CT’s Changing Face | Computer Telephony

What began as a tentative effort by PBX vendors to open their switches has turned into a more solutions-based set of technologies.

Part of that is due to the widespread adoption of technical standards for interoperability between vendors and industries. At the basic level, there are standards for the operation of component hardware at the board level. There are also specs put out by individual vendors that enable applications to work correctly on particular board sets — SCSA and MVIP, for example, by companies like Dialogic and Natural Microsystems.

There are also standards, created largely by the computer software industry, for the creation of applications that work with operating systems. The key standards, TAPI and TSAPI, were offered up by Microsoft and Novell, respectively, as a way to push the switch vendors into compatibility so that developers could use those companies’ OS platforms as the basis for CTI applications.

Some of these apps focused on call control (the movement and tracking of calls around a phone network). Many others were apps that took advantage of the growing LAN/phone system connections to bring data to the desktop at the same time as the phone call arrived. Wherever voice and data networks come together, you need standards to assure that the integration goes smoothly (or works at all).

The Internet, of course, forced things to become even more standards-based. Building apps that combined call control and data manipulation became a lot easier with the adoption of Java, TCP/IP and ODBC as standards for data communication.

While most apps assume that there will be a dedicated piece of hardware for the pure telephony switching, it’s becoming clear that nearly all the add-on functions of value — the call center specific applications — can reside quite nicely in a “telephony server” hooked into the phone switch. More and more, that telephony server is a Windows NT box.

In addition to raw interoperability standards, there are more different kinds of links to consider. In any data/voice application, there are a lot of possible combinations. When I say “voice,” for example, I could be talking about a lot of different kinds of “calls” — traditional phone calls, for one, but also recorded calls in the form of messages, fax traffic, even the digits that callers enter when they pass through a voice response system.

As for data, this starts with the host information that resides in the back office databases. But it also includes the subset of host data that moves to the desktop (and back). And MIS data that passes through the corporate LAN, through intranets, over the Internet (including company Web traffic), and emails.

It used to be easy to isolate the two streams — to break it apart and clearly identify what was voice and what was data. Today, though, a corporate CTI system might also be dealing with strange combinations that include elements of both sides: things like voice-over-the-Internet, fax-over-the-Internet, browser-based transaction processing, “call me” buttons that appear on Web pages. Even speech recognition — all these combos are the result of standards that make it easier to push data and voice across each other’s pathways, and that make it increasingly irrelevant in what form a piece of information comes in. What’s more important is how that information is acted upon, and who has access to it.

CTI is now so broad that it is best defined as any technology that combines some form of real-time, person-to-company communication with a background of data that adds-value to that communication.

When I first wrote about this, in the first edition of this handbook, computer telephony was a developer’s technology. It was still being thought out, and built into the core telecom. It needed to be embedded in the switch, or the switch needed to be open to one or more of the “glue” products that sit in between the switch and the applications you wanted to run (“middleware”). Now, that integration is showing up more behind the scenes, as products (particularly applications) for call centers assume a greater degree of CTI readiness on the part of the switch and the center infrastructure as a whole.

The phenomenon of switch-to-host integration represents a total transformation of what you can do with a center. Thanks to this category of product, the most sophisticated call center features are no longer only available to the biggest, highest-volume centers. Small companies can now avail themselves of once prohibitively expensive technology, taking advantage of ANI, DNIS and other network-provided services to do a lot more with each call. This places them on a more level playing field with their most mammoth competitors.

Some of the most obvious benefits that computer telephony offers the call center:

• Have shorter calls. Cut hold time dramatically. Speed information to the agent’s desktop, then to the caller. Reduce your telecom usage costs (the second biggest expense in a call center).

• Have happier customers. Simply put, your reps solve more of their problems the first time out. And faster.

• Make more sales. You have more information about the caller. And, more important, you can bring the information that’s hiding in the corporate database to bear on that particular call exactly when it’s needed. You know what they like to buy, and what problems they’ve had in the past. You can appeal to them on their terms. They don’t get passed from agent to agent. And you can cross-sell or up-sell them while building their loyalty.

• Make better use of staff. Gain efficiencies through blending, and other ways of creating dynamic, responsive group configurations. Slow period for calls coming in? Move some of those reps to the outbound side. A dialer seamlessly starts sending them calls, a script pops on the screen, a whisper prompt in their ear tells them the name of the person they’re talking to. Presto — no more down time. Now, it’s not quite as simple as that, but you get the point; computer telephony puts more information — meaningful information, not just raw data — into the hands of the people who can use it most.

• Improve customer service. Put more information into the hands of the customers, with or without agent intervention. Customers can often serve themselves. This costs less and frees up company resources for more complex tasks.

• Connect with the Internet, or with company intranets, with all sorts of multimedia sales and service tools. All of the traditionally expensive tasks, like order processing, literature fulfillment, interactive faxing, are made easier through computer telephony.

Where Did CTI Come From? | Computer Telephony

Consider, for just a moment, the historical anomaly of telecom. Despite the conventional wisdom about the rise of the microprocessor and the computer revolution, the fact is that the national telephone network built incrementally by AT&T over the course of decades was a feat of computational and networking engineering unmatched in the 20th century.

And yet, when microprocessors begat PCs and PCs begat client/server networks, the companies that made the phone switches for average businesses remained curiously unmoved. Notwithstanding the fact that the business phone system is one complex piece of computational hardware, there was a great deal of resistance to making the phone act more like a computer.

The computer, though, was easy enough to make act like a phone. The computer industry was better at implementing the things that make disparate technologies talk to each other — important things like vendor-independent standards.

Computer telephony has its origins in the fact that if you wanted to add on to a typical office PBX, you had to buy the add-on from the original vendor, or from a third-party company that wrote to the proprietary spec promulgated by that PBX vendor.

Good applications were hard to find because for a software company writing these add-ons, the cost of developing for multiple switch vendors was prohibitively high. If you wanted a reporting system to complement your switch, you had only a few options: buy from the vendor or the vendor’s approved partner, or build it yourself. None of the options were particularly attractive.

