How to Slash Your Customer Escalations with AI
Traditional escalation practices are a resource drain on your support organization. This session will reveal how AI can revolutionize your escalation management, cutting response times while ensuring only the most critical issues reach higher levels. We’ll challenge the notion that escalations are merely reactive and show you how to anticipate and manage potential crises before they arise.
You Might Also Like
All right. So just a brief, brief intro is right. As Judy said, my name is Max.
I'm a
senior customer success manager with SupportLogic. I've been here at SupportLog
ic for about
three years now, working with many of our, many of our customers in their
journey to
reduce escalations and also to just become more proactive with AI. Right? And I
'm joined
today by one of our esteemed customers, Ken, from 8x8. And Ken, would you mind
introducing
yourself for the audience here? Sure. Hi, I'm Ken Thon, the director of digital
support
at 8x8. I've been a customer of SupportLogic now for upwards of, I believe, in
almost four
years. And it's been a fantastic journey thus far. Awesome. So I'm going to
assume that
this is the clicker. It is excellent. Right? Okay. So what we're going to cover
today,
challenge and impact of customer escalations, a bit of proactive escalation
management
and how SupportLogic predicts escalations. And Ken's going to tell a bit of the
story
of how 8x8 has been successful. And as well as we'll talk a little bit around
the practical
way of how do you do this, right? How do you implement a proactive workflow for
reducing
escalations and being preventative in that capacity? So here's up on the screen
here.
You'll see here are some of the kind of typical challenges that we see pers
isting in support.
This should be very familiar to some of the many of the folks here. Things like
constant
firefighting, not understanding the true voice of the customer and cost center
versus revenue
center. Ken, what do these kind of mean to you? How have you seen these
impacting your
organization? I mean, these are major pain points for us. And that's what drove
a lot
of our escalation rates as high as they were. You know, inability to really
kind of identify
what the customer issue was. Inability to really manage our own time sometimes.
It was
just one of the biggest pain points for our customers was just timely updates.
Getting
back to them in a timely manner to really recognize that they had issues they
needed
to resolve quickly. And we weren't able to kind of take action on them as fast
as quickly
as they wanted them to us too. Yeah. How about how do you typically, like how
do you
typically, you know, you see here we talk about focusing on the wrong metrics
for making,
for dealing with escalations, right? What would you typically measure beyond
just escalation
rate to try and improve a proactive experience? A few things we really looked
at were, you
know, total time resolution milestones. And one of the things that we noticed
when we
began using the support logic tool was that we weren't really quite clear on
what our
milestones were. Neither were our customers. And we may have been measuring our
milestones
incorrectly, setting the wrong expectations based upon what customers wanted to
see.
Awesome. Okay. So, yeah, I mean, these are typical challenges, right? Near and
dear to
all of us. One that I actually don't see here that I want to call, that I'd
like to call
out, right? And now you've been asking a question about Ken. Sure. Is the
impact of escalations
on morale, right? On the employee experience, right? From what I've seen, like
support engineers,
support managers, support teams in general, they maintain a kind of a pride of
ownership
in resolving a case, right? They want to believe that they can resolve a case.
Engineers
in particular. And so they often don't ask for help quickly enough, right, to
get ahead
of things. And that can lead to escalations, which can have an impact on morale
, right?
There's nothing worse than being an engineer and being the one who is owning a
case that
all of a sudden is escalated and you're having leaders coming to you about it,
right? So,
how have you seen the impact of escalations in general on morale in the
organization among
the support teams? Absolutely. It's actually kind of coincidentally, you asked
that question
because I was thinking about it earlier today when Krishna was talking about
kind of the
hard cost and the soft cost of escalation cases. And beyond the fact that you
have a
cost per case and that if that case is escalated, you have exponential cost on
top of that because
of other people you have to have involved, SMEs, managers, specialists, whoever
. There
is that inherent cost to the person as well, that mental health deficit also,
that each
time a case becomes escalated, whoever is handling it gets chipped away a
little bit.
