The once-humble call centre is taking on strategic importance with the convergence of help desk, telesales and customer service operationsAlthough saying the Australian call centre market is a booming one may be a cliché, it is no less the case. According to the Annual Australia and New Zealand Call Centre Report, 2000 from www.callcentres.net, there are 4000 call centres in Australia, employing 200,000 people, from casual agents to senior managers. Although the annual growth rate has fallen from 30 per cent last year to 20 per cent this year, it is still a growth rate many sales managers would envy.
It is not hard to see why. According to the report, a transaction processed by a call centre agent cost $7 compared with between $25 and $50 for one processed by a person at a branch. Call centres can also be called from anywhere and, subject to staffing, at any time.
Call centre managers have a score of issues to deal with - staff turnover, agent empowerment, skill classification, marketing advertising that generates calls about which the call centre was not informed are just some. However, when it comes to technology, these people often seek some advice from IT.
The core technology in any call centre is the Automatic Call Distributor, or ACD, a capability offered, for better or worse, by most PABXs, Telstra's Spectrum centrex service, and a few Computer Telephony Integration (CTI) products. ACDs manage the assignment of inbound calls to agents, based on a combination of the pilot number on which the calls have been received and the skill classifications of the agent, and the queuing of calls when there are more calls that available agents.
ACDs also typically have their own management information system (ACD MIS) to provide both real-time and historic reporting on a plethora of metrics.
In addition to performing the basic functions well, the better ACDs typically:
Support extension-side E1 circuits to interface to an IVR.
Support the transmission of DNIS (Dialled Number Information Service) to an IVR allowing the IVR to assign the queue-specific menu to that call.
Use an ODBC-compatible database in the ACD MIS allowing easy downloading of reports.
Use an ACD MIS that supports both configuration changes and reporting.
Support skill-based routing, networked ACD, call progress detection for a software predictive dialler and CTI-controlled call recording.
Of the telephone systems available in Australia and New Zealand, there is, of course, considerable product differentiation. In responses to requests for tender I have administered, bidding the Alcatel 4400, Avaya (formerly Lucent) Definity, Intecom E, Interactive Intelligence EIC, NEC 7400 with Genesys, Nortel Meridian 1 and Rockwell Spectrum for call centre systems have all scored well against ACD criteria.
Small Business Uptake
As detailed in its report, Market Engineering Research for the Australian Call Center Market 1996-2006, Frost & Sullivan forecasts a growth in ACD system sales this year of 20 per cent, with growth primarily driven by increased sales to the small business segment.
Beyond telephone systems themselves, the most widely deployed call centre technology is interactive voice response (IVR) systems, which ask callers to press one for sales or two for service and to enter one's customer number, followed by the hash key.
The reasons for implementing IVR systems are reasonably well known. To begin with, they save money. According to the Annual Australia and New Zealand Call Centre Report, 2000 from www.callcentres.net, the cost of an enquiry or transaction performed by an IVR averages at $2.50, compared with $7 for a call centre agent. By performing the most frequently requested and simple transactions, IVR systems enable agents to spend more time on more complex enquiries. In addition, IVR systems can perform such functions 24 hours a day, seven days a week.
IVR systems have been used in Australia for more than a decade. The market is thus well established with between 2200 and 2500 such systems installed collectively by the 35-odd suppliers present in this country. The products themselves offer considerable variation in both capabilities. About half of the suppliers offer a GUI drag and drop, point and click facility to develop menus and the other half use command languages. This differentiation is significant because, as most users of IVRs make changes infrequently, a GUI facility allows them to be made easily.
Other differentiating features are the ability to be powered from DC batteries, mainframe computer interface capabilities and mean time between failure (MTBF) figures, and the ability to generate reports from a database of records that have been automatically collected of each caller's progress through the menu.
There is also significant price differentiation: systems with a single E1 link to the PABX range in price from $45,000 to more than $300,000.
Despite the prevalence of IVR systems in Australian call centres, a much more topical technology is computer telephony integration, or CTI. According to CTI guru Harry Newton, CTI "adds intelligence to the making and taking of phone calls". But it must not be forgotten that CTI is an enabling technology, not an end in itself. CTI applications are designed and deployed to provide one or more time-saving and service-improving functions, the most common of which are:screen pop: populating an agent's PC screen with information about the call purpose (for example, a sales enquiry about widgets) and, if an existing customer, the identity of the caller;screen transfer with call transfer;screen/keyboard outbound dialling to a phone number displayed on-screen;semi-automatic collection of wrap-up codes.
Richard Allen, former CEO of the Australian Telemarketing Association, estimated that between 25 per cent and 35 per cent of call centres were CTI-enabled. This matches the 30 per cent penetration reported in the Annual Australia and New Zealand Call Centre Report, 2000, which noted a jump from 16 per cent in 1998 and 1999. However, Allen went on to estimate that 60 per cent of call centre seats were CTI-enabled, reflecting a greater use of this technology by larger call centres.
