Follow Us

Close
Like what you're reading?

Subscribe to receive periodic updates about new posts by email, or follow us via Twitter or RSS.

Please enter a valid e-mail address to subscribe.

Close
Close

You have subscribed.

How US Airways Delivers Proactive Customer Service Using Their IVR

 

Recently, the Customer Service Investigator posted an article with tips on proactive customer service that became one of our most-read articles to date. Since the post resonated so strongly with readers, we decided to revisit topic by drilling into one such strategy for proactive customer service: programming your IVR to respond to customers intelligently based upon prior data analysis.

To provide our readers with details on how this is done, I reached out to one company that has done this very effectively–US Airways.

They implemented a new IVR system about two years ago that proactively suggests solutions to callers based on data about their purchase history and the reason they are most likely calling.

For example, if I bought a plane ticket from US Airways a month ago and called their customer service line two hours before my flight, their system would recognize my phone number and most likely say, “Hello Ashley, would you like to check the status of your flight to Phoenix this afternoon?” This would save me from having to dial through several prompts to find the answer I’m looking for, or wait on hold to speak with a live agent. US Airways benefits from this system in several ways that I’ll discuss later in this article.

So, how does this process work? Here are four steps US Airways took to implement their proactive IVR system.

Step 1: Create a List of Top Reasons Customers Call You

To start, you will need to compile a list of your most common customer service scenarios. This will eventually help you create a list of solutions to proactively suggest to people who call you. For US Airways, the most common reasons customers call are to check on the status of a flight, book a flight, change a reservation, or request information on general policies (e.g. baggage fees and allowances).

To find out what customers commonly need help with, the US Airways team went through recordings of calls between live agents and customers. They also did something called “Wizard of Oz” testing, where they set up an automated system to respond to customers who called in, and listened behind the scenes to how these customers responded to prompts. This process allowed the US Airways team to create buckets for the common reasons customers call, as well as identify the prevalence of each issue or question type.

Nuance Communications, the company that provides US Airways’ IVR technology, said this process can take days or weeks, depending on how much data you already have, and whether you’ve conducted any prior analysis. In US Airways’ case, they had amassed countless hours of recordings, but had done very little analysis on any of it.

“Most [companies] do not have the data readily available. They have tidbits here and there, but it’s spread out in five different systems and they don’t know how to coalesce the data,” says Dena Skrbina, senior director of hosted solutions marketing for Nuance.

Skrbina says some companies assume they can find the most common reasons customers call just by looking at which options in their IVR system callers choose the most. This can be misleading, she says, because this doesn’t account for callers that “don’t play in the IVR,” or callers that simply press “0” to talk to an operator instead of choosing an option from the automated prompt.

There are cases, however, where companies might not have to analyze call recordings to get data on why customers call them most. For some industries, such as banking, market researchers have already compiled this information. Or, if you use customer service software, the system might allow agents to tag common customer issues by type, then pull a report that shows which tags occur most often.

Step 2: Plot Your ‘Top Reasons’ Along the Customer Lifecycle

Next, you will need to determine where these most commonly asked questions typically occur in the customer lifecycle. This will eventually help you determine when you should proactively suggest a specific solution to customers. US Airways’ analysis of customer calls, for example, found that flyers who call anywhere from a month to two weeks before their flight typically want to change their reservation (as opposed to checking the status of their flight, or booking a new trip).

To make these determinations, you will need to have a system for tracking customer purchase and interaction history. This could be in a CRM system, spreadsheets, or another type of documentation. Regardless of what you use, you need to record the date (and ideally time) when each purchase and interaction occurred.

Next, you’ll want to dig back into your call recordings and pull a sample of customer interactions that fall into the buckets you’ve created for identifying top reasons customers call you. For each call in this sample, pull up the customer’s purchase and interaction history. Then, record the time between the call you’re listening to and the interaction your company had with the customer immediately prior (whether via phone, Internet, etc.). Here’s a hypothetical example: you note that the customer on the call you’re listening to wants to check on baggage allowances. Looking back at the previous interaction your company has had with this customer, you discover they booked a trip overseas eight months ago.

