A few weeks ago I purchased a very expensive piece of software from a company I'll call Baked Mud Software. As part of the purchase I am entitled to get a free upgrade to the next major version, which was released just last week. Here are the steps I've gone through so far (more or less):

1. Look on the company's website for upgrade information. Fail to find it.
2. Use Google to find a web page buried deep inside Baked Mud's site with upgrade instructions. So far so good.
3. Fill out a web form, including a copy of my scanned receipt. Click "Submit." This is the point where I descend into Customer Service Hell.
4. Get the page: "We're sorry, we encountered an error processing your request."
5. Click the "Broken link? Send us an E-mail" link.
6. Get the generic top-level "Contact Baked Mud" web page.
7. Click the "Contact Service and Support" link. This goes to a generic web page about how to search the help files and access the online forums. In other words, useless.
8. The other link on the "Contact Baked Mud" page is labeled "Visit Product Help and Support Centers." This link goes to another generic web page about searching help files and accessing forums. Also useless.
9. Try going back to the web form from step 3. Now the form itself is unavailable and gives the same error message.
10. At some point, a popup window appears inviting me to take a survey. I give Baked Mud a "1" on all 20+ questions. I also provide my e-mail address and a long description of the problem I'm having in the comment box.
11. Nothing happens in response to the survey in step 10 (it's been over a day).
12. Several hours later I go back to the form in step 3. The form is available again, but still gives an error when trying to submit it. Then the form becomes unavailable again.
13. Oh look, there's a box in the window offering online chat support.
14. It also says that all chat agents are unavailable and won't let me click it.
15. Try steps 3-14 a few more times over the course of two days.
16. Finally give up on the website. After clicking around a while, I finally find a phone number for customer service.
17. The first message I hear when dialing customer service instructs me that if I'm trying to get my upgrade, I should go to a particular web page. This is the same web page which has been broken for two days.
18. Spend about five minutes navigating the phone menus.
19. Spend about ten minutes on hold.
20. Finally reach a person. He's nice and his English skills are good, but he clearly isn't a native speaker. I spend five mintues spelling my e-mail address before he gets it right.
21. After explaining my problem, the response I get is "our systems are really busy, I can't submit this request for you, try the web page again in a week." To his credit, the customer service representative is very patient and doesn't try to rush me as he does absolutely nothing to help.
22. Give up. For the time being.

I understand when a big company has a major product release things can get crazy. But some companies manage events like this with a whole lot more grace.

The irony is that Baked Mud recently won an award from a very expensive consulting company for its "Voice of the Customer" program. I'm not sure what they did to earn that award, but it's pretty clear from where I sit that there's still considerable room for improvement.

In our business office, we've used Comcast's business-grade cable modem service for Internet access since 2007. In general we've been satisfied: the service is much faster than the DSL connection we had before, and the price was reasonable and exactly what we had agreed to in the contract nearly five years ago.

Over the past year, however, we've noticed the quality of our cable modem connection declining a little. Where it used to always be wicked-fast, we now see occasional slowdowns and overnight outages.

Then this month, a surprise: a new $7 fee in our bill for "equipment rental." $84/year is not going to make or break us, but as a general rule you should only pay bills for money you actually owe. Companies which break this rule tend not to stay in business.

So I dutifully called Comcast customer service to try to get to the bottom of this. It seemed a little strange that our five-year-old cable modem was suddenly worthy of a monthly fee when it hadn't been for so many years before. Here's more or less what happened:

1. I called the phone number printed on the bill. It might be reasonable to expect that someone calling the phone number printed on the bill might have a billing question, but I had to spend about five minutes navigating the menu to find the option for billing problems.
2. I then waited on hold for fifteen minutes. Normally one might expect that business customers calling with billing disputes are both high value and upset, but Comcast apparently feels the time of their customer service representative is more valuable than mine.
3. Once I got to speak to someone, the representative told me the new charge was because of a "policy change" which had been disclosed in a prior bill. I dug up the old bill. There was one sentence in 8-point type on the back page of a bill from six months ago (I am not exaggerating this in the slightest). That's not disclosure, that's active concealment.
4. The Comcast representative said he couldn't do anything about the charge, but he could transfer me to a "retention specialist" who might be able to do something. Keep in mind that by this point I had been on the phone for nearly half an hour.
5. When I told the retention specialist I wasn't happy about having new fees randomly added to my bill, he first tried to argue that Comcast had the right to add the fees because of a contract on Comcast's website. Leaving aside the legal question of what makes a binding contract, getting into a "Yes I can/No you can't" argument with a longtime customer over less than a hundred bucks a year is not going to win any loyalty.
6. Finally, after telling the retention specialist that I was absolutely not going to argue with him about whether Comcast's website constitutes a legally binding contract, and I just wanted the fee removed, he said he couldn't take the fee off. Instead, he could offer me a $10/month credit for 12 months.

In general I don't have a problem if Comcast wants to push through a modest price increase. They have never raised their price to us in almost five years, and on the whole they've provided good value. But if they want to raise their price, they should do it the right way: send us a prominent notice or an insert in our bill announcing the change. Trying to sneak in a new fee like this is just, well, sneaky. Instead of looking like a legitimate price increase it looks like a mistake, which I had to spend a fair amount of time tracking down. Then when it wasn't a mistake it just looked sleazy.

As it happens, there's a new carrier pulling fiber to our office building this summer. A week ago I wasn't really interested in switching from Comcast. Today I am.

Many companies like to collect customer feedback on a regular schedule: monthly, quarterly, yearly, etc. This is a good idea, but it's an even better idea to spread that survey out over the entire period.

For example, if you plan to measure your customer service performance through 450 customer surveys once per quarter, you should not do all the surveys at once. Instead do 150 surveys per month, or about five per day, each day of the quarter.

The cost should be about the same, but spreading it out like this has several advantages:

1. Continuous surveying can uncover problems as they are developing, rather than at the end of the reporting period. You can catch things when they're small and nip them in the bud.
2. An intermittent customer survey is more disruptive to the organization. The quarterly or yearly customer survey is a project which everyone has to make room for, but an ongoing survey is simply another business process which goes on day to day.
3. Everyone tends to be on their best behavior during "survey week." Even if there's no direct compensation for getting good scores, people still want to do well.

Free response questions (aka open-ended questions, aka verbatims) are a powerful tool for finding out what's top-of-mind with customers. The traditional method for dealing with them is to have an intern perform the mind-numbing task of reading each comment and categorizing it, in order to do some statistical analysis. This is what we did for the recent NCSS report on customer complaints, though I and one of my managers played the roles of interns.

In the past I've played around with some text analysis software to try to reduce some of the burden of this chore, and found that it didn't help much for the size of data set we're working with (usually under 10,000 free responses). It took a little too much time and effort to train the software to give good results. On the other hand, if we had hundreds of thousands (or even millions) of comments I could definitely see the value.

For surveys in an interview format it makes sense to have the inteviewer provide a preliminary categorization of the customer's comment. The interviewer won't know as much about the company's internal business processes as an insider, but will probably get it right 90% of the time. The interviewer can also flag ambiguous cases for further review.

There's also a limit to what you can get from a free response question. When you ask for comments, most customers will give you just the one thing most important to them at that moment. So this is not a very good way to track the actual incidence of specific problems, since you miss customers who may have experienced the issue but didn't happen to mention it.

So when something interesting develops in the comments, it's a good idea to start asking about it specifically in the survey. You will get a much better read on whether there's a problem with (for example) language skills or the IVR if you ask every customer whether they experienced that issue or not. If it turns out you don't have a problem then the question can be removed.

Issue #56 of Vocalabs' newsletter, Quality Times, has been published. This issue has some more discussion of the just-published NCSS Cross Industry Report on Customer Complaints.

As always I hope this is interesting and informative, and if you like what you see please feel free to subscribe to the newsletter via e-mail.

One graph we did not include in the recent NCSS report is the complaints about how hard it was to find the phone number to call customer service (at right).