Computer telephony was an attempt by the more perceptive members of both the PBX and computer industries to come to grips with the notion that they were more alike than different. Switches were really high-performance communications servers, after all. If the specs could be opened up, if standards could be developed, both sides would benefit from the flood of applications that would be developed.

Today’s switches come with CTI hooks built in, and a suite of applications from the vendor and its partners that take advantage of the connections.

Computer Telephony

Any company’s main focus should be its customers: fielding their calls, delivering service, getting orders out the door, making sales. The easier it is for a customer to get in touch with you, the better the relationship will be. Companies that do the best job of opening the door to customers, those that make it as easy as possible for customers to find out what they need to know, are the ones that have the best track records in the long term. Small and medium-sized companies that have adopted customer-focused attitudes have, over time, become giants of their industries.

Over the past few years there has been much discussion of the pros and cons of a new set of technologies called CTI, or computer/telephone integration (or just computer telephony — they all mean the same thing). Computer telephony was designed specifically to enable better contact between companies and their customers.

It is a loose but complicated amalgamation of interlocking technologies. It isn’t any one thing, not any one piece of hardware or software. It’s a way of combining the two streams of information — voice and data — through open, standards-based systems. It has uses in all areas of modern business, but its most dramatic possibility is in the call center. If implemented well, it can improve the way a company interacts with its customers, which of course, is the whole point behind the call center.

Computer telephony is a way of reaching beyond the traditional limitations of either of the component technologies (phones and computers) and bringing them together in a way that improves them both, by bringing more information to the person on the phone, and making the data behind the scenes much more flexible.

Think about it. The ability to integrate your computer and telecom system could bring the customer’s phone call along with his datafile right to the agent’s desktop, as the call comes in. This translates to massive savings in 800 line charges and agent labor.

In practice, implementing computer telephony has been a dicey proposition. Until very recently, it was largely custom, with each venturesome company taking the plunge using a systems integrator to cobble together all the necessary links, proprietary interfaces and special connections to applications. The benefits are easy to see, but sometimes difficult to achieve. Most call center CTI experiences begin with good intentions. Somehow, they don’t all end up that way. Imagine this scenario:

Your company is facing stiff competition and is growing rapidly, resulting in a certain amount of customer tension — people have a hard time getting you on the phone when things go wrong. It takes too long for sales reps to respond to good leads. Emails come in from customers and go...you’re not exactly sure where. Same thing for fax traffic. You don’t even have time to think about the traffic coming in from your website (and whether your site is connected to your call center).

You hear that there are technologies out there that promise relief. They promise to tear down the walls between you and your customers by bringing voice and data together. You swallow the bait. You hire a consultant and they present you with a plan. Screen pop, says the consultant. When a call comes in, shouldn’t the agent have all the information? Sounds good, you say. Single point of contact, he says. So when a customer calls, whoever handles it has all the relevant info. Makes sense, you say. Links between the switch and the host. Connections everywhere.

Of course it makes sense. And before you know it, you are in the middle of an implementation. The months drag on. The consultant puts a dollar figure on the technology, but once he’s gone from the scene you realize that his number didn’t include things like training, or coordinating what happens in the center with what goes on in other departments. Of course, the technology works, but do the people know how to work the technology?

A year later, you are staring at the prospect of starting all over, with a different set of technological priorities, a different consultant, but the same basic feeling in your gut that yes, you do need to get closer to your customer. You just need a better way to get from Point A to Point B — one that has clearly defined cost and benefit signposts along the way.

For most of the 1990s, installing CTI systems was an incredibly custom job that involved detailed on-site “fixing” to make sure that everything worked together. Luckily, things have changed a lot.

Computer telephony is simply defined as “adding computer intelligence to the phone call.” When you think of it that way, everything from simple screen pop to predictive dialing becomes, at one level or another, a computer telephony application. Depending on your call center’s level of sophistication, and the capabilities of the underlying telecom infrastructure, you may already be using core computer telephony technologies.

What to Look For | Outdialing Systems

• Databases. The dialer must connect to your host system to access your call lists. It’s got to switch between lists and campaigns and it’s got to reschedule callbacks for busies and no-answers.

• Continuing improvement in software and applications. With a clear focus from the vendor on integrating the dialer with inbound switching and with the total customer management software infrastructure. It’s not going to do you much good if a sales person works from one isolated island of information, and the customer service department works from another.

• Call blending. You don’t have to divide agents into inbound and outbound pools anymore. Once, it was necessary. But why should you have outbound agents rushing through a call list while inbound agents sit idle?

What happens when an automated outbound center starts to receive callbacks from the people not reached on the first call? (That happens a lot in collection environments.)

If the ACD and the dialer don’t communicate well, the calls could go to an inbound-only group. But you may want the same agents to handle inbound and outbound. Or a particular account may belong to an agent or a group.

Intelligent features are available for blending inbound and outbound calls on the dialer. You adjust for peaks and valleys by dynamically switching agents from one group to another. It reduces staffing inefficiency and maximizes agents’ talk times more than ever before.

There are two kinds of call blending: reactive blending, where agents are switched around when the system detects an overflow; and predictive blending, in which you predefine when the pools are switched.

• Computer/telephony integration. Dialers should be linked to the host telephone switch, as well as fax and voice mail systems. Monitor ACD activity through server-based connections to the switch. These reports help control workflow in a call blending situation. All this information is vital — how many calls come in, how long they are on hold and when the call center is busiest.

• Business-to-business impact. Predictive dialers are great when you have large lists of customers or potential customers. But they don’t work well when you have to call businesses.

  That’s because companies almost always answer their phones, even if the specific person you want to reach is not there. You lose the efficiency gained by screening out no-answers and busies. The same problem applies to people whose job it is to screen calls.

  If you need to do this kind of calling in conjunction with consumer telemarketing, consider a dialer that lets you shunt several agents into a separate campaign, then run that campaign with the predictive features turned off. Lots of dialers let you do this.

  If you do this kind of calling exclusively, try a sales management software program that controls your lists, feeds the sales person the phone numbers, but lets them keep control of the call from dial to hang-up. This is called preview dialing and it’s common to most contact management or sales automation software programs.