Once you do this for every call in your sample, calculate the average time between interactions. Through this process, you should be able to pick up trends around the sequence of events that leads customers to call you. Here’s an example of what US Airways’ customer purchase lifecycle might look like:

US_Airways_Lifecycle

You might have several customer lifecycle maps, depending on whether your “top-reasons for calling” vary by customer type. For example, US Airways’ frequent flyers’ lifecycle might include other types of calls, such as checking their mileage and benefits.

Step 3: Program Your IVR to Deliver the Right Solution Based on Customer History

In this step, you will actually program rules into your IVR that tell the system what solution to proactively suggest, and when to suggest it. Completing this step will depend on the capabilities of your IVR and whether it can integrate with your CRM, or whichever system you use to manage your customer information.

If you are unsure whether your IVR has the necessary features, you can share the following description with your IT staff and/or technology providers that explains what you’d like the system to do:

  • When someone calls, we’d like the IVR to be able to recognize whether that phone number exists in our customer database.
  • When someone calls, the system should identify when the last interaction was with the customer (if there was one), what type of interaction this was, and the time elapsed between this last interaction and the current one.
  • If the system recognizes that the caller’s last interaction was one that usually leads to a “most-common reason customers call,” the IVR should suggest the solution that solves the reason they’re calling. This will require having the ability to create a library of solutions that the IVR can choose from and writing the appropriate “if / then” rules. (“If / then” rules = If someone calls with “x” information in their customer profile, then suggest “x” solution from the library).
  • If the caller calls back on the same day, the system should recognize this and not offer the same solution. Some systems can also be customized to say “welcome back.”
  • If the caller is already a customer, we’d like the IVR to address them by name, or “nickname,” if the customer profile includes that information. Also, if they have a language preference in their profile, the IVR should automatically speak to them in that language.
  • If the caller is a customer, but their phone number wasn’t saved in their contact profile, we’d like the system to ask the customer if they’d like us to remember their phone number for next time. If they reply with “yes,” the system should save this phone number in the customer profile.

Step 4: Measure Success Rates to Continually Fine-Tune IVR Rules

This final step should be something your team does on a consistent basis–measure the success of your IVR’s proactive support system, and mine for opportunities to improve.

“Too often, we see companies that launch their new system, pat themselves on the back and move on to the next thing. You should always be looking for ways to improve the experience and intuitiveness of the system,” Skrbina says.

To measure the effectiveness of their IVR system, US Airways does a thorough success analysis approximately twice a year. Specifically, they look at the following metrics:

Transfer Rate
This refers to the rate at which callers must be transferred to a live agent because the system doesn’t understand what the customer asked. This happens when the solution that the system proactively offers isn’t correct, and the caller has to request something else. This could mean that the IVR’s speech recognition vocabulary is missing terms needed for understanding the caller’s intent.
Repeat Information RateThis refers to instances where the caller requested that the system repeat options, including solutions that are proactively suggested. If you see the rate increase, this could mean that the solutions that are proactively suggested need to be worded more clearly. It could also mean the solution they are were looking for wasn’t there.
Repeat Caller AveragesThis refers to instances where the caller hangs up, then immediately calls back again. This often happens when the caller chooses an option, but then wants to go back to a previous menu. This could be an indicator that there is a mismatch between the customer’s needs and what the IVR is proactively suggesting.
IVR Exit RatesThis refers to instances where callers dial “0” to exit the IVR instead of using your automation options. If you see this rate increase, it could also mean there is a mismatch between the customer’s needs and what the IVR is proactively suggesting.
Average Time in IVRThis refers to the overall time callers spend in the IVR before finding the solution. Ideally, you’d want callers to get their answer from the first solution that is offered (instead of having to key through several options). If the average increases, or is longer than the time it should take to get an answer from the first solution, this could be another indicator that there is a mismatch between the customer’s needs and what the IVR is proactively suggesting.
Task Completion RatesThis refers to the rate at which callers solve their issues through the automated system. For example, if you were a bank and a customer wanted to transfer funds, were they able to do so just using the automated system? For scenarios that don’t require an actual action (e.g. checking the status of a flight), you can make this determination by asking, “Can we help you with anything else?” Then you can measure instances where the customer responded with “no.”