Serving the Market
For those 70-odd per cent of call centres that have yet to implement CTI applications, the market is well served with 23 specific CTI products and five customer relationship management (CRM) systems with a CTI capability, as well as a number of telephone systems offering integral capabilities that perform some of these functions.
There is considerable product differentiation. Some are designed to work only with specific PABXs. Many of these systems offer a number of features beyond the above-listed functions, such as offering predictive dialling and audio call recording capabilities, and support for collaborative browsing by an Internet site visitor.
Another capability that some CTI systems offer as an option is control of call queuing. This capability is essential if the switch's own ACD facility is mediocre and can facilitate interleaved queuing of calls and e-mails. However, it has the crucial disadvantage that a server is far less reliable than a PABX and, if powered from UPS, will typically operate for no more than 15 minutes if mains power is lost. The CTI thus becomes a critical point of failure. The CTI link can also potentially impose a throughput limitation, and some switches do not support the facilities necessary for external call control. Also significant is that with call control by the CTI server, the switch has no visibility of the queues, thus requiring the provision of an ACD MIS and reporting package provided by the CTI application in addition to that typically required on the switch to perform configuration changes. It is preferable to use the switch's own call control if possible.
There is also considerable differentiation in price and the capability of the local vendor.
In contrast to a CTI application that is largely invisible to the caller, the use of audio call recording must generally be announced to the caller. Audio call recording systems have traditionally been deployed to record all calls and retain them for legal reasons for services such as stockbroking and emergency services; more recently they have increasingly been deployed to selectively record calls for agent coaching.
Traditionally, such systems used cables to parallel all trunks or extensions, a configuration which, although inelegant and inflexible, suited call logging applications. However, where calls are to be recorded selectively, a far more flexible configuration is to have a small number of dedicated circuits between the recording unit and the switch, with a CTI unit that instructs the PABX to establish a conference call with one of these circuits. Such conference calls can be established by schedule or on demand.
The market for audio call recording systems is well established, with more than 2000 of them in place, although many are in non-call centre applications. Intending buyers are also well served, with 10 products from which to choose.
A far more application-specific technology is the predictive dialler, a system which is used by organisations that make large volumes of calls to sets of residential contacts. The dialler generates call attempts to telephone numbers in a target list, monitors their progress and connects only those calls answered by a person to an agent, displaying the relevant information on the agent's screen. Using an algorithm based on the ratio of call attempts to successful answers, the dialler attempts to minimise both agent idle time and successful calls when there is no agent free to take it.
Traditionally, the predictive dialler was a stand-alone unit with its own trunks and extensions, an inflexible and expensive configuration that is rarely deployed these days. Predictive diallers are increasingly deployed in a configuration in which the dialler has neither trunks nor extensions and is simply a server-resident software application with a CTI link to the switch. Through this CTI link, the dialler instructs the PABX to generate outbound calls and monitor and report on call progress. When a person answers the call, the dialler instructs the PABX to connect the call to a nominated agent. The PABX also notifies the dialler of any status change, such as when a call is transferred or terminated.
For this to work effectively, the PABX must monitor each trunk carrying an attempt at an outgoing call and detect change of status, such as from ringing to being answered, or ringing to engaged, and must immediately notify the dialler of this change of status. Most but not all PABXs in the Australian market can do so.
There are four predictive dialler systems and six CTI and CRM products with a predictive dialler capability available in Australia. Dedicated predictive diallers are more expensive than dialling modules of other products, but they also typically have far more features. Because of both the cost of the system and the specificity of their application, the total number of such systems installed in Australia is about 60 to 70.
As useful as these technologies can be, it must be stressed that with the possible exception of audio call recording, none of them are plug and play applications. They all require development, usually significant development, to be able to deliver the benefits that feature so prominently in vendor presentations.
Take IVR systems. Despite the maturity of the product category, some primitive systems continue to be sold. More significantly, one would expect most organisations using IVR systems to be aware of the need to provide an easy means of transferring to an agent, keyahead, the option to repeat a prompt, and of changing the prompts when the call centre is not staffed. It does not take a rocket scientist to realise that IVRs providing 24x7 service must themselves be supported by enterprise computers operating 24x7. But adherence to such non-revolutionary principles of good practice is far from universal. The good news is that a more enlightened organisation can easily differentiate its service by providing menus that are specific to each customer: for example, by not asking the customer to choose from savings account, cheque account and credit card if the customer has only a savings account.
Nonetheless, development is time and time is money. Although application complexity and hence cost vary widely, the cost of an IVR application can often equal the cost of purchasing the system.
What Screen to Pop
By far the most widely known CTI application is termed "screen pop", populating an agent's PC screen with information about the purpose of the call or the caller's identity. This can be used to improve customer service and reduce call duration and hence cost. However, there are provisos.
This has the potential to be effective only if there is a significant volume of calls of short to moderate duration for specific functions for which a screen can be popped. Although a screen can be popped based on the 13 or 1300 number dialled, to really be effective, the identity of the caller and purpose of the call must be obtained from an IVR. For this function to be effective, the organisation's IVR must have a good menu that callers are prepared to use.