In our research we found a lot of these complaints, and it wasn't surprising. Companies have been "hiding the ball" for years, and it's rare to find a company's phone number on the homepage of its website. Customers also frequently complained that they couldn't find a number to call on their statements or bills.

The differences between companies in these complaints was statistically very significant, with Verizon, Chase, and AT&T posting particularly large numbers, and this was clearly a source of great frustration for some customers.

As much as I like this data, I cut it from the final report because we recruit NCSS survey participants by advertising on Google. That means there is likely a bias towards customers who had a hard time finding a way to reach customer service. The recruitment process for each company is identical, so I think the comparisons between companies are valid. Nevertheless, I can't prove that. So, regretfully, I felt I had to cut one of the more interesting comparisons from the published report.

We have just published our most recent National Customer Service Survey report. This quarter we're doing something a little different: we've done an analysis of all the customer complaints from the past 15 months (since the beginning of 2011) in all the companies we follow, and our Executive Summary shows what kinds of complaints are most common at each company.

You can download the Executive Summary, and contact us for more information about how to subscribe to the underlying data.

In business we talk a lot about "best practices." The general idea, which is a good one, is that industries can develop and standardize certain ways of operating which are demonstrably better than the alternatives.

I have some problems with how this works in the real world. Ideas promoted as "best practices" are rarely backed up by the kind of rigorous research the phrase implies. Too often, a "best practice" works well for some companies in some circumstances, but doesn't work elsewhere. At worst, "best practices" are sometimes nothing more than interesting ideas from consultants looking to drum up business. Perhaps my biggest complaint, though, is that the phrase "best practice" implies that the "best practice" can't be improved and that no alternative approach would be better.

I think most businesspeople understand this, and know that every situation (especially in the customer service world) is unique. What most companies are looking for from "best practices" are ideas which have worked elsewhere and might be adopted (or adapted) in their own operation where it makes sense.

So I no longer talk about "best practices" with my clients. Instead, being from Minnesota, I tell my clients about "pretty good practices."

"Pretty good practices" are ideas which work in some companies and make sense to consider for other companies. Pretty good practices are not set in stone, so they can be refined to fit the unique culture and situation of each organization. Pretty good practices are things I consider pretty good, but I'll keep my mind open in case a better idea comes along.

Most importantly, though, where prescribing a "best practice" tends to end the discussion (whether the practice is really the "best" or not), taking about "pretty good practices" is the beginning of a conversation about how to improve.

On average the ATM machines in the town of Timisoara, Romania are installed at exactly the right height for customers to use.

The problem is that one of the machines is six feet above the sidewalk and only usable if you bring a stepladder. The other ATM is at about the right level for a toddler.

But on average, they are perfect.

(From Austrian Times, via BoingBoing)

Consumerist today has a story of a car buyer who was told he had to give the dealer top marks on the customer survey or else the employee could be fired and the customer might not be allowed to bring his car back for service.

Clearly this dealer's behavior will not lead to increased customer satisfaction, nor is Ford getting a realistic measure of how well the dealer is performing.

This story may be unusually extreme, but this behavior seems to be endemic  in the auto industry. Anecdotally, I haven't talked to any recent car buyer who hasn't been subjected to some sort of blatant attempt to manipulate the customer satisfaction survey.

Over the years what started as an honest effort to measure and improve customer satisfaction seems to have morphed into a hollow exercise which penalizes employees who don't cheat. But the process is so baked in that change is extraordinarily difficult.

Change is necessary, though. The existing customer satisfaction surveys are not only ineffective, they actually encourage bad behavior. Here's how I would fix a badly broken process like this:

1. Immediately stop doing customer surveys. If the customer satisfaction process is this badly broken, it's not only a waste of money, it is actually making customers less satisfied and encouraging bad behavior. Ending the program will send a clear message that things must change, and force the organization to overcome its inertia.
2. Reconsider the ultimate goals of the customer feedback program. The goal is (probably) not to get good survey scores for the sake of good survey scores. The goal is to provide excellent customer experiences. In the case of auto manufacturers, the goal is to make every customer as satisfied as possible with the purchase and service experience, and strengthen the relationship the car owner has with the brand.
3. Develop a new feedback process from the ground up around the new goals. To develop that deep satisfaction and relationship with the customer, you don't just want to ask if they were satisfied and punish the dealer if they weren't. Instead, rebuild the customer feedback process as an opportunity to identify and correct mistakes the dealer might have made. Don't just ask how satisfied the customer was, ask what the customer needs to resolve the problem, and then have the dealer correct the issue. Give the dealer incentives to fix things, and make the goal "eventual customer satisfaction" not "zero problems ever."
4. Make it a two-way street. The people being evaluated should buy in to the process--which means listening to dealer's concerns, being fair and open about how they are measured, and also giving them a way to contest unfair feedback. No process is perfect, and there are customers who threaten bad feedback if they think they can blackmail the dealer. Have a review process for the dealer to demonstrate that they did everything possible to satisfy the customer.
5. Don't tolerate cheating. Even the best customer feedback process will be manipulated if someone thinks that's easier than providing good customer service. Customer feedback programs need to be actively managed, and anyone caught trying to cheat must be punished.

Conventional wisdom--and a moderate amount of academic research--holds that better customer service leads to increased customer loyalty.

The effect takes some time, however. If customers immediately abandoned a company after a single bad service experience, no company would dare skimp on service.

Instead, a less-than-satisfying customer experience gives customers a reason to explore taking their business elsewhere. If the alternatives look better, then they may leave. The effect is an increase in customer defections over time.

In some research we did for one of our clients we tracked customer defections for several months after a customer service call. We found that after six months the defection rate among customers who were "Very Satisfied" with that original call was one quarter the rate among customers who were "Somewhat Satsified" or less.

That's a striking difference, but it took months for the gap to fully open. That one less-than-fully-satisfying service experience probably was not the only thing which led the customer to leave. More likely it was part of a chain of disappointment. Had the company broken the chain they could have kept the customer.

Fortunately this client now has the data to quantify what disappointing customer service is costing. They can choose to make the investment to improve servie and retain those customers--and they know exactly what it's worth.

Issue 55 of Vocalabs' newsletter, Quality Times, has been published. E-mail subscribers should have a copy by now.

This issue contains a pair of essays on the theme of performing deep vs. wide surveys and how to construct a hybrid program which combines the best of both worlds.

As always I hope this is interesting and informative to my readers. You can get each issue e-mailed to you as soon as it is published by signing up on the newsletter page, or you can read it in your favorite newsreader with the RSS feed.

T-Mobile is in the news today for announcing that it plans to reduce its customer service operation by 1,900 people over the next several months.

This is a company which, several years ago, was near the top of our customer service survey rankings. Over the past year our National Customer Service Survey data shows T-Mobile getting considerably worse, while the company has faced the misfire of its AT&T merger, a declining subscriber base, and the continued lack of the iPhone.

Unfortunately, T-Mobile seems to be stuck. It needs to invest heavily across the board in order to be competitive with the other three national carriers. But the resources to do that just aren't there.

I've heard some speculation in the industry that perhaps another foreign carrier could buy T-Mobile as a way to enter the U.S. market. Since T-Mobile USA is already owned by Deutsche Telecom, this would not be a case of an American company being bought by a foreign company. Nevertheless, it doesn't take much to imagine the political problems with, say, China Telecom trying to acquire a major U.S. phone company.

So it's hard to see where T-Mobile goes from here. The current vicious cycle of declining customer base and reduced service levels will not end well.

Advanced analytics tools have made it easier than ever to identify patterns of customer behavior. It's now much easier to learn things like how often customers call on the phone after visiting your website, or whether certain customers consistently try and fail to use self-service options. This often uncovers some obvious places where you're leaking money through excessive support costs and repeated customer problems

Understanding why these patterns exist is harder. You can guess, but it's much better to ask the customer directly.