And let's be frank, we love our customers. But customers are terrible. I'm a
customer
myself, I recognize how terrible I can be, especially when I'm at that
escalated state
and I can take a good chunk out of a person when I'm on the phone with them.
And that
happens to our agents day in and day out when you have escalations. So, the
fact that we've
been able to reduce our escalation rate to the point we're at means that the
teams that
have been handling these escalations are feeling much better. They're happier,
they're more
content. And because of that, you see an improvement in not just their
performance, but the CSAC
for our customers as well. So, it's been definitely a huge win for us.
>> All right. So, what are some of the long-term customer impacts of, what are
some of the
long-term impacts on customers of firefighting? Right? 96% of customers say
support experience
is important in their choice of loyalty to a company. 82% of support leaders
are actively
looking to shift to more proactive support models that directly drive revenue
growth.
Ken, what do these figures mean to you? >> They mean everything. Obviously, we
put in
time, we put in effort to ensure that our customers are happy, their issues
resolved.
And every time we're able to save a case from becoming escalated, that's just
that better
word of mouth that can reach our customers and potential customers from that.
So, it
just really helps to reinforce the brand value that we've been trying to drive
day in and
day out. >> Terrific. Yeah. Okay. So, let's talk a little
bit about the impact of escalations on revenue. Right? So, why do escalated
cases carry a higher
cost for the business? Now, there's a number of different aspects of this,
right? One of
the biggest problems with an escalation is when a member of C-suite or leaders
get involved,
all of a sudden the cost of the case itself skyrockets, right? The hourly rate
of your
support engineer who's working the case may be manageable, right? But the
hourly rate
of your CEO or your CCO is a very different thing, right? Plus the human cost
on morale
as well of getting a senior executive involved directly with the team members
trying to troubleshoot
the issue, right? >> Yeah.
>> Solution development on the engineering side when you have to bring in
engineering
resources, right? As well as just the duration of the case and obviously the
risk to the
bottom line of customer churn, right? What do these kind of call out? What do
these mean
to you and how do you see them impacting the business?
>> It's funny because I see C-suite in the box right away and that hits home
pretty hard
because with, you know, platforms like LinkedIn, it's become very easy to reach
a CEO or a
CCO and it's happened to us multiple times where we've had to take action where
very
simple cases suddenly have, you know, C-suite level attention which is bad for
the customer
but it's really bad for us too since we're supposed to be handling this all in-
house.
On top of that, rapidly trying to come up with solutions that we could have
really spent
time and invested time and effort and resources into building, you know, from
the start as
opposed to rapidly trying to do it right away impacts our ability to scale and
control
our costs. And obviously each time this happens repeatedly with customers, the
greater the
chance of churn and, you know, for us having, you know, trying to protect our
revenue, protect
our subscriptions and our customers, it is paramount to success.
>> Great answer. Thank you. Okay. So proactive escalation management, right? We
know that
escalation in general, escalation management in general has typically been a
very reactive
process, right? And part of the promise of AI has been to make it a more
proactive process,
the ability to get ahead of these things, right? And so just kind of, when we
think in
terms of this, I imagine 8x8 initially, right, was very reactive in their
strategy around
dealing with escalations even if you wanted to be proactive. So what does
proactive escalation
management actually mean from your perspective?
>> For us, it meant identifying potential cases that were on the potential of
becoming escalated,
identifying, you know, some root causes of why these cases were on the cusp of
becoming
escalated and trying to address them beforehand. The idea of being proactive
and seeing these
issues that customers had not kind of flame up into something much bigger than
they necessarily
needed be. Really kind of adapting our service to ensure that we weren't making
the situation
worse than it needed to be also.
>> Right, yeah. What are some of the outcomes that we would hope to achieve
from being proactive
on escalations? Obviously the reduction of escalation, right? We talked about
the improvement
to employee experience as well, right? The improvement to see that. Any other
things that
kind of jump out at you?