• Do a 30-day test of the dialer with each vendor you’re considering. With a 30-day test, managers will experience not only what the system can do but what impact it will have on your people.

• Make sure the system supports the number of users you need now and in the future. Take a close look at the stability of the company — those that made just dialers in the early ‘90s have been dropping like flies.

• It’s important that there is at least one person within your company whose primary job is to see the project through. This shouldn’t be just a part of their job but a major function of their job.

• Beware of slow transfers. If the prospect has to say hello three times before the call reaches the agent, you might lose the sale before you’ve made the pitch. Your system should transfer the call fast enough for the agent to hear part of the first “hello.”

• Beware of call abandonment. Some lower-end predictive dialers hang up on 20% or more of prospects. The last thing you want is to generate complaints: for people with Caller ID to be calling your center, or calling the phone company, or the state regulators. List owners who receive complaints may stop renting you their lists. Try to buy a system capable of a zero abandon rate. This is a necessary precaution in case future regulations require it.

• Get a system that’s capable of inbound/outbound call blending. That’s the ability of every agent to handle both incoming and outgoing calls from the same station. This leads to higher list penetration and more sales. The outbound agent can leave a message explaining what the call is about. Calls that come back in based on that are more likely to end in a sale. Likewise, inbound-based agents can handle outbound calls during slow periods.

• More leads can mean a lower sales conversion. Perversely, when the supply of fresh leads is unlimited, agents sometimes don’t try as hard for the sale as they do when the supply is limited. Especially with very little wait time between calls.

• Get call tracking and custom list loading. Your system should keep a record of call attempts to be certain that each attempt is at a different time during the day and on different days. Some weaker systems have no call tracking abilities at all, and every time you load the dialers the same people are always called first.

• Keep transition times short. Predictive dialing should reduce the after-call work time (that time between when your agent hangs up the phone and when they are back on the phone) at a given low abandonment rate, say, 2%.

  Only a few predictive dialers can achieve list penetration level of 50% with daytime calling, while maintaining an average wait time between calls of nine to 20 seconds at a 2% abandonment rate.

• Get full branch scripting. The best scripting systems allow data and calculations to be performed in the script and the screen routing is based on the answers given.

  Also when an agent hits a key to change screens, it must be instantaneous, no matter how many people are on the system. A multiple-second wait time is unacceptable. The system should also provide function keys for access to questions most likely to come up at each specific place in the script.

• Ensure campaign flexibility. There should be no limit to the number of campaigns or number of projects that can be called. Every operator should be able to call a different campaign at the same time.

  Make sure it’s expandable. What if your company becomes a big success? Plan for future growth. (I always have to say this, but it’s true.)

So Predictive Dialers Head For Software... | Outdialing Systems

I found one company that put together a custom installation, using their own programmers, with a software-based predictive dialing system and Dialogic boards. It may not be the best system for everybody, but it is definitely possible to get predictive dialing for less than you expect.

Dialing is by definition software. It always has been. For years, the predictive dialing vendors (rightly) competed with one another on features — answering machine detect, speed of answer, and fundamental algorithm — that were software. The boxes were of secondary importance. They were proprietary because you needed lots of processing horsepower to drive those software applications.

Nowadays you want to have more flexibility with your agents, inbound or outbound. You want to link your hardware systems together: switches and computers, dialers and voice systems.

The logic behind it is overwhelming: if dialing features are mainly software, and powerful generic processors are available to run them, there’s no reason why they can’t be part of an overall inbound and outbound call routing system on a client/server platform.

So what should you be thinking about when buying your predictive dialer? Integration — with every other piece of hardware and software in your call center. Mostly software.

What’s happening in the call center now is the marriage of voice and data. Call centers are using open dialing platforms to take advantage of other niche technologies in the call center. It’s a powerful means of taking the benefits of predictive dialers even further.

Why is integration important to call centers? Primarily because of the increased control call center managers have over their technology. Essential call center equipment like ACDs, PBXs and predictive dialers now work in concert, allowing for greater efficiency and productivity.

Companies are driven to make better use of their resources. There are so many technology directions that companies can easily fritter away resources and not really improve the service they provide or the bottom line that they protect.

Smaller centers have basically three options:

1. Buy a turnkey system (which may be proprietary) and build a telecom and computer system around it. This is good for companies that want to dump older equipment.

2. Go for an integrated solution, combining the power of PCs and LANs with software or hardware dialing processors and a phone system.

   Using off-the-shelf parts, you can put together inexpensive solutions. You can grow into it slowly, without sacrificing the dialing features you need: swift answer detection and screen transfer.

3. Or lay a dialing solution on top of the existing telecom and data infrastructure. Your best option here: talk to the vendors who make your existing equipment and software. Chances are you might find a dialer maker among the vendors of your ACD, VRU or call management software.

Software-driven predictive dialers integrate into the call center environment because they’re based on multipurpose minicomputers that let users run other software and because they employ industry standard computer-telephone devices to perform predictive dialing.

Besides predictive dialing, many dialing systems let call center agents perform preview dialing where agents call up data and review it before the call is placed. Preview dialing mostly benefits small call centers making business to business calls.

As for the future of predictive dialers, most agree about the importance of integration. For some call centers, integration means less dependency on mainframes, while others see it as a way to tie dialers into a national database of people who don’t want calls. Even more adventurous is the theory that full function predictive dialing will be possible from an agent’s home phone.

Regardless of what happens in the future, one thing is true: forward thinking has turned a once limited piece of hardware into a versatile and vital piece of technology. As long as that persists, its value in today’s (and tomorrow’s) call center remains undiminished.

But Wait - Aren’t They Complicated? | Outdialing Systems

Yes, and no. The story of dialing in the last few years has been one of a technology that matured, and then was overrun by changes in technology outside the box.

By that I mean that the basic functions involved in predictive dialing (or any other dialing, for that matter) were long ago created and encoded into software. The rest of the cost of a predictive dialer was the cost of the high-powered dedicated box needed to make it happen, and to integrate it into the list system, and to the agent desktop.

