Everyone pays lip service to the idea that they love their customers — some companies even make corporate slogans out of it. It's easy (and very tempting) to be cynical about it, but the true fact is that you ignore customer satisfaction at your own peril.

In the call center, generalized customer dissatisfaction will show up in two quick ways: declining revenue (for sales centers) and poor quality assurance scores for all but the top tier of agents.

You'll also start to see a disconnect between the numbers you use to measure call activity and those that measure transaction success. Call activity might look normal, or even better than average, if what you're measuring is things like duration, handle time, number of calls per hour. There may be an inverse relationship between how long a call lasts and how successful it is.

You may be getting more shorter calls because of a structural customer problem but not recognize it because if you're measuring success based on activity, it looks like your agents are doing a good job of handling what customers throw at them.

You are using two distinct criteria to measure your call center — call handling metrics that tell you how long and how many; and outcome-based metrics that tell you how well the call went. The growing disconnect between these two types of metrics will be the first clue that something has gone wrong in customer satisfaction.

While you're poring over anomalous activity/satisfaction data to determine what's going wrong, your customers are out in the wild, spreading word of their dissatisfaction to everyone but you.

Customer dissatisfaction may, or may not be, the fault of call center operations and policies. But regardless of who is to blame, you will be blamed.

Another effect of dissatisfaction that will have rippling ramifications outside the center is the decline in cross-sell and up-sell opportunities. You can't finesse a cross-sell to someone who still has a problem with you. Again, that's going to lead to declining revenues, the first indicator I mentioned a moment ago.

But worse than any of this, perhaps, is the series of silent defections that you'll experience. Customers will quietly pack up and leave you to go somewhere else. There are so many downsides to this scenario that I don't know where to begin: yes, you lose immediate revenue. You lose the entire lifetime value of each customer, and you incur costs in having to replace that customer.

More important, though, you lose the opportunity to learn what went wrong, and to fix it. Those who vote silently with their feet deprive you of the feedback you need to make process improvements. Here's some advice for even the best centers: put in place some mechanism to capture the information about failures. Lots of centers will survey customers after they leave to see what went awry. Most customers won't respond to those surveys. Their interest in helping you out is zero.

You could record all your calls and try to identify patterns and trends in failed customer relationships. Or you could train agents to see the signs of failure and then take some prescriptive action, even if it's just to acknowledge that something's wrong and needs to be fixed.

You will always make some people unhappy, and those people will leave — minimizing those defections is a worthy goal. But since you can't reduce defections to zero, the only thing you can do is to ensure that each time a dissatisfied customer does leave, you know what it was that sent him to the exits. Let a few leave so that you can save the rest.

One thing that I fear happens a lot in call centers is a knee-jerk response to a long series of bad data: when managers detect several months or quarters of anomalous activity that indicates customer unhappiness, they throw resources at it to try to get the stats back into alignment. They don't necessarily try to solve the underlying problem of customer dissatisfaction — they try to rein in the aberrant activity statistic to bring it back to a benchmarked "normal." This has the same effect as taking an aspirin for a brain tumor — it might provide a little short-term symptom relief, but it's not going to do much good in the end.