Automation technologies like natural language interaction and IVR have revolutionized the way customers interact with the customer service function.

As well as cutting costs, these technologies can also boost customer satisfaction. That may seem counter-intuitive, but a growing number of customers prefer to “help themselves” by using natural language interaction for routine transactions rather than wait for a human agent to attend them.

But over-zealous or badly implemented self-service initiatives can also backfire badly, according to a recent survey by research firm Vocalabs.

In identifying common failings with automation strategies, the firm also reveals areas where natural language interaction could be used to dramatically improve the self service processes and so boost satisfaction.

One of the most common mistakes with automation is to make it difficult for callers to reach a person — the menu of IVR options is missing the all-important “speak to an operator” for example.

Intuitively, you would expect that this tactic would encourage more customers to use self-service – out of desperation if nothing else. But Vocalabs says “hide the cheese” is almost completely ineffective as a customer service strategy.

Among the customers surveyed who reported that an automated system made it hard to reach a person or find the right option, only 2% successfully used self-service. The majority did eventually reach a person, and the rest hung up without getting what they needed. Vocalabs expects that most of the latter group called back to try again, driving up customer service costs.

Put another way, for every customer who got what they wanted from the automated system, twenty customers hung up in frustration, and thirty more customers figured out how to get around the barriers.

Another common mistake — and one which personally I find infuriating — is that when a dissatisfied customer calls back, he or she is required to explain the problem all over again. These repeat calls are often more challenging than customers’ initial calls, which is why the contact center industry attaches so much importance to first-call resolution.

Indeed, repeat calls from the same caller are often treated as separate incidents, even though for the customer it’s all one long continuation of a single incident.

Customers who have to call multiple times to get what they need make up a significant fraction of call volumes and so business are driving up operating expenses needlessly by making repeat callers start from the beginning each time they call.

I can see a potential application for natural language technology here, spotting tell-tale words of phrases in the customer’s dialog — “I’m calling again” to use an obvious example — that show they have called previously.

Interestingly, the study shows that customers who have to call multiple times to get what they need are just as satisfied as first time callers if they can pick up where they left off from the earlier calls.

But those who have to repeat their problem all over again when they call back are likely to be extremely dissatisfied.

The third key finding reinforces the importance of giving customers a choice about automated channels.

When a customer uses self-service by choice, they rated the experience better than among customers who spoke to a person, Vocalabs’ survey reveals.

But when customers wanted to speak to a human agent and got routed to an automated channel instead, the experience was “dramatically worse”.

According to Vocalabs, customers are much better than automated systems at deciding when they need to talk to a person and when they can use self-service. However, most speech and IVR systems decide whether to route the call to a person based solely on what the customers say they need — not the channel they would like to use.

Clearly, this is an area where natural language interaction could easily be applied to unobtrusively detect which calls would benefit from being routed to human agents.

For example, an IVR system asks the caller “why are you calling?” If a person replies “it’s about my bill” then they are probably happy to continue using the IVR system. But if the person replies “I’m calling about my bill but please let me speak to a human being”, then it makes sense to respect their wishes.