It wasn’t so long ago that predictive dialers were a simple purchase — you bought the one that gave you the most talk time per hour, or the one that had the best answering machine detect. What you looked for in a dialer was dialing features. That’s changing immensely.

Like most other hardware technologies, predictive dialers are responding to changes in the nature of the call center. Nowadays you want more flexibility with your agents, inbound or outbound. You want to link your hardware systems together: switches and computers, dialers and voice systems.

More than anything else, you want to choose the software applications that make sense for your business, and get cost-effective hardware to run them. Decoupling the software apps from the hardware is the most impressive development to come along in years.

Predictive dialer vendors, like PBX and ACD vendors before them, have been forced to adapt to a changing world. People are less inclined to choose a standalone system they can’t program and that can only be connected to a limited range of compatible peripherals.

Predictive dialing has always been a software application. It required a great deal of processing power, so vendors put their specialized software onto high-powered computers, most of them with a closed architecture. But the research and development was always geared to better dialing algorithms, more sophisticated call tracking features, and better database management — essentially software apps.

What started as a great idea for outbound telemarketing and collections — fire out more calls than necessary to maximize agent productivity — became the platform on which software companies continued to refine and develop new features for handling calls.

It was such a good idea that companies in other areas (telemarketing software, especially) began adding predictive dialing modules to their systems. The logic was good: if dialing features are mainly software, and powerful generic processors are available to run them, there’s no reason not to create a whole new category of product — the PC-based (or at least client/server-based) dialer.

The traditional hardware/dialing vendors are now changing to match. Several of them have taken their core technologies, enhanced them, and are presenting them to call centers in a new light. They are creating systems for managing all aspects of the call flow. They let agents make calls in predictive mode, and receive incoming calls as well.

To facilitate that, dialer makers have incorporated a technology to blend agents; this allows a single station to handle either incoming or outgoing calls. And although it’s not used widely yet, it’s growing. The dialer is steadily losing its identity as a purely outbound object. It’s got to act like, and interact with, inbound call routing systems. Because it’s increasingly unlikely that a given center will be doing all of one kind of calling, or all of another. Recent information from Datamonitor suggested that the market for outbound dialers was actually expected to increase in the next few years.

Since few call centers are now dedicated to outbound traffic, integration with inbound is the highest priority for the vendors of high-end outbound dialers. Their strengths is clearly in the software that routes the calls, downloads the lists, tracks the results and coordinates the customer information on the back-end. If this sounds an awful lot like the new CIS software, you are right. If it sounds like computer telephony integration, you are also right.

The most successful predictive dialer companies right now — the ones making the most interesting and useful technology — are the ones that have rethought the logic of the outbound call center and recast their dialer as an indispensable component of the inbound and outbound center. For all of them, the selling point is not the power of the dialing engine, but the value-added capabilities of the companion software.

Outdialing Systems

As an outbound call center manager or supervisor, you get more than a little annoyed when your agents can’t reach the people on their call lists. You know it’s not your agent’s fault. Much of their time is taken up trying to get through to a prospect to make a sale or collect a bill, and the longer it takes them to do their job, the more it costs.

Even if you have a small call center, a typical agent only reaches 25 to 35 people per 100 attempts, which could take hours. Enter predictive dialing: automation provides the same 100 calls in about 90 minutes, routing your agent only the ones that reach a human voice.

Today’s dialers are much more sophisticated than they were fifteen years ago. Predictive dialing automates the entire outdialing process, with the computer choosing the person to be called and dialing the number. The call is only passed to the agent when a live human answers.

Predictive dialers screen out all the non-productive calls before they reach the agent: all the busy signals, no-answers, answering machines, network messages, and so on. The agent simply moves from one ready call to another, without stopping to dial, listen, or choose the next call.

True predictive dialing is merely one kind of automated dialing — there are others; but predictive is the most powerful and the most productivity-enhancing. True predictive dialing has complex mathematical algorithms that consider, in real-time, the number of available telephone lines, the number of available operators, the probability of not reaching the intended party, the time between calls required for maximum operator efficiency, the length of an average conversation, and the average length of time the operators need to enter the relevant data.

Some predictive dialing systems constantly adjust the dialing rate by monitoring changes in all these factors. The dialer is taking a sort of gamble: knowing that these processes are in motion, and knowing that there is a certain chance that a call placed will end in failure, it throws more calls into the network than there are agents available to handle them, if all the calls were to succeed.

Sometimes the prediction is wrong, and there are fewer failures than expected. In this case the called party will pick up the phone, say hello, and be hung up on when no agent is available. One of the intricacies of predictive dialer management is fine-tuning the aggressiveness of your dialer’s algorithm.

Predictive dialing has been nothing short of revolutionary in the outbound call center. When agents dial calls manually, the typical talk time is close to 25 minutes per hour. Most of the rest of that time is non-productive: looking up the next number to dial it; dialing the phone; listening to the rings; dealing with the answering machine or the busy signal, etc. Predictive dialing takes all that away from the agent’s desk and buries it inside the processor.

When working with a predictive dialer, it is possible to push agent performance into the range of 45 to 50 minutes per hour. I’ve heard of centers going as high as 54 minutes per hour. (You can’t really go higher than that, taking into account post-call wrap up time.)

There is more to the technology than just the pacing algorithm. Predictive machines excel at detecting exactly what is on the other end of the phone, including the ability to differentiate a human voice from an answering machine. They typically decide that the call has reached a person within the first 1/50th of a second — the start of the word “hello.”

Here are just a few of the important ways predictive dialing systems can help you.

• They completely automate outbound consumer calling. That includes the actual dialing, assigning agents and controlling the list you call from.

  You can run multiple inbound and outbound campaigns, and you can specify names on a list not to call. It also schedules automatic callbacks for nonproductive calls. Dialers let you set the parameters for the dialing algorithms to meet the needs of a particular campaign, like the percent of overdials the system sends out.