Even what would appear to be the relatively straightforward application of audio call recording has its challenges. For this to be effective, there must be evaluation criteria and forms with which calls can be evaluated; coaching and training for agents who are not up to scratch; and, above all, an overall process to schedule call recording and evaluation, and the time of call coaches and agents to participate.
Rivalling CTI for publicity are the technologies collectively referred to as customer relationship management (CRM) comprising help desk systems, sales force automation systems and customer contact management systems. CRM systems perform the broad function of tracking and managing each customer's contact with the organisation, which uses it for such functions as:
Tracking each customer's contact and sales history by sales channel;Customer base segmentation;Automating an on-road sales force and delivery facility;self-service sales functions; andManaging service requests and fault report tracking.
As useful as a CRM system would appear to be, the report Customer Relationship Management - Strategy & Implementation in Australia & New Zealand from www.callcentres.net, found that only 4 per cent of companies responding to its survey have successfully implemented a CRM software system. It would be interesting to know what percentage of the companies surveyed had purchased CRM software.
Five Times the Cost
There are between 25 and 30 CRM products in Australia that fall within the domain of CRM. These offer a range of capabilities and come with widely ranging price tags. Most significant, however, is the investment, both in supplier fees and staff time, in implementation.
As with IVR and CTI technologies, a CRM system can only implement practices, processes and marketing strategies from within the organisation. No processes, no successful implementation. And CRM isn't cheap. A successful implementation may incur an implementation cost five times the purchase price of the software licences.
The one other technology that deserves mention is the Internet. Internet integration into the call centre takes on a number of forms, but the most common are collaborative browsing, processing of e-mails and Web forms, and WAP (Wireless Application Protocol).
Collaborative browsing is the facility some Internet sites offer by which a visitor clicks on a button and subsequently speaks with an agent while both view the same pages from the Web site. This is potentially a powerful technology if - and this is a big if - interaction with an agent is likely to convert a curious visitor into a buying customer. However, with the cost of sale via the Internet being between 10 and 50 cents (Annual Australia and New Zealand Call Centre Report, 2000) compared to $7 with a human agent, it is easy to see why most sites offering e-commerce do not offer this facility.
Other technologies manage e-mail requests and the processing of Web site forms. From the perspective of technology, these systems need to be able to assign such requests to agents according to service-level objectives that, supplier claims notwithstanding, are far less stringent than those for waiting telephone calls.
There's also the human perspective. While call centre agents are almost always proficient at speaking to callers, only some are proficient at both composing written responses and typing them - skills necessary for responding to e-mails. Even those proficient may not work well when presented with a chaotic mixture of phone calls, e-mails and form processing requests. Agents will be less stressed if they get such requests in batches, uninterrupted by phone calls.
Still within the domain of the Internet is that technology of the month, WAP, a specification that defines the means by which users interact with service providers through their mobile phones. If the user has a WAP-enabled handset, the user can use his or her handset to, for example, check a cinema's timetable from a mobile without having to listen to menus or wait for a human operator.
The use of WAP can only increase; after all, it is new. Nevertheless, an organisation considering its use must consider whether it is appropriate to tits line of business.
Picking and Choosing
So how does one select and implement call centre technology?
It all starts with an analysis of user requirements, and by someone who understands that users do not need technology, they need solutions. For example, users never need an IVR, but they might need a means to automate the tens of thousands of credit card payments that are being processed by agents each month. Similarly, users never need CTI, but some call centres may demonstrably benefit from having the customer details on screen for the hundreds of thousands of customers who call regularly and whose calls are, on average, short.
The outcome of such an analysis will be a documentation of the functions required, with a business case that estimates both the expenditure required, and the savings and/or intangible benefits that are expected to be achieved. If done well, such a document can be used to select the vendor and system that best meets the company's requirements. It also helps the selectors against glamorous but unnecessary capabilities.
Perhaps most significantly, a thorough user requirements analysis will document what is to become the application. Although true for all call centre technologies, it is particularly the case that buying an IVR system, CTI system or system to integrate the Internet with the call centre have parallels with buying a Fortran compiler. One must still design and write the application.
Kenneth McKenna, Periphonics Corporation's director of CTI products, commented: "From our experience, we often get part way with an implementation and the client realises that they must go through business process re-engineering [to get the benefits of CTI] which has not been anticipated and prepared for."
Although not the only means to select a system, there is much to be said for a competitive tender. Provided it is well written - which many are not - and includes the application (I have issued tenders for IVR systems including the entire menu), a request for tender (RFT) is a sound means for not only selecting such a system, but having it implemented to the point of delivering a working application. This may or may not involve a consultant, but never get any help from a vendor - an RFT written for a specific product wastes everyone's time. And time is money.
Stephen Coates is an independent telecommunications consultant and author of Computer Telephony Integration: Integrating Enterprise Communications (www.ctrcorp.com). He can be reached on email@example.com
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.