If US Airways’ team sees any of these metrics changing in a negative way, they manually dig into calls where error occurred (the caller didn’t complete a task, exited the IVR, asked the IVR to repeat information, etc.).

For example, in calls where the customer asked for instructions to be repeated, the team would want to know if it was simply a matter of the customer not finding the answer in the list of options that was presented. In this case, US Airways knows they need to adjust what solutions are suggested to customers like the one who called in that instance. Similarly, the team can dig into data for IVR exit rates to find out what caused the customer to bail out of the system.

Analyzing repeat caller data can be extremely useful for discovering new opportunities to be proactive. For example, if a bank customer called to change their address and was able to do so using the IVR, their experience would be counted towards your “task completion rate.” However, that same customer might call back the same day to order checks, so they would then be counted as a repeat caller.

If the bank in this hypothetical scenario dug into their data and found that customers who called back after using the IVR to change their address typically wanted to order checks, they might adjust the system to offer this option to all customers who call requesting a change of address immediately after the change is completed.

“You have to remember that even the caller might not know the full scope of what they need. It’s up to you to spot patterns–this call always leads to this call–and adjust your system accordingly,” says Dan Enthoven, the chief marketing officer for Enkata, which provides performance optimization software for call centers and other organizations.

What’s the Return on Investment?

I mentioned previously that implementing this strategy will depend largely on the abilities of your IVR. Admittedly, this technology can require a significant investment to launch and customize (although, some systems come with templates for proactive solutions, depending on the industry).

But for systems that are effectively implemented, you can expect a return on your investment in a few ways.

The first is an improvement in your “task completion rate,” also called call containment. This metric measures the number of customers you’re able to help without connecting them with an actual person, which reduces the number of tickets that are created in your call center. US Airways has experienced an increase in task completion of about 8 percent since implementing their new IVR. This has also increased their first-call resolution rate, or the number of cases where the customer’s issue was resolved in one interaction.

“We really want to be respectful of our customers time. These changes have really reduced the time that customers have to spend on the line with us,” says Tim Lindemann, vice president of reservations and customer planning for US Airways.

Another way the system has reduced costs for US Airways is by reducing Average Handle Time (AHT), or the amount of time it takes agents to serve customers when they do have to talk to a live person. Because the system has already recognized the customer via their phone number and collected other information using automated prompts, the live agent doesn’t have to spend as much time figuring out why the customer is calling. US Airways has seen their AHT decrease by about 5 percent since launching their new system.

Finally, these systems can increase your call routing accuracy. Again, ideally the customer will solve their problem through the automated system; but in cases where they don’t, the system gathers more information that can be used to send them to the right person who can help them.

Being Proactive Doesn’t Stop Here

When I finished researching this article, I called public speaker and consultant Adrian Swinscoe to get his thoughts on the topic. He regularly provides advice to companies wanting to become more proactive, and recently wrote an article for Forbes on the topic. While he said the IVR system I described is indeed proactive, it still requires the customer to make the first move in calling you.

“To be truly proactive, you really need to solve the customer’s problem before they have to call you, or even before they even know they have a question,” Swinscoe says. “I think too often companies think buying technology is the solution to becoming more proactive. Really it starts with a change in mindset.”

So while this type of IVR does provide a means for serving customers more proactively, your efforts shouldn’t stop here.

Thumbnail image created by John Spade.

Share this post:  
Ashley Verrill

About the Author

Ashley Verrill has spent the last six years reporting and writing business news and strategy features. Her work has been featured or cited in Inc., Forbes, Business Insider, TechCrunch, GigaOM, CIO.com, Yahoo News, the Upstart Business Journal, the Austin Business Journal and the North Bay Business Journal, among others.

Connect with Ashley Verrill via: 
Email  | Google+