  With collections applications, for example, you may not care if the dialer has to hang up on a “customer” if there is no agent available. You’ll trade the customer’s good will for a higher volume of calls. But for a sales promotion, you’d want to keep those hang-ups to a minimum.

• You can manage your call center more effectively.

  Standard features include real-time statistics about how each agent, group of agents or list is performing. Also, trunk pooling, which reduces operating costs by processing both inbound and outbound over the same trunks.

• They reduce agent burnout and turnover. Just imagine all the tedium they avoid: finding the phone number, typing it in, waiting for the phone to connect and the number to ring.

  The dialer makes sure that the only calls an agent has to deal with are real calls, with a live customer on the other end. No busy signals, no endless ringing, no answering machines.

  Cutting out that stalling doubles the time spent talking on the phone. Talk time, which is about 20 minutes an hour without a dialer, jumps to 40 to 50 minutes with one. Agents like their jobs better when they don’t have to wait around for the phone to be answered.

• Reach more people in less time. You penetrate lists more deeply in a fraction of the time.

Predictive dialers adjust the balance of agents from one list to another, taking into account factors like list performance, time of day and the success of particular agents.

What They Have To Offer | ACD

There are literally hundreds of companies out there that sell switching and routing systems for call centers. This is just a scratch-the-surface look at some of the major companies, particularly those that tend to be market leaders in developing new features and seeking out third party partners. These few companies sell the lion’s share of standalone ACD systems.

Aspect. In late 1997, Aspect radically changed the platform underlying its key switch offering. The change was partly one of positioning, and partly one of hedging technological bets as the call center industry becomes less tolerant of proprietary technology.

The core ACD, formerly called the CallCenter system, now known simply as the Aspect ACD, incorporates two new “platforms” for call processing. The first, called the OpenMedia Module, is built out of MVIP and SCSA cards. Co-existing with that inside the box is another new component, called the Integrated Applications Module — a standards-based (NT, Pentium, Oracle) processing platform that’s scalable.

This combination, it is hoped, allows Aspect customers to use a broader range of software applications, including some that appeal to the smaller, departmental or distributed call centers. In theory, you can link multiple IAMs off the main unit to create an ever-growing chain of connected applications, all processing in tandem.

At the same time, they brought out a construction and maintenance tool, the Aspect Architect Software. This places all the call routing templates on a single drag-and-drop desktop, including such things as IVR integration and call delivery channels from multiple media.

Shortly thereafter, seeking to position their switch as a serious one for the growing electronic commerce/Internet call center market, Aspect debuted its Web Agent software — a real-time browser-based information sharing tool that can be added to a call center for about $1,000 an agent, including CTI connectivity. Web Agent lets the person on either side of a Web transaction navigate through different pages, and gives the call center rep the ability to guide the customer through a series of screens according to a script. It works through a choice of multimedia options, including a desktop IP connection, Web-based text chat, or a traditional two-line callback using the public network.

The single most important thing about this software release is that it is an open system. It will work with ACDs other than Aspect’s — Lucent, Nortel and Rockwell, and connects on the other side with existing IT and telecom infrastructures. The advantage to Webifying an existing call center seat is that it allows you to leverage the already-trained rep and equipment to sell existing products — even complex ones — to someone that might not have even been a caller. Converting Web “lurkers” into callers is the first step to turning them into customers.

People used to talk about the “call center in a box,” as if there would come a day when a single vendor would create, and shrink down, a total application, hardware and software, for running a call center. Instead, it looks more like a collection of cherrypicked integrated apps, certified to work under the aegis of a single vendor, will do the job just as well.

For switch vendors, the imperative to grow by adding features has in years past created a spiral of really interesting technology that often goes nowhere operationally — fantastic ideas (skills-based routing, universal agent blending, “call-me” buttons on Web pages, etc.). However, they can’t be made to work in the real world because there are so many operational and cultural hurdles on the ground in real call centers.

With fierce competition on features, and a frightfully high cost of development, it’s only natural that a larger company like Aspect would find it more logical to add value to its switch through an aggressive acquisition strategy. For the small companies that develop call center and CTI apps, the costs of developing in a multi-vendor environment are so high, and the need to spend huge sums on marketing to get the word out (even when the technology is superior) make partnering up a no-brainer. One large software company’s CEO told me that he gets — and turns away — at least one request a day to join his firm’s application partnership program.

They’ve come to the conclusion that in a world of open systems, it’s better to be known for the enablers and the apps than the switches, and they are probably right about that.

Intecom. This Texas-based company has been a champion of the little guy, making advanced features available to the small- and medium-sized call center market. That’s not to say large users are left out in the cold. Their switch supports up to 8,000 agents. Addressing the need for increased call processing and application options planned for high volume call centers and PBXs, Intecom introduced its E Millennium Server in early 1998.

The original E was conceived as an incremental system, the kind that you buy a small piece of when you’re small, and then add to — emphasizing the multi-site linking capabilities.

Tremendously increasing the processing power of its communications platform, the Millennium Server has the sustained ability to complete over 500,000 calls and support 3 million CTI messages per busy hour in optimum configurations.

Serving up to 8,000 agents in a non-blocking, single database system, the Millennium Server supports an Ethernet CTI interface with 63 IP addresses to handle the most intense CTI capabilities in the industry.

Intecom adds to the E switch with an NT-based application platform, called CallWise Centergy. It uses SQL Server as the database upon which it serves up statistics, reports, and customized templates for seeing what’s going on in a call center, and making changes based on that information.

It traces call event details from start to finish, supporting single or multiple call centers from a single administrative point. (The reports are provided by Crystal Reports 6.0). The whole system is standards-based, extremely open and promises to bring more custom options to managers who use the high-powered E switch.

CallWise Centergy captures call center events second-by-second in an ODBC-compliant relational database to construct the information and statistics needed for effective call center management and quality customer service. The GUI reporting interface allows users to configure, layout, schedule and run reports. The single solution also provides data redundancy and fault tolerance, recovering data from the main Intecom platform after any loss of communications. The scalable Windows NT server allows customers to size the system for current requirements and expand it as their needs grow.

