Finally, we would be remiss if the consideration of all the ‘‘other stuff’’ here did not to include at least a passing reference to the role played by cost in the users’ ultimate determination of the value of a particular service. Although the assessment of cost/effectiveness or cost/utility of telecommunications services is an entirely different matter, well beyond the vision of this book, there are several valuable principles as to how to formulate and analyze the trade-offs between QoS and economy of service (EoS) suggested by the treatment of problems of measurement and evaluation quality herein.
For example, the discussion of assessed QoS in this section clearly shows
that service providers should not try to trim costs at the expense of deleterious effects on the factors that determine user assessment of overall QoS. This goes almost without saying, but it does become important enough to warrant the admonition to cost analysts to be sure that any cost/effectiveness trade-off studies explicitly recognize the value to users of the intangible factors besides perceived QoS that determine assessed QoS, stated simply as the principle:
Attempts to reduce costs at the expense of provisioning, customer service, and billing is penny wisdom and pound foolishness; to ignore what produces assessed QoS is to invite bankruptcy.
Another principle of this kind is:
The value of an improvement to a telecommunications service must be gauged in terms of perceived, rather than intrinsic, QoS.
In view of the definition of perceived QoS, this may seem like an attempt to belabor the obvious for the unconscious. However, the subtle point here, which has been made several times with respect to relationships between quantifiers for perceived and intrinsic quantifiers of measures of quality, is that there are levels at which further improvements in a measure of intrinsic quality produce little or no improvement in user perception of quality. If these relationships are not recognized in cost/effectiveness analyses, it is entirely possible for a service provider to be encouraged to invest in something, like an increase in call completion rates from 99.0 to 99.5%, that is very attractive from the viewpoint of performance or technology, but has very little impact on user perception of quality.
Finally, for all those involved in analysis of cost and pricing of delivery of service, I offer the ultimate heresy that the never-ending argument between advocates of economy and quality may, in fact, be a totally meaningless one, because:
It is possible to improve perceived quality with no increase in the cost of delivery of service.
The message here is this. In evaluating alternatives for telecommunications services, it is usually presumed, if not explicitly assumed, that there must be a trade-off between EoS and QoS. Such presumptions lead us to expect, for example, that: investments in quality improvements must be justified by an expectation that better quality will attract and retain more users, or warrant a higher price for the service; a lower QoS will be less attractive to the user community and must therefore be delivered at a lower cost and price if the service is to be competitive; etc.
services probably do account for much of the telecommunications market mechanisms, it is also true that there are many cases in which QoS and EoS are, in fact, not in conflict, and may actually be complementary, in the sense that there are service delivery options that simultaneously enhance both QoS and EoS. This means that there are opportunities in the telecommunications world to improve quality with no increase in cost, to reduce costs with no loss of quality, or even to improve cost and quality at the same time.
The truth of this observation is best seen from an example that readers can test for themselves. Consider one of the big problems assessing the presumed trade-off between quality that is confronted in deciding how best to provide for those conditions when additional capacity or alternative routes outside of the provider’s own network are needed to deliver traffic. For any particular desti- nation there may be many different alternative carriers, each offering a differ- ent unit price for use of their networks and facilities. Because the lower price alternatives also tend to offer poorer performance, there is always a concern that the ‘‘obvious’’ solution of using the lowest priced routes will materially degrade user perception of quality, with dire consequences in network management centers and/or the marketplace. As a consequence, there tend to be on-going arguments between service provisioning and finance as to whether cost or quality will be the determining factor in the selection of providers of alternate or overflow capacity. When major customers begin to complain about the QoS after the bargain-of-the-month reseller’s trunks are moved to first choice in the overflow routing plan, the operators will readily cite that problem as clear evidence of the fallacy of the lowest cost strategy; when use of the bargain-of-the-month has no apparent impact on user satisfac- tion, the advocates of the lowest cost strategy are quick to latch onto that experience as evidence that service does not have to be ‘‘gold-plated’’.
To see how the adoption of the appropriate evaluate concepts and measures might allow for an alternative that is satisfactory to both the finance and provisioning personnel who are at loggerheads over the choice of strategy, pretend for a moment that we are they, and suppose that there is some destina- tion, D, for which we expect to be offered substantially more traffic than our network can carry. Suppose, further, that it has been decided that simply blocking the excess is not an option, because our customers reasonably expect a better grade of service to D. In this circumstance it is tempting to conclude that our objective is to locate sources of enough extra capacity to D to ensure an adequate grade of service, and selecting from the alternatives identified the source(s) which will carry the traffic to D that we expect to hand off for the least cost.
In other words, it is tempting to posit that the objective is to procure the minimum amount of extra-network capacity needed to assure adequate QoS
with respect to handling the expected traffic to D, and do so at the least cost. However, it is more useful, and probably more accurate, to posit that the objective here is actually to realize the greatest income from the potential revenue represented by the offered traffic to that destination.
