All 4 books by Edward Tufte now in
paperback editions, $100 for all 4
Visual Display of Quantitative Information
Beautiful EvidencePaper/printing = original clothbound books.
Only available through ET's Graphics Press:
catalog + shopping cart
All 4 clothbound books, autographed by the author $180
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer:
Visual and Statistical Thinking $5
The Cognitive Style of Powerpoint $5
Seeing Around + Feynman Diagrams $5
Data Analysis for Politics and Policy $9catalog + shopping cart
Edward Tufte one-day course,
Presenting Data and Information
Princeton NJ, August 1
Brooklyn NY, August 2, 5
Philadelphia PA, August 6
Chicago, September 23
Chicago IL, September 24
Minneapolis MN, September 26
Work-arounds and hacks for voice mail menu systems:
-- Edward Tufte
-- David Magda (email)
I agree that there are a lot of terrible interactive voice response systems out there, but I think that automated systems are an essential part of modern life. Few consumers have a real understanding of the cost of dealing with a support call or customer query - I spoke to a Dell senior manager recently and he mentioned that the cost is around USD 40 for a human to deal with a call. When modern goods and services are generally pretty cheap (which is the way we seem to like them) the reality is that having a human route and answer every call is unrealistic.
How do we solve this without driving customers crazy? Well, for a start IVR systems have to be designed a lot better. Usability testing has to be taken a bit more seriously. But that's not enough. IVR will never be a great solution. Companies have to look at all their touch points with customers, in particular their websites and the instructions and bills they send out. Telcos in particular have the habit of sending out complicated bills.
But still that's not enough. Companies really have to simplify their products. A common reason for calling customer service at a telco is to enquire about or to change your 'calling plan'. Most of these calling plans are pure bunkum. There is little or no point to them either from a customer or a management point of view. As far as I can see they are perpetuated by busybodies in marketing and billing departments. Something similar could be said about banks. A lot of the charges that banks make are hardly worth the bother they cause.
Here's my idea: My ideal telco (speaking both as a potential investor, employee and customer) would have only one plan, entitled 'the more you buy, the cheaper the rate'. It would be the same plan, no matter how big or small a customer I was, and my quantity discount would be made explicit near the bottom of the bill. If I had this, I wouldn't need to worry about ringing up and getting the customer service reps to explain the various plans to me.
I think a lot of commentators (including a lot of the comment in the articles linked above) put too much value on human interaction in the delivery of services. The telecomms example is really easy to deal with. In the period after the introduction of direct dialling and before the introduction of competition, telephone customers hardly needed to talk to the telco at all. The demands on the customer service department were much less, because the product was easier to understand (though much more expensive) and the companies were operationally focused rather than sales focused. Equally, most utility services and most goods are provided nowadays with only a marginal level of personal customer service being required.
-- Antoin O Lachtnain (email)
Dell's remarkable inefficiency doesn't provide evidence on this point. Does the "$40 per call" include overhead markups?
This particular "Senior Manager" is confused about how to use his allocation based costing system...
I am going to change professions and start answering and dispatching calls for DELL at 30 dollars a piece.
-- Sandeep (email)
I don't think we have to fret about Dell's ability to manage costs. It isn't a question of having to pay out USD 30 or 40 every time there's a call. It's not a variable cost. The issue is the amount of overheads you have to set up to be able to deal with calls in a barely adequate way.
Let's say you are expecting 100 calls per day, each one lasting 20 minutes (a computer manufacturer's support centre, say). That's 2000 minutes of calls, or a little more than 33 hours, around four shifts. At USD 15 per hour, that's USD 495 per day, or USD 4.95 per call. So far so good.
However, this doesn't take account of the fact that most calls happen during peak times of the day and that certain times of the year are busier than others. In practice this means you need more staff to cover the peaks, otherwise customers will be on hold for hours. Added to this is the fact that support calls are extremely variable in length (it's impossible to predict how long or short a particular call will take) and you will see that it is important to have plenty of capacity to provide even barely satisfactory service. Also, you will probably need to have at least two or three shifts, to cover the evenings, the timezones and if necessary, the weekends.
You also have to think about the cost of expertise. It is hard to get all these different shifts of people trained up to understand complex products and computer systems. In practice, this results in further costs. In a computer support centre like Dell runs, the staff often leave after about a year. Also, it is difficult to retain 'second-line' support who have deep expertise of the products. You have to pay a large premium to keep really good staff and this drives salary costs up considerably.
It is worth reading the Wikipedia article about call centres. It rambles a little, but brings up good points and has a good reference to a paper about the maths of call centres. (http://en.wikipedia.org/wiki/Call_centre). It also brings up good points about the nature of call centre work. Do we really want to build more, bigger call centres?
As with any other company, there are additional overheads involved in employing all these people. You have to provide space, heat, light, health insurance and all the other benefits which typically doubles the cost. All of these factors conspire to drive up the cost of a support call far beyond the cost of the salary.
The answer to the issue is to redesign the product and the process, not just throw more IVR systems or more staff at the problem.
-- Antoin O Lachtnain (email)
This Microsoft-sponsored article says it costs $50 or more for a human agent to field a customer's call:
While this MSN.com article says they are $5 each:
Who is right?
The high number is surprising considering that most of the call centers for large corporations like Dell, AOL, and Microsoft are now outsourced to countries like India, where the average daily wage of a worker is a few dollars.
-- George Subrebost (email)
It depends on how complex the query is. If you ring up and ask where the nearest Kinko's is, it will cost less than if you ring up and ask for guidance through reinstalling the network drivers on your PC.
Outsourcing takes away some of the salary costs, but many of the other costs and issues stay exactly the same or are exacerbated. Also, many consumers consider offshore call centres to be unsatisfactory, to put it politely. Some of the reasons for this are legitimate (cultural clash, lack of understanding of in-country issues, strong accents), some are not (xenophobia, racism).
Why can't we look at some of the mega-businesses that are out there that hardly depend on phone support at all? What about Ebay auctions and Google Adwords, which manage to purvey a complex service with a minimum of telephone support? These should be our model, not the telcos of the 50s and 60s.
-- Antoin O Lachtnain (email)
One of the key reasons for miserable support and annoying IVR systems is that "customer support" is seen as a cost centre for companies, not a profit centre. Or, more appropriately, as an element of its marketing.
So the person in charge of it focuses on minimising the cost. Losing callers to IVR irritation reduces the cost of the call centre. The fact that it might lose a company a customer who's cost thousands to acquire and will earn thousands in profit isn't the problem of the call centre manager.
It would be easy to skip at least one layer of IVR if, for instance, the "bill enquiry number" immediately went through to the right department. Instead companies love having a single point of contact through which we have to fight through a few layers of ill-thought-out IVR.
My favourite is a technique mastered by our largest telco here in Australia. The "please enter your phone number so we can deal with your call promptly" prompt. Guess what the first question that's asked by the operator is. They're obviously well trained as they never so much as smirk when the hundredth customer of the day says "but I just entered that".
Not that I ever think about things like this....!
-- David Glover (email)
These annoying systems just say the company doesn't care about the customers' time and level of satisfaction, so try to change companies -- of course this is not always possible. [The test of the company's opinion of your time is to ask the "customer 'service' 'partner/team member/associate'" if they have a display of how long you have been waiting or have used up to get to them and what the average wait/process time is.] I believe many of these "inefficient" systems are darkly intentional and so must be intended to act as a buffer and/or disincentive, e.g. 1-800-555-1212 "Toll free directory assistance powered by 'tele me'"; which has NOT once understood my requested listing the first time and so always requires at least two (2) requests before I either get a listing or am able to get to an "operator" to overcome their (failed?) voice recognition system. An example of an excellent, efficient automated system is Greyhound Bus Lines routes/times/prices automated information system, 1-800-231-2222, which will inform of route and departure/arrival times and/or fare price with just six (6) choices which can be worked through in less than one minute. j.d.mccubbin
-- jdmccubbin (email)