Lucent. Lucent is a spinoff of AT&T. In the breakup, AT&T got the wires, and Lucent got the phone systems. The Pelorus Group claims that Lucent led their call center rivals in 1997 in both revenue and agent positions shipped. In fact, Lucent shipped the most agent positions in five of the six market size segments they tracked, especially at the higher end. At the very highest end (centers of 151 agents and up), Lucent shipped a reported three times as many agent stations as its nearest competitor.

Though it’s almost impossible to pin down with any certainty, another study showed that Lucent appears to have the highest market share among call center switch suppliers to the US market. A Dataquest report entitled US Call Centers Market Share and Forecast showed that Lucent’s domestic shipments increased from 197,000 agent positions in 1996 to 251,000 in 1997.

Apparently, Lucent has 30%+ market share in several categories: small (more than 21 agents), middle (more than 75 agents) and high-end (401 agents or more). In that last segment, Lucent was reported to have a 52% market share.

However you count it, there’s no getting around the fact that Lucent’s Definity is one of the most popular switches for call centers.

The Definity ECS Call Center is the core switching platform on which Lucent overlays a very broad software suite called CentreVu. One application, for example, is called CentreVu Exchange. It can assemble info from as many as 20 Definity switches and 40 multi-vendor voice response units. It can tell you exactly who touched a call, even if that call was transferred from site to site. It also tells you the length of time a call spent in a given application. One use: gauging which voice prompts have the best effects under different circumstances. This reporting tool lets you see exactly what those circumstances were.

It’s so precise, for many call center managers this will open up a whole new way of looking at data — drilling down to the rawest call path information, rather than looking at aggregates. It’s being touted as an alternative do data mining, as more of a way of selecting out “boutique” batches of data.

The CentreVu Call Management System provides more than 100 real-time and historical management reports designed to help you achieve critical sales and customer service objectives, while boosting the productivity of your call center employees and resources. It offers extensive historical data storage capabilities, keeping detailed intra-hour data for up to 62 days, daily reports for up to five years and weekly and monthly data for up to ten years.

An optional Forecasting package allows you to use trended data and “what if” growth scenarios to forecast the number of agents and the economic break-evens of staffing and abandoned calls.

Their CentreVu Call Management System, which provides call center operation performance measurement tools, will, from a centralized site, use graphical tools to manage all agents connected via the ATM backbone. For CTI apps, customer records in one city will be instantly accessible by agents in other locations. Enterprises will no longer need to concentrate on integrating or linking databases and CTI applications.

The solution includes the integration of the CentreVu line with its Definity ATM switch and PacketStar Access Concentrators, which enable voice, data and video traffic at each call center location to be supported over the ATM backbone network.

Lucent also has analytic tools for call centers that combines center data with information from outside, in the rest of the enterprise. CentreVu Visual Analyst is software that helps companies understand the correlation between call center activity and business results. It collects information from a company’s multiple databases and presents it in easy-to-use interactive, graphical formats. CVA lets managers look at their call center performance data while simultaneously viewing data from other parts of their business.

With its integrated reporting capabilities, users can analyze complex networked databases to make smarter and faster enterprise-wide business decisions. It helps uncover trends, patterns and problems buried in large amounts of disconnected data. It works in conjunction with call center performance reporting systems, such as CentreVu Call Management System and CentreVu Explorer. A bank might learn from its report, for example, that agents are making strong sales on a loan product to a certain customer segment. At the same time a database outside the call center, with company revenue information, might indicate the bank is receiving more profit on that product from a different customer segment.

Basically, the more information from various sources that you can combine in a meaningful way, the better your call center will perform from a whole-business standpoint.

Nortel Networks. Nortel’s Symposium Call Center Server is a call switching system that sits inside a Windows NT server, and that allows users to layer a suite of NT apps on top. (The suite itself has been bundled with a host of other products, notably the WebResponse Server and an IVR, into a portfolio called Symposium Internet Call Center.)

SCCS puts the switching into a client/server box (an NT one, to be precise). That makes it an interesting choice for small companies looking for a platform upon which they can grow their center.

For the small center, this system offers a wide range of benefits, including advanced call control and reporting, skills-based routing, a good scripting language, and the ability to network existing call center apps with new Symposium sites.

Perhaps most important, Symposium is open to connection with third party products (Nortel has been a leader in this area since the days when the Norstar phone system was the most open, call center-friendly key system around).

The Symposium system can be networked together with existing Meridian or Norstar call switching systems (so companies expanding their call center infrastructures can do so incrementally). Capabilities included in the Symposium are skill sets (that is, a form of skills-based routing), and ease of integration with third-party products through open interfaces like ODBC, Host Data Exchange, Meridian Link and Real Time Interface.

The Symposium Agent software offering acts as something of a bridge between older platforms and Symposium — it runs on Win 95 as well as NT, and integrates with other Nortel call center products, such as the Call Center Server and the Meridian 1 communication system. Calls come into the Meridian 1; the Call Center Server routes them to the most appropriate agents; and Symposium Agent helps agents manage the customer interaction.

Further evidence that all the various streams of messaging input will ultimately be collapsed and handled by one “super-ACD”-type of box: Nortel’s enterprise messaging system, called CallPilot incorporates speech rec, call management, CTI hooks and an app gen for building the kinds of apps you’d need for a call center.

While it’s unlikely to be a call center app in and of itself, unified messaging (especially on the enterprise level) is likely to be the kind of glue that connects the business’ phone system with it’s data messaging traffic.

It’s likely to provide some of the structure for creating small-scale departmental call centers that are needed for specialty applications, or that connect to larger, standalone centers.

CallPilot manages voice, fax and email messages. It integrates with a wide variety of email clients and creates a multimedia mailbox. And you can use it with voice commands, saying “play” or “print” to activate functions.

While that kind of thing won’t be of much use to the call center rep, what will be good will be the background call tracking, and things like the TAPI and S.100 interfaces, which provide the hooks into coordinated call center apps.

If an app is intelligently designed on top of this, it literally allows any desktop or person in the company to act as (or interact with) a call center agent.

