Fundamentals of AI in CS
Join industry leaders and AI gurus for a dynamic panel discussion on how to fundamentally think about and leverage AI is CS. Our expert panelists will share their experiences, strategies, and best practices for integrating AI into CS operations, helping you understand the core principles and potential pitfalls. Whether you’re just beginning your AI journey or looking to refine your approach, this session will provide valuable insights into leveraging AI to elevate your customer success efforts.
Well, thank you all so much for joining us for an amazing panel on the
fundamentals of AI and customer success.
I hope everyone had a good lunch. Yes, everyone gets some good food.
Pokebowl was amazing. Any other options? Was it curry good? Pasta good? I heard
all good things. So awesome. Pasta was good.
We'll hope you've all had a chance to eat, grab some more coffee, maybe get
some water.
But super excited to kick off our afternoon sessions. We have an amazing panel
of folks,
some of which are on our keynote stage, Ori and Amit this morning, as well as
Maria,
who are very excited to have with us as well. I'll let everyone do a quick
intro,
but just wanted to kind of kick things off. Two quick housekeeping things.
We have a slide-o polls for questions. So in the app, if you open this session
in your agenda,
you can leave your questions in the Q&A and polls feature. We'll have a ton of
time to ask
amazing questions of our panelists, so make sure you keep those coming
throughout the session.
And then we also have those breakout surveys after each session, so just make
sure you fill those in
for a chance to enter our raffle. I think those are my only housekeeping things
, so I'll go ahead and have a seat.
But super excited to kick off the panel. Just to introduce myself, my name is
Tori Jeffcoat.
I lead the marketing team at Gaincide around our CS and AI programs. I don't
know if anyone is
here for my first session of the day, but have been in this track all day
hearing such amazing
things about AI. Really excited to hear from all of our panelists here today. I
do have a quick ice
breaker question for everyone to answer as we introduce ourselves. So our ice
breaker is,
what is the most exciting or funny use of AI that you personally have had in
the last couple of
months? I can also answer this question myself. I have a three and a five year
old, so I've actually
used it to write bedtime stories with their name and based on their interests.
My three year old
loves dinosaurs, so we have a lot of dinosaur stories. But that's my own answer
to that question.
And with that, I'll hand it over to Ori. So I'll actually start with the ice
breaker. So it was
Halloween in the US recently, and I had to get on another call, and there was
another ice breaker.
So I needed a joke for Halloween, so the joke that ChatupT gave me was, why don
't skeletons
fight with each other? Does anybody know? Because they don't have the guts.
I'll worry, by the way, now as a VP of product at Gainsite, overseeing AI,
customer success in Staircase, and prior to that, I was the co-founder and CEO
of Staircase.
So my story with AI, my father just celebrated his 70th, so we took a bunch of
old photos from
the 50s from when he was a baby and a toddler, and we animated and revived him.
That was a lot of fun.
Actually brought tears to his eyes, so that was really cool. And my name is
Amit. I am the global
VP of healthcare customer experience with Clarity. It's a very long title
because we're a verticalized
company. We do cybersecurity for critical infrastructure, and I run a specific
sector,
which is healthcare. So this is why so long. Awesome, nice to meet you.
Thank you. So my name is Maria Bandarenko, and I'm working for SAP Signavo, and
I'm doing
mostly operations and bringing all the data dreams come true. And my funny
story, I want to say
it is funny. First of all, I'm using AI a lot for my cooking ideas. When I'm
out of ideas, I'm just
asking, I'm asking, "Chadgipiti, okay, can you just come up with something?"
But yeah, the corporate
funny story is that I found the AI that is actually coming up with the songs as
well as with the music,
and I just fed it with some names of our teammates and then fed a few phrases,
and then it came up
with a very nice song, and then I also posted it on the channel, and they were
very surprised
with my hidden talents, but of course, I thought that is actually not me. I'm
not that talented
yet, but it was quite funny. Awesome. Well, it's so great to meet you all and
have you as part of
this panel. I'm going to kick us off with a question around building and
choosing the right AI for
post sales. Everyone on this panel is either building AI in your case, or
having to choose
and navigate AI for CS teams. Curious when you're thinking about AI, how do you
prioritize what
are the most important critical ways to use AI to get the most value within
customer success?
We can just go through the panel, or if you want to kick us off. Yeah, so I
think a couple of things.
One is when we talk about AI, I think everyone today goes straight to Chadgip
iti and LLMs, but
there's a lot of other AI out there from various kinds of machine learning that
are more statistical
oriented or things like that. I think it's really important, like anything,
that you're going to
do in your business is first of all to choose the problem that you want to
solve. It's just,
there's no point to just spend time. A lot of times you can waste time in AI
also, going down
what I call rapid holes in data science, and we can talk about that. It's
important to choose a
problem that you think you can, one is important to you, two is that there is
some technical
feasibility around it that you can see early results soon. I give you an
example from staircase.