This objective, then, immediately suggests that the criterion for selecting among alternate sellers of capacity to destination D should be based neither on a measures of QoS nor costs. Rather, the appropriate measure is expected return ratio (ERR), which can be defined generically as the ratio:
ðamount of revenue expected from a call overflowed to DÞ
=ðcost of providing for the overflowÞ A pretty good quantifier for this measure can be defined as the ratio:
½ðASRÞðBDÞðPCÞ={ðCSTÞ½ð1 2 ASRÞðPDD 1 UTÞ
1ðASRÞðPDD 1 AT 1 BDÞ}
where ASR is the answer seizure ratio; PC is the price per minute of conversa- tion charged to the customer for a completed call to destination D; BD is the average billable duration of an answered call to destination D in minutes; CST is the cost per minute of use of the overflow route; PDD is the average PDD for calls completed via the alternate route; AT is the average time to answer for calls answered at destination D; and UT is the average ring time for calls not answered.
Now, if we look at the factors in this quantifier of the expected return ratio, PC, BD, AT, and UT will be stable and fixed for calls into destination D. The other three will be characteristics that may vary from seller to seller. And, note in particular that CST is the measure advocated by the ‘‘least cost’’ strategists, while PDD and ASR are two of the measures most frequently cited as bases for the criteria advocated by the ‘‘best quality’’ strategists.
I could at this juncture, then, go on to concoct examples of situations where analysis of this ratio for competing sources of overflow capacity would alter- nately favor the lowest cost source, the source offering the best ASR and PDD, or some compromise in between. However, I think I’ll leave that exercise to the curious reader, and be content with the obvious conclusion from the definition of this quantifier of ERR that since CST, PDD, and ASR are inde- pendent variables, it is entirely possible that an analysis of alternatives will show that the source that would have been chosen on the basis of least cost may also be the one that would have been chosen on the basis of the best values of PDD and ASR…
Afterword
Since Part I of this book ended with a description of the concerns of persons who use telecommunications services, it is somehow perversely fitting that Part II ends with a similar description and discussion of the concerns of those who have to deal with the service providers in paying for telecommunications services and assuring that the quality of what is paid for is satisfactory. If the intent of these descriptions of such concerns has been realized, readers will have found them to be so intuitively credible as to be self-evident. The user concerns in Part I, for example, should be immediately recognized and appre- ciated by anyone who has ever used a telephone, absent any knowledge or understanding of telecommunications technology. The ‘‘other stuff’’ described at the end of Part II should be readily recognized and appreciated by nearly anyone who has had experience in dealing with a telephone company, inde- pendent of any knowledge or understanding of the sophisticated management theories, organizational concepts, processes, procedures, and policies that determine how the service providers interact with their customers.
Description of those readily apprehensible concerns in terms devoid of technical language may, in fact, have smacked to some as ‘‘belaboring the obvious’’ or ‘‘unnecessarily tutorial’’. Yet, I dare say that few would argue with the premise that these simple, concrete, kindergarten concepts have served us well in the effort to characterize quality of service (QoS) for tele- communications, illuminating and motivating definitions and derivations that might otherwise have been nightmarishly obscure. To the extent that the reader has found this to be true, this book conveys by demonstration the message that the key to credible, cost-effective, scientifically defensible measurement and evaluation of QoS is the preliminary characterization of what is, or may be, important to those who will ultimately determine whether QoS: Measurement and Evaluation of Telecommunications Quality of Service William C. Hardy Copyright q 2001 John Wiley & Sons, Ltd ISBNs: 0-471-49957-9 (Hardback); 0-470-84591-0 (Electronic)
perceived and assessed quality will be acceptable, expressed in terms that are meaningful to those who will be making the judgments.
Such a characterization is, in technical terms borrowed from philosophy, an ontological model of the service being analyzed. I have on occasion tried to stress the importance of such ontological models in analysis of QoS by appeal to the maxim, obviously formulated on the premise that the worse the pun, the greater the likelihood that it will be remembered, that:
If you want meaningful measures of quality of service, ask not what de tech would use; ask rather what the user will detect.
It has been my experience in nearly 35 years of defining measures and quantitative evaluation schemes that analyses predicated on a good ontologi- cal model proceed almost unerringly to operationally meaningful measures, easily calculated quantifiers, and evaluation criteria that are credible to users, readily acceptable to decision-makers, and accurately predict the likelihood of satisfaction with perceived quality. The usefulness of such analyses has, more- over, survived the competition (or non-competition, depending on one’s view- point) from such proffered replacements for the discovery of truth as mathematical programming, logit regression, expert systems, ‘‘data mining’’ and neural networks.
It is such experience, clearly evinced, to some extent at least, in what has been presented here in Part II, that returns us to where we began in Part I, to my ontological model of analysis and the corollary admonition to begin each new QoS analysis effort with something that cannot be implemented on a computer, no matter how fancy and colorful the graphical user interface – a trip into the minds of the persons who will be assessing quality, to determine what they will experience and how those experiences will shape their concerns…