Rockwell. Seeking to grab a portion of the underserved, but exploding, small call center market, Rockwell took its call handling expertise and shrunk it down to the size of an NT box. Transcend is a call management system that’s based on their larger Spectrum switch; it runs the same software, but scales much smaller. A complete system can be had in 10, 20, 40, and 80 agent position configurations.

Upgrades are available in 10 agent increments. Transcend is also available in a 20, 40, and 80 agent position kit form, which includes all the software and just the ACE 360/ISA-2E processor board.

It’s built on an Oracle Universal Data Server and Dialogic’s CT Media middleware in a layered approach that separates out the application, switching, routing and call handling functions, making it easier to integrate the whole thing into existing data and telecom infrastructures. That aims it squarely at the informal and departmental call center.

The open, standards-based approach ultimately makes it easier to add on IVR or Web or any other data-oriented application, or connect to existing apps (which is the more likely case in the smaller center).

Rockwell has historically had leading edge technology but a hard time defining exactly where that technology fit into the real world activity of the call center. The small center market, however, needs no rationale; it’s a clear growth area in an industry that many are now seeing as somewhat mature.

The Spectrum is a switch that serves centers of between 50 and 150 seats and somewhat larger, depending on which model you use. All Spectrums, including the mid-sized version, incorporate Rockwell’s software like the Telescript Graphical Editor (for designing call flow), Total Recall Reports (for historical feedback on call center activity) and remote agent features.

This switch connects with other open Rockwell products like the Call Center Command Server and Transcend, based on the S.100 standard.

Perhaps it’s a sign of how far we’ve come, but this version fits inside a cabinet the size of a PC server, and has a greater degree of “plug and play” software configurability than older versions.

The Call Center Command Server (3cs) is a software suite that sits in front of the Spectrum. It’s a development tool for creating mini-apps that coordinate the flow of data in and out of the call center. Completely open and object-oriented, 3CS functions as the interface between business apps and call center systems. The control platform’s open architecture features reusable object-oriented components such as Microsoft’s ActiveX Controls, making it easier for call center supervisors to rapidly develop custom, enterprise-wide applications.

One component, an app, called Commander, consolidates separate call center data elements into a single control point, letting supervisors, managers and administrators immediately respond to changes within call centers. Commander also graphically manages the dynamic relationships between agents, supervisors and applications. Information screens are customizable and scaleable. Administrative features include the ability to view equipment configurations for the call center. Application telescripts can be created and modified efficiently, in both text and graphical modes.

Small Center? No Problem | ACD

First, let’s start by thinking of the needs of the smallest centers, those for whom the purchase of an expensive, standalone ACD is too much to handle. Luckily, call routing is available as part of the PBX configuration, bringing large call center tools down to the level of even the smallest centers.

Think of the five or ten person collections department, the customer service area of a larger company. They have many of the same needs — and problems — as larger centers. But until now, there have been few call handling tools that deliver state of the art features at a reasonable price.

Their personnel are not always dedicated phone reps. They need flexible solutions that build on the systems already in place, that give them room to grow without putting the company in the poorhouse. The response to those needs is a new variety of call handling system — the ACD without the box, or the PC-based ACD. Thanks to the new-found openness of switch vendors, developers are offering a host of software products that add ACD features to key systems and hybrid switches.

For one thing, it’s far less expensive to bring ACD features into an existing business phone system. There’s no capital expenditure on a big piece of hardware. With larger ACDs, it’s very difficult to justify at the six-agent size or, for that matter, anything below thirty agents. It’s also a lot more flexible than it used to be. You can easily integrate top-notch systems like interactive voice response or voice mail, giving your small center a highly professional appearance.

Critical to call centers is that you be able to add third-party call control. You don’t need to know how to program to set up a rule-based system for getting the right call to the right agent. With the PC it becomes a low-end solution. You can do a lot of things, like provide special treatment to customers based on the language they speak, route calls based on skill sets, or based on time of day for full 24-hour coverage.

What the PBX ACD does is let you dabble in call centers without having to go full bore right away. Make no mistake — one of these low-end ACDs built off a PBX will only get you so far. If you’re going to grow beyond a certain point (50 agents is a good ballpark), then explore the larger, standalone systems. At least explore the low-end offerings from those vendors, because those vendors are offering much smoother upgrade paths than ever before, hoping to capture some of the small center market.

PC or PBX ACDs allow small installs (typically 10 to 15 people) to be placed on the same technological plain as bigger centers. Since many companies already have PBXs that can be enhanced with available software, they can dabble. It’s possible to convert a few users and then decide that if things go well (they usually do), to expand further.

For example, Cintech offers Prelude, by its very name a starter system that encourages people to step up to Cinphony, Cintech’s more advanced software ACD. Prelude is aimed at retail stores, pharmacies, universities, car rental places — places that until now might have gone without features like call categorization and advanced routing.

Comdial’s QuickQ software lets you set overflow patterns between multiple small groups, and lets you change those parameters quickly, with a minimum of required knowledge.

These systems do not deliver everything you’d expect if you used a dedicated ACD, but they don’t have to. Departmental needs are different. Few need multi-site routing, for example. Department heads (who may not be telecom people) need different kinds of reports that have more to do with sales and costs than with call traffic.

There are many small call centers that are just beginning to realize they are call centers. And that they need the same kind of technologies big centers have been using for years. Customers demand the same kind of service, no matter how big you are. This small-scale solution lets them reach more of their potential for pleasing customers at reasonable cost.

Automatic Call Distributor

Automatic Call Distributor. A simple way of describing not just a piece of hardware, but really everything that goes on in a call center. That’s the basic function, anyway: taking incoming calls and moving them to the right place, the agent’s desk.

Over the years, things have changed. The ACD is responsible for more than just moving calls. That’s probably not even the best term for it anymore; I should probably be using something closer to telephony server, though that term is already in use for the LAN server that moves call control commands from client workstations to the attached ACD or PBX.

Now, the ACD’s job is not just to route calls, but to manage the information associated with those calls as well. “ACD” is really a function that can be carried out by a wide variety of different kinds of processors.

