaction), which a single magic number will hide. Consequently comparing two different operations using a single metric will not be “apples to apples”.

Similarly, the inclusion of handling time will skew the results when comparing dissimilar operations. AHT will be very different if the call center supports standard MS desktop applications, custom UNIX software applications, answer HR questions, or handles hardware related problems. A very efficient operation with an intrinsic long AHT will still get a low score.

Cost-per-call metric attempts to reconcile all the factors into a single number, assuming that operational costs reflect internal efficiencies. But besides the inconsistencies in measuring cost per call (see www.DiagnosticStrategies.com/papers/benchmarking.pdf), assigning monetary value to customer acquisition/satisfaction/retention is far from being a scienceÂ… Moreover, customer satisfaction is influenced by a number of factors, the more important one being first contact resolution, and much less ASA as many suggest. Another factor that is rarely measured is an outcome of inconsistency in service delivery.

A reasonable way out of this is to use a balanced scorecard approach to assign strategic and business values to each of the key metrics and define your own value equation. The formula may not be transferable to other organizations, which is impractical anywayÂ…

Many call centers flood management with too many charts of the various key performance indicators. This unnecessary and often confusing because many of these KPIs influence each other and the information may be too detailed for senior management (but very useful for the call center manager). If management is interested in a more concise report, customer satisfaction, abandonment rate and agent utilization (instead of AHA or even calls/agent) may be a reasonable compromise.

I agree with Bob that it is more important to emphasize effectiveness over internal efficiency, which may also lead to a more feasible definition of target performance metrics. (BTW, the proposed 6 months measurement period is fine only if you business is not seasonal; many are).

Joe
www.DiagnosticStrategies.com

[RickB]

Thanks to all for your input, many echo thoughts similar to mine.

To Joe:

Your comment on first call resolution as opposed to ASA hits upon another issue, and that is the belief held by some that service level measurements such as ASA can be replaced by a simple Answer Rate (i.e. the inverse of abandon rate). Where do you stand? (We are not talking about a combined metric here - just the replacement of ASA with AR.) To be blunt, my position is that AR only measures your customers tolerance to the service being provided, and not their satisfaction (or delight) in that service. I also would maintain that you don't control the customer's tolerance, while you do control the level of service (ASA) that you decide to deliver.

Once again, your thoughts?

[JoeB]

Interesting discussionÂ….

I agree. Metrics should be used primarily to gauge the quality and efficacy of service we give customers, and then the efficiency in which we do it. I also agree that managing for efficiency (e.g. answer rate) may miss the mark. Customer satisfaction is driven primarily by the quality of resolution (i.e. first contact resolution) and to a lesser degree by ASA. The latter is kind of interesting. We tend to look at *average* speed of answer, yet the abandonment threshold (i.e. the time the caller hangs up) is not distributed normally (in which case an average figure is relevant), but I have found that most callers hang up consistently after a certain a certain amount of time (which varies by organization & industry)

BTW, is your definition of answer rate the inverse (1/AR) or complementary (total calls - abandoned)?

Joe

[RickB]

Joe,

Answer Rate = (Answered Calls) / (Offered Calls)

Interesting comment on the time of abandonment and it's related distribution. I have always maintained that you can not get an unbiased analysis around the tolerance to hold because the activity will always reflect the service level you are giving at the time.

Best Regards,

Rick