Customer surveys designed for this kind of in-depth insight are a little different than a routine customer service tracking survey:

• Sample: This is not the place to use a random sample. Instead, target just the people in the group you're trying to understand. So if you're trying to figure out why some customers bypass your IVR and try to go directly to an agent, those are the customers to survey.
• Method: It's important to get to the customer as quickly as possible after the experience, while the memory of what happened and why is still fresh (but not during the same phone call please!). Have a human being call the customer within an hour. The quick response and the rapport between the customer and interviewer yields the most detail and specifics about what happened and why.
• Questions: Even though this survey isn't intended to track your top-level metrics, you still want to include those satisfaction or loyalty questions in order to compare against your entire customer base. In addition, ask detailed "what happened" and "why" questions with an emphasis on open-ended questions.
• Size: For these purposes, several hundred completed interviews is a good starting size. 500 interviews will give you a broad cross-section of the customers you're interested in and some meaningful statistics about the different root causes and their prevalence (while keeping the budget manageable). Going up to a few thousand interviews gives more granularity and better statistics on particular root causes.

The challenge today is often trying to take too much data and generate insights and understanding. Modern analytics tools are a powerful start. Targeting a detailed customer survey to a specific pattern of behavior is often the quickest and most effective way to understand what's really going on. 

In customer service surveys it's fairly common to find customers who need more attention from the company. If a customer expresses dissatisfaction during an interview, or indicates that they have an unsolved problem, we strongly recommend giving the customer the option to have a supervisor call the customer back and try to resolve the issue.

Not all customers will take this offer, but when they do there needs to be a service recovery process to make sure the customer is taken care of:

1. Notification: As soon as the customer asks for the follow-up, someone at the company should be notified. There may be a dedicated service recovery group, or these may be assigned to a regular manager or supervisor, but there needs to be someone who has clear responsibility for investigating each case and reaching out to the customer. Most customers will expect to be contacted within a day or two,  and the responsible supervisor should get the notification within a few hours.
2. Follow-Up: Be flexible in how you are willing to resolve customers' problems, because the range of issues which turn up can include just about anything. The responsible supervisor should be empowered to go outside the company's normal policies and processes if necessary, because problems which land here may have been caused by those policies and processes. If the customer has a genuine issue, standing behind "this is our policy" is about the most infuriating thing the company can do.
3. Accountability: Many employees find they have more enjoyable things to do than trying to placate upset customers. There needs to be a process to track when customers have been contacted, what resolution was offered, and what the ultimate outcome was. Otherwise, service recovery may take a back seat to other routine tasks and customers can slip through the cracks. If you can't reach someone on the first try, there should be a clear expectation of how many times to try to contact a customer before giving up.
4. Approval: Every customer's case should be reviewed to make sure the follow-up was adequate. A random sample of customers should be contacted again to confirm that the resolution met their expectations. This serves as an important cross-check to make sure you really are delivering the service you think you are.
5. Tracking: The issues which appear in the service recovery process often represent the worst service failures in an organization, so this can be a goldmine of information about how to improve overall service levels. Tracking the root cause of customer complaints shows you ways to both make your customers happier and often save money at the same time.

An effective service recovery process benefits both the customer and the company. This isn't something that will happen by itself, however. With the right process and accountability, you can be assured that customer complaints are not being ignored.

I'll be in San Francisco next week for Mobile Voice 2012. If you're at the event be sure to say Hi, and maybe even stop in on my panel discussion on Wednesday afternoon. I'll be talking about designing a customer feedback process to improve self service applications.

Key to encouraging survey participation is making the customer feel that you are taking the survey seriously.

Among other things, that means asking only relevant questions and taking the time to make sure the survey communicates the right message to participants.

And don't do this: (click the image for a full-size, easier to read version)

While I'm sure that some companies really want the "None" service, after some deep introspection I decided that "None" was Unimportant to my organization.

Coincidentally, "Unimportant" also seems to describe how seriously this company took its customer survey.

We just published Issue 54 of our newsletter, Quality Times. In this issue we discuss things companies can do to improve the effectiveness of a customer service survey, and also why "Call Containment" needs to be banished from the customer service dictionary.

If you find the newsletter useful, please feel free to subscribe to receive it by e-mail.