>> One of the things that happened was, well, outcome wise, we were seeing
escalation rates
anywhere from 10 and we hit a high about 13%. And as we began to use the
support logic
tool, we saw those rates drastically fall. In fact, we're at a point now where
we're
probably sub 2% on any given day, which is fantastic. But what the tool allowed
us to
realize is that we had a lot of procedural engagement issues. One of the key
pieces was
we were talking about how we were speaking to our customers. Our porting team
began to
use support logic. And our porting team typically had the high escalation rate,
but they also
had very low morale. And one of the reasons we've gotten to notice was porting
numbers
is a very procedural task. It's very mundane. It's basically you fill out a
form, you submit
it, it's been correctly submitted, and you wait for the carriers to take action
and actually
do the number porting itself. But during that period of time, the customers
left kind of
waiting in limbo, wondering if things are working, if things are progressing
right.
And we did what we do. We had a very procedural means of doing this. We had tem
plat, templatized
email messages. And that wasn't enough for our customers. And we didn't
recognize this.
From our quality assurance perspective, we did everything we needed to do. We
checked
all the boxes. That could be 100% of QA. Doesn't necessarily equate to 100% CS
AT. So it really
let us realize that we weren't engaging our customers appropriately. Customers
wanted
to have that warm fuzzy feeling. They wanted to recognize that their issues
were being
taken care of, and that even if it was taking a long time, even though it was
technically
out of our control, that we were kind of with them. And we understood, we were
monitoring
them, letting them know that things were progressing in the manner they were
supposed
to. And that we recognized that it required some patience. And there's a way
for us to
make that engagement nice, painless, as opposed to what we were doing, which
was letting them
suffer in silence a little bit. And kind of seeing that happen again and again,
being
shown this through the predictive model made us a lot of us to make that change
. And in
that sense helped us drop our escalation rates, but also made the team far more
proactive,
far more engaging. And they're a much happier team as a result of that.
Awesome. Yeah, you talk about suffering and silence there, which I think is
always, it's
an interesting thing, right? We don't want to provide empty updates to
customers necessarily,
but we do want to make them feel like we're constantly right alongside them in
this process
of getting their issues resolved, right? Yeah, customers don't want that single
here, follow
the knowledge-based article a solution anymore. And with the advent of
technology, we shouldn't
be providing that as a solution. Yeah, great. Okay. So I think some of you have
seen this
slide, you've probably seen this slide already this morning. This is kind of
the benchmark,
this is a benchmark in bedrock of what support logic does, right? So we analyze
all of the
unstructured text in your conversations, interactions with your customers,
right? We heard in some
of the panels earlier that 90% of customer communication is unstructured data,
right?
Metadata makes up a small, an important portion of it, but a relatively small
portion of it,
right? So the way support logic looks at this, right, is we extract and categor
ize all of
the, a variety of different sentiments and signals across your cases, right?
Those range
from positive sentiment to negative sentiment, things like frustration and
confusion, as
well as just customers expressing urgency on getting their issues resolved or
that they're
facing a critical issue. And what support logic does with this information,
right? It
takes it and rolls it up into a series of scores, a sentiment score, an
attention score,
a support health score. We've talked about a little bit, we've talked about
these things
and you've heard about them a little bit already, but one of the key things
there is with the
scores that we create, right? You're maintaining context over the life of the
case, right?
We know that this case is trending in a positive or a negative direction, right
? Escalation
prediction, it's not a binary thing, right? It's a series of actions and
activities and
feelings that happen over the life of a case that take it in a direction that
can eventually
lead to a case being at risk of escalation and actually escalating. The beauty
of this
is that many of those factors are measurable and they're measurable in advance
and that
allows for the ability to predict these things in advance and at the very least
point out
cases that are at risk, right? And if you can be notified of cases that are at
risk,
you can start to be proactive. You can build proactive workflows around
addressing these
issues before they come to a head. So in kind of a nutshell, this is really
what support
logic does in how we predict escalations, right? It's based in the actual
conversations
with your customers and there's a wealth of information there that we leverage
for this
purpose. Kind of you're pretty familiar with this at this point, right? Pretty
much. But
anything specific about this that jumps out at you?