We had a rule in staircase where we would try a data science feature, and if we
didn't get results
in two weeks, we would drop it. So it was, and again, if it was a few more days
here and there,
but the reason behind it was because very often we found that we could easily
spend months on a
project and not get anywhere. So I think it's really important to kind of
balance those two
criteria. So to me, it's about two things. One is increasing operational
efficiency. That's one
way of showing ROI, and the other one is like direct impact on revenue and
finding churn,
because I am in CS after all. So increasing operational efficiency, trying to
get any
manual labor, repeatable repetitive manual labor out the door as much as I can,
liberate my CSMs, if you will, note taking, inputting data into CRMs and
systems. Nobody likes that
stuff, and it's a time sucking machine. So this is the first thing, and
staircase is not the only
tool we're using in AI. We have a how-to video creation tool based on AI. We
have like an internal
AI knowledge engine inside the company that we use for CSMs and other employees
to ask questions
about the company, about our customers, whatever, natural language questions.
So really keen on
achieving this operational efficiency. This internal, so it saves headcount and
so on and so forth.
And the other thing, obviously staircase in that context is actually insights
that impact
revenue and churn. And I just spoke about it earlier today on stage. Some of
these insights are
really key in saving accounts before you lose them. So these are the two main
areas for me.
Thanks for sharing. We have quite similar concepts. So especially in operations
, we first evaluate
if AI would bring impact, valuable impact for our customer success. And if we
see that it would
reduce manual work, it would automate certain things, then we go for it. That's
a very important
thing, as you already mentioned, that you need to fail very fast. And our
directors are also
pushing a lot to fail very fast. And then if it's not working, then just go for
that. And that's why
we are adopting this approach and also doing a similar concept there.
So it sounds like fail fast and it's not working and it has to drive actual
results.
Better succeed fast, but if you can't.
But it sounds like getting ROI is kind of the real value of AI for all of you
when you're
considering choosing tools. Curious when you're thinking about what those
results are. If you
could share some examples of how AI has impacted your own teams. And Maria,
maybe we start your
end and come back this way this time. Sure. So one of the recent examples that
we are leveraging is
the cheat sheet that is also done by Gainsite, where we have the crisp overview
of the customer
profile and where the CSMs as well as directors can speed up their routine by
saving the time.
And then they can also see the whole overview, the whole profile of the
customer in seconds.
They can be prepared for the call in a minute. Just by reading the whole
outcome. What are the
key projects? What are the key risks? Is there any renewal discussion going on?
And then with all
this information, they are already prepared for the call. And okay, it's
already done. And then
they can finally start focusing on the strategic part of the conversation so
that they can prepare
more for this strategy and be in the real partner for the customer so that they
can
work on the success together with the customer. So we have a very similar
concept with their case.
So you get this cheat sheet before a meeting, it saves a lot of time. But I've
had issues with
CSMs that were a little bit overloaded. And so we used other functionalities
like
reports on open items at the end of the week, stuff that really helped them
make sure they don't
drop anything. And also we have just like last week used the efficiency report
and efficiency report to actually work on our headcount growth plan. So we compared
how much effort and
energy were actually investing in certain customers of certain tiers against
what they should,
what kind of investments we should be making in these customers according to
their tier.
So definitely use that as well to show ROI like actual financial discussion
with the CFO.
And then lastly, I mean, even if you say one customer thanks to an insight that
came in on time
that's already the best ROI ever, right? So it only takes one customer to prove
it.
Awesome. Or obviously working against it, but curious if you have any good
results from
customers of staircase or key use cases you want to share here too.
Yeah, so I think when it comes to ROI, I look at it like from two different
aspects. One is
cost savings, which I think we talk about, which is actually probably easier to
measure and to,
if you need to build a business case internally is to do that one. The second
is obviously
revenue. I mean, to like saving a customer or being proactive around a downsell
or something
like that or selling a customer. The problem there is what we call the
attribution problem.
It's very hard to kind of narrow down to one insight or one thing. So it's more
of an overall
view of what happened. I think a bunch of things that we see in the field. So
one is we see incidents
where an alert came in time. So an example could be one of our customers. The
CSM was on PTO two
weeks before renewal. And a very upset email came in on a Friday from a very
important customer
and that customer. And basically we caught that extremely negative sentiment
and sent that email
to the chief customer officer who then pulled in the VP of engineering and both
of them jumped on a
call and they saved the customer. Now you could say, hey, you know what that
didn't save the customer,
but definitely it had a very material effect on the overall outcome. So that's
an example on the
revenue side. I think on the efficiency side, we had another customer