Since the beginning of the IVR industry (that would be something like 30 years ago) the industry has been talking about "Call Containment." This is typically defined as the percentage of calls which end in the IVR and don't go to a person.

It's time to put a stake in the ground and declare the word "Containment" as forbidden in the customer service dictionary. Instead, the right phrase to use when talking about IVR performance is "Self Service."

"Containment" is a terrible, awful word which confirms all the negtive stereotypes of automated customer service. It leaves the impression that the design goal of an IVR is to put customers in a place they cannot escape from. And let's face it, we all know systems where that was the goal, stated or otherwise.

When you think in those terms, you also design and measure in those terms. A "Containment" metric only looks at whether a call stayed inside the IVR. It says nothing about whether the customer was served, or had a positive impression of the experience. About the only virtue of Containment is that it's easy to measure: calls contained divided by total calls.

"Self Service," on the other hand, contains the notion of service. If the customer was not served, then the IVR failed at self-service even if it succeeded at containment. Self-service is much closer to what both customers and companies actually want an IVR to do.

When you design and measure an IVR based on Self Service, you think in terms of empowering the customers to help themselves, rather than limiting their freedom to exit the system. Self Service is what we do at an ATM or when buying stuff online, and in many cases customers prefer it.

If you're in the customer service world, next time you're tempted to talk about "Containment" stop yourself and substitute "Self Service." And when you hear colleagues talking about Containment, you can try the phrase I use:

"Actually, I prefer the term Self Service, because it better reflects what we're trying to accomplish."

Effective customer feedback programs need to consider the depth and breadth of the data they collect:

• Depth is the amount of data the survey collects on each individual participant. Depth is important for understanding what's driving the customer's opinions and making the feedback actionable.
• Breadth is the number of customers who participate in the feedback program. Breadth is important for generating statistically meaningful results and calculating accurate metrics.

Deeper surveys generally cost more than shallower ones, because getting the additional feedback from each participant usually requires a live interview instead of an automated survey. A program's budget will determine how much depth and breadth you can afford.

The most effective feedback programs provide a balance of both depth and breadth. At the extremes, focusing entirely on either depth or breadth often represents wasted effort:

• Narrow and deep feedback gives a very detailed look at the opinions of each participant, but the budget normally won't allow a statistically meaningful sample. For example, a focus group or usability test where you physically bring participants to a study center and interview them at great length will give fantastic insights and generate many ideas for improvement. But this can cost hundreds to thousands of dollars per participant, making it financially impossible to include the hundreds of participants necessary to gather meaningful statistics. So it's impossible to know with certainty whether the insights you get are problems common to all customers, or represent the quirks of the individuals who happened to participate.
• Broad but shallow feedback is the opposite. Automated surveys are nearly free so it's possible to collect feedback from thousands or even millions of customers. However, many consumers today have little patience for these surveys, so its not realistic to expect to get answers to more than a handful of questions (and response rates tend to be poor). Getting thousands of people to answer the same questions is fantastic for precisely calculating the statistical results for those specific questions, but everything else is pretty much speculation.

There's a common fallacy that you don't need depth if you have enough breadth, and vice-versa. This is mistaken: getting more detailed information from one person tells you nothing about the opinions of people you didn't talk to; and asking the same questions of more people doesn't give any information about the questions you didn't ask.

Balancing depth and breadth takes more thought than just doing a bunch of surveys, and it may require collecting feedback in more than one way. But it's the only way to make sure you're getting reliable and actionable results from a customer feedback program.

Apple is not the biggest company in the mobile phone industry, as measured by metrics like employees, customers, and revenue. But it is far and away the most profitable--to the extent that the company actually makes up the bulk of the profits of the entire mobile phone industry. According to today's Wall Street Journal, Apple is so insanely profitable that it is distorting the stock market as a whole.

How does one company manage to suck the profits out of an entire industry?

I think the answer is obvious: every other major company in the mobile phone value chain is focused on winning business. Apple is focused on winning profitable business.

So where other handset manufacturers and the carriers are beating each up on price and trying to outdo each other on feature-by-feature comparisons, Apple is delivering an entire customer experience. Apple's experience isn't just about the hardware: it covers everything about the phone including hardware, software, support, upgradeability, battery life, and the list goes on. Even the experience of removing a new iPhone from the box is meticulously designed.