So one of the failures that we had that jumps out at me, it was that when we
initially adopted
and began to implement support logic, we weren't quite using it the way that we
was suggested
to us. We had a little bit of arrogance on our side and we thought, you know,
we're
just going to inject this into our current case management process, manage it,
figure
it out, it'll be fine and it really wasn't. And a big part of that was we were
very set
in our own ways and we used email alerts to notify us when cases were
potentially going
to escalate and we may have taken action on them, but one of the missing pieces
was we
weren't letting the tool know that whether it was correct or not. It wasn't
learning
from what we were doing. And we worked with support logic more, you know,
support logic
was great in partnering with us and began to build in buttons in the email
alerts to
allow us to do that. But we still weren't fully adopting to that, which meant
that for
a better part of a year, year and a half, even though we were using the tool,
even though
it was actually identifying potential escalations, it wasn't really learning
from it. And it
took us probably a year and a half of adoption before we began to really feed
back into the
tool, began to kind of build upon a lexicon so it can begin to identify even
greater
expanses of cases that could potentially escalate. So it's a big learning part
for us and it's
a big piece that really stands out to me now.
Yeah, it's interesting. Adopting a new tool is, you know, it's only, technology
is only
part of it, right? The change management piece is the other part of it. And
building
a proactive escalation practice when you're used to just having a reactive one,
right?
That's a big part of that change management as we go through this.
Yeah, pretty key.
Okay. So why don't you speak to us a little bit more in detail about kind of
what this
looked like for you from kind of inception to actually achieving the results,
right? Talk
a little bit just now about the changes that you have had to make, but maybe
get into a
little bit even more detail about the teams that are using support logic, how
they're
using it, the workflows that they've followed, all of these things.
Sure. For us, as I mentioned before, escalations was a huge pain point for us.
We were seeing
anywhere from 10 to 13% escalations rates for our support cases. So we needed
something.
We needed some way to really begin to reduce that and throwing people at it,
throwing processes
at the time wasn't really working for us. So finding a tool like support logic
was key
to us. And for me, it's near and dear because it was the second AI tool that I
implemented
for our company. The first being a chat bot that was customer focused and I
always felt
support logic was kind of internally focused. Something that helped us drive
productivity
and efficiency for our teams, which was key for me. In finding the tool, in
using it,
in demoing, in kind of the POC, everything checked off in terms of boxes. The
biggest
challenge for us was the adoption. And it wasn't just a brand new tool. It was
the fear
of the tool, the fear of AI in general. Is this going to replace me? Is this
going to
replace me as a manager? Do I need to now look for another job?
So for period of time, it was trying to negate the fact that AI was replacing
people. This
was really just a tool that was used by people to really drive efficiency and
productivity.
So that was a key piece of it. Beyond that, it was breaking the chain of what
people were
used to doing. People had a very formulaic way of managing their cases. And
unfortunately,
because of that formulaic way, you begin to get complacent with making
improvements and
trying to inject a brand new tool into that process flow was difficult. So it
took time
to adopt, it took time for people to comprehend and recognize value. And we had
to do that
by breaking up into smaller teams. So instead of a global release, it became
smaller groups.
Groups like our number-porting team, which began to use it and adopted it whole
heartedly.
It's had massive change in that team. Before you would hear number-porting talk
about the
team, people had that grown. That grown that you wouldn't hear, but it was just
part of
them. And now it's a team that's proud of the work they've done. They've been
recognized
globally by the teams in terms of the numbers they produce. And the morale on
that team has
switched over, just based upon the usage of one tool.
How did you leverage the success of one team to -- and did you to drive
adoption among
the other teams? Because that's always -- you're always trying to get that
initial success
to get over partially the trust barrier that you're talking about. So how did
you leverage
that in that capacity and how did that go?
We actually did. Not so much with support, but aside from technical support
cases, the
other cases that tend to be problematic, as you can imagine, are billing cases.
And our
billing team began to see the drastic change in our number-porting team going
from 13%
down to 2%. And just the change in morale, the change in how they were being
spoken about
by the company, and began to say, "I think we want to take a look at this tool,
too."
So over the past several months, we've been looking at how they've been
adopting as well,
how they've been using it, how they've been applying it to monitoring their own
billing
cases also. So it's had this kind of a domino effect in terms of other teams
adopting and
really wanting to begin the user tool to drive their own success.
Excellent. What about these specific numbers that we're talking about here? 75%
reduction
in escalations. We've talked about that a little bit. But improvements in
backlog and
things like first response, where have those kind of fit in into what you've
been doing
with support logic?
So the reduction in backlog, I mean, obviously, we've had massive reduction in
terms of our
escalations. Those escalations have an impact in terms of resources and time in
looking at
other cases. And oftentimes, escalations create additional
cases down the road. So it's given us a chance to work on the massive backlog
that we had
before, reducing that considerably, streamline that process so that we really
are back to
a point where it's one case, one solution as opposed to one issue generating
multiple
cases, which is really streamlined and really made us a more efficient
organization.
So solving, in a sense, one problem has made it easier to tackle other problems
Paying attention to the one problem has allowed us to reduce the need for that
one problem
festering into multiple problems.
Awesome. Anything else that you'd like to call out just about the team's
experience?
You know, it's not just been the team, but the company itself has recognized
the value
so much so that we've partnered with support logic. It's now a part of our
product. So
integrations into our tool to not just deliver the same value that we found to
ourselves,
but deliver that same value to the customers that we have also.
The teams have been, as you can imagine, the reporting team feels like support
logic is
a bit of a godsend because it really did change their own trajectory beyond
just the
demand of being happy with performance. The employees are much happier too. And
that's
that one piece that really can't kind of quite quantify or place value on. It's
just
a happiness of those employees to come to work and to work on the cases they do
Excellent. Yeah. As far as how the team specifically is using it, right, you
talked initially
about how they're using it, the alerting capabilities, but now that you've had
them
really start to perform the actual a bit more of the kind of those recommended
workflows,
right, what does that look like? So they've actually really begin to dive into
the tool
a little more, building their own, so each teams are now looking at building
their own
specific alerts. So based upon a different segmentation of customer, how they
want the
tool to really begin to react based upon what it's seeing, alerts, words,
phrases, in order
to generate those concerns in terms of escalations or potentially can turn. So
everyone's really
been pretty key in terms of finding different ways of customizing using the
tool.
Excellent. Now, if you were to make a recommendation to any of the folks here
who are looking to
implement like kind of a proactive solution to escalations, like what would be
kind of
what are the top two or three things that they should think about? And I'm not
talking
about just what as far as like implement support logic, right, talking more
about what are
the processes that they need to implement in order to be successful in becoming
more proactive
in their approach to escalations? Well, if you were to work with a company like
support
logic, first off, I'd say, no, trust the process. Obviously, they're told they
know
what they're talking about, so that they make a suggestion. It's at least worth
considering
and maybe even trying out. Start small. Trying to implement something globally
at this scale
is going to be difficult to do. Not to mention difficult to adopt for everybody
involved.
Being small, showing a proof of concept, showing the kind of success, allows it
to begin to,
with our other teams, domino out. And that has a greater chance of not just
being successful,
but identifying gaps that you haven't thought about initially and then filling
those gaps
in as you begin to spread the scope farther and farther.
Awesome. Thanks. Yeah. Okay. So, just kind of briefly with regards to support
logic itself,
one was certainly talking about how there are some recommended workflows, right
? And
to trust the, let's say, trust the vendor a little bit when they're the expert
in something,
right? And how they recommend you should go about, especially when we're using
our platform,
right? And so, what does that typically look like with support logic? Support
logic generates
a number of distinct signals, right? I talked about the sentiment scores and
the needs
attention scores and the escalation predictions, which are a part of that. All
of these signals,
there are workflows that you can perform within the platform, right? Which are
incredibly
important. However, you can't be in another platform all the time, right? Your
teams are
inundated in all of these different tools. And so, what we always recommend is
leverage
the support logic alerting capabilities extensively, but use those to drive you
into the tool,
right? So that you can then perform those workflows that feed information back
to the AI,
as you were talking about, and ensure that there's a continuous improvement
processing
process happening, right? So, this is, it's actually relatively straightforward
, right?
Support logic points things out to you. Take a look at them, determine if
something needs
to be done to drive this towards a better outcome, and then based upon that,
right?
So, let's just move on to the next one, right? And our customers have seen
again and again
that simply by implementing some rigor around this, right? You're going to
catch cases that
would have snowballed into an escalation, right? So, that's basically, it's
kind of
as simple as that with regards to, you know, using the product. And so, maybe
let's just
open it up to a question or two. If anyone has a question for Ken about what 8x
8 is
done with support logic or proactive escalation management that they've
implemented, please.
All right. Any, yeah, please, go ahead.
[inaudible question]
So, it became clear to us during multiple QBRs, multiple sessions at the office
that escalations
was a problem. And we weren't clear why, because we felt like we had a pretty
good handle on
terms of our case management and our milestones, but nevertheless, we were
still seeing a high
rate of escalation rates. And it was coming to conferences like this where we
began to
see different tools where we discovered support logic. And it began to make
sense to us. At
the time, we had just started to scratch the surface of what something like AI
could do
for us from a customer experience perspective, from a resolution, solutions
perspective for
our customers. We hadn't really considered what it could do internally as a
tool for
us yet. And we began to kind of take a dive into what it might potentially do
and how
it could drive some productivity and efficiency. And it really seemed to make
sense. And at
that time, we didn't have any reason not to give it a try. So the team was very
open with
letting us try a POC, which, by the way, it's what I do now. With any AI tool,
I won't,
I'll be happy to take a demo. I'll talk to you for as long as you want, but
unless you
can show me what you're telling me and claiming, I'm not going to necessarily
believe anything
you have to say. So we did a POC with the tool, and it really allowed us to
visualize what
the value could bring. And that kind of sealed the deal for us. So that's how
we decided,
if we have a tool for our customers, maybe it's a good idea to have a tool for
our agents
as well.
>> You reported the interaction or as far as doing the portfolio, you are
likely to be
the many predictors in the technology, anything like that.
>> Yeah. It's actually been a bit of a pain point, to be honest. It's trying to
determine
how many predictions have resulted in escalations and trying to do that
historically. And one
of the problems we've had before is getting that data. So very early on, it was
a, it
was a, I almost call labor a love because it required so much effort from the
engineering
team on the SmartExide to deliver us the raw data and then, you know,
generating our own
reports based upon that. And I'm very happy that the tool has grown, innovated,
and now
we're much, we're able to get a lot more of that data and that those metrics
out near
real time.
>> Do you understand the leaders there?
>> Absolutely. For us, we've been on a digital transformation journey for quite
some time
now. And I've almost kind of want to say that we've also been on this separate
journey
of AI adaptation. So the work that we're doing from an AI perspective is very,
very crucial
to our continued success and support and our executives are very keen on
understanding
what we've been doing. And, you know, honestly, being able to share that with
our customers
as well, what our journey's been.
>> Any other questions? Please, go ahead.
>> [INAUDIBLE]
>> It is.
>> [INAUDIBLE]
>> I think you said from an account.
>> From an account perspective? Not quite yet. We began to think about that
when originally
the, well, not originally, but when the account health score began to pop up.
And that's partially
what's been integrated into 8x8. So it is an area of interest for us. We just
haven't
done it quite yet.
>> All right. Well, thank you, everyone. Thank you, Ken, for so much for
sharing your story
on here and being a part of this conference.
>> It's been fun.
>> Yeah. Absolutely.
>> Thank you.
>> All right.
Thanks.
[BLANK_AUDIO]