At the very low end, you can buy a PBX that has “ACD” (read: call routing to agents) built in, or you can choose to add it on through a PC application. (Yes, the switches are generally open enough to third-party or other add-on apps.) You can route calls in the network, thanks to intelligent features built into the carrier networks.

The ACD is the heart and soul of the modern call center. It is the engine of productivity — the single piece of technology without which the whole edifice of inbound sales, order taking and customer service all crumble. What the ACD has done is enable the volume of calls you take to escalate intelligently, and in ever more specialized complexity. It has matured beyond call routing. It is the brain and control point for the call center, for both inbound and outbound, for voice calls and data traffic. It’s a call center’s arbiter: setting priorities, alerting supervisors to patterns and crossed thresholds.

Once the term “ACD” meant a very specific type of telephone switch. It was a switch with highly specialized features and particularly robust call processing capabilities that served at least 100 stations (or extensions). It was purchased mostly by airlines for their reservations centers and large catalogs for their order centers.

Companies with less specialized needs bought different technologies that didn’t offer the same specialized features. Currently, true ACD functionality is found in telephone switches that range widely in size and sophistication.

Today, there are PC-based ACDs, key systems with ACD functions, key systems that integrate with a computer and software to create a full-featured ACD, PBXs with ACD functions, PBXs with ACD functions that are so sophisticated they compete with stand-alone ACD systems, stand-alone ACDs that serve centers with less than 30 agents, traditional stand-alone ACDs (don’t misunderstand, these are usually the most sophisticated), ACDs that integrate with other call center technologies, and nationwide networks of ACDs that act as a single switch.

There is simply no technology more suited to routing a large number of inbound calls to a large number of people than an ACD. Using an ACD assures your calls are answered as quickly as possible. It can provide special service for special customers.

ACDs are capable of handling calls at a rate and volume far beyond human capabilities, and in fact, beyond the capabilities of other telecom switches. They provide a huge amount of call processing horsepower. Using an ACD assures your human resources are used as effectively as possible. It even lets you create your own definition of effectiveness. An ACD gives you the resources to manage the many parts of your call center, from telephone trunks to agent stations to calls and callers to your agents and staff.

With all these call handling options, many of them surprisingly open and modular, why would anyone still want an expensive standalone ACD for their center? Two simple reasons:

Power. Nothing has more raw call processing ability than a first-tier standalone. Nothing else is so uniquely suited to the needs of today’s reservation or financial service megacenters.

Technology. When it comes to integration with other call center systems, like IVR, data warehouses and intranets, nothing can beat a powerhouse ACD. The same goes for multi-site networking and skills-based routing — two of today’s most sought after inbound features.

No knock on smaller systems like PC-ACD and PBX/ACD hybrids (which account for much of the industry’s phenomenal growth in small centers), but there is no substitute for the call-crunching strength of a standalone ACD in many high-volume applications.

The ACD is being changed by two dramatic trends in the call center. First, it is being asked to channel more information, of many different kinds, in more directions. A decade ago, there were two kinds of information: the call itself, and raw log information about the calls in aggregate. Little else was needed, and if you did need more details about what was coming through the ACD you could analyze the data (which often came out a serial port) on your PC with a cumbersome third-party tool.

Now, call center managers need information — presented in a form that makes it easy to grasp quickly. High-end ACDs vendors have added data management modules at a rapid clip. Also there are many outside programs that can connect to the ACD and funnel data in and out. These include workforce management tools that forecast load, and software systems that put real-time and historical data into any form needed (like reports and readerboard displays).

ACD vendors are improving the tools the supervisor has to tweak the ACD while it’s in motion: things like creating groups on the fly, moving calls and personnel around, monitoring for quality, to name just a few things.

The other dynamic change is in what kinds of calls the ACD has to route. Sometimes this is referred to as alternate methods of call delivery. Call centers have been integrating ACDs with IVR for years, and the same goes for the combination of ACD and fax. Now I’m seeing vendors grapple with the Web and the Internet, with calls that come in from PCs and that terminate in databases instead of agents.

The call center itself is giving way to something more amorphous known as (for lack of a better term) a “customer contact zone,” in which what’s important is the transaction between the customer and the company, not what wire that transaction passed through.

Another thing that’s shaken up ACD design is skills-based routing. At first, this was a feature that was added to switches more because the technology was possible and cool than because call centers were clamoring for it. It’s taken a long time for call centers to figure out how to make it work because (for reasons that I’ll go into in more detail in another chapter) skills-based routing rubs the wrong way against the proper use of workforce management software.

Be that as it may, skills-based routing is a highly interesting and advanced system for distributing calls that come into an ACD. Traditional routing is based on two factors — an equitable distribution of calls among available agents, and the random nature of incoming calls. Skills-based routing changes this somewhat: it routes calls to the agent “best qualified” to handle the call, measuring “qualified” by agent parameters you set.

The ACD does this in two steps. First, some front-end technology must be used to identify the needs of the caller. That’s usually accomplished through DNIS, ANI or an IVR system. Then that information is matched against the sets of agent skill groups. There are two ACD advances that let you run skills-based routing effectively:

• Leaving a call in an initial queue while simultaneously and continuously checking other agent groups for agent availability;

• Or allowing an agent to be logged on to more than one agent group (in this case a skill group) at a time, assigning priorities to those groups by skill type.

And finally, changes have been hastened by the need and desire to link call centers together into multi-site call center networks. In some ways, this is operationally an extension of skills-based routing: it’s not enough to choose the best available agent — often you have to choose the best available agent at the most appropriate location (based on time of day, traffic at one or more sites, skill clusters or call priority).

To a greater or lesser degree, vendors of standalone ACDs are pushing the technology envelope with their switches. Some are concentrating on software development to add value to the core switch. Others are paying more attention to integration with third-party call center technologies like the Internet and IVR. Still more are adapting their switches to smaller, departmental call centers, hoping to catch some of the growth in the industry that way. In all, it’s created a dynamic atmosphere — one in which if you want a feature, all you have to do is ask.