Not every customer wants what Apple is selling, but a lot of them really want it and are willing to pay a premium (or put up with a carrier they don't like) to get it. As a result, the most profitable business flows to Apple and the carriers feel obligated to make crazy deals to get the iPhone or lose customers.

I don't think there's anything magic about Apple's formula. But it requires a company to focus on what it really does well, think about the whole customer experience, and not get distracted trying to respond to every ripple in the marketplace. That's very had to do in our business culture where the goal is usually profitability measured in three-month increments and the mantra is "win every battle at any cost."

Doing a deep dive on the National Customer Service Survey on Banks we recently discovered a startling fact: in our survey, of customers who reported a problem with their bank's IVR, 20% of the time the problem was somehow related to authentication. For example:

• Nonsensical Authentication: The customer was required to enter an account number to find the hours of a branch.
• Impossible Authentication: The customer was calling to report a lost or stolen card, and did not have his account number readily available.
• Technical Failures: The customer gave a valid account number and PIN, but the system didn't recognize it--either because of a speech recognition problem or because of some other failure
• Prospective Customers: A prospective customer wants information, but is asked to enter an account number first.

In all, these problems affected about 5% of all the people calling one of the four large banks in our survey. 5% may not sound like a lot, but it represents tens of thousands of customers every day who have to call back, don't get what they want, and have a negative view of their bank as a result. It also represents millions of dollars in expenses to these four companies due to repeat customer calls and unneeded time with a live agent.

So why does this happen? I believe it's because these IVR systems were designed from the company's perspective, not from the customer's perspective. The company decided that it would be helpful for them to know the customer's account number, and so they ask for it even when it makes the customer's experience worse.

The solution is to ask how the company can best serve the customer, not how the customer can be helpful to the company.

Automated customer service surveys (via e-mail, web, or IVR) have one huge virtue: they are cheap. Big companies can afford to collect feedback from millions of customers, creating a broad statistical profile which can be used to measure customer satsifaction and other key metrics with a lot of granularity. However, automated surveys often suffer from poor response rates, and can usually only have a handful of questions before customers start abandoning the survey. That severely limits the quality and depth of the feedback.

Phone interviews, on the other hand, are almost exactly the opposite: customers are much more willing to participate when there's a live person on the other end, and they are willing to answer a lot more questions as long as the interview is well-structured. Interviewers can also probe and go in depth in a way a computer can't. But it's inherently more expensive to conduct a live interview, and there's just no way to escape that fact.

In some companies I advocate for a hybrid approach, combining the vast ocean of customer feedback afforable with an e-mail or IVR survey, with the depth and reliability only attainable through inverviews.

This takes advantage of the fact that in most companies, most customers have a reasonable experience most of the time. If a customer was satisfied with the experience, there's usally little to be learned about how to improve things. On the other hand, if the customer was not satisfied you can get a lot of useful information. Running both an automated survey and an interview process in parallel focuses the survey dollars and effort where it is most effective.

The automated survey is used to establish high-level statistics across the organization and identify underperforming segments. The live interviews are used to generate actionable feedback about how to improve. Customers are selected for one survey or the other based on whether the statistics suggest a high probability of dissatisfaction.

The end result combines the best of both worlds: tons of high-level statistics across the entire organization, and highly detailed in-depth feedback where the company needs it most, all delivered in a coherent package to the people who need to make both strategic and tactical decisions about how to deliver the customer experience.

In case you missed it, we issued two news releases this week: NCSS Mobile Phone results, and NCSS Tech support results.

AT&T also issued their own news release after ours came out. We didn't have anything to do with that, but we are glad to see them paying attention.

We've published Issue 53 of Quality Times, Vocalabs' newsletter.

In this issue we discuss the 2011 results for the National Customer Service Survey in mobile phones and computer tech support. These reports compare customer service quality at AT&T, Sprint, T-Mobile, Verizon, Apple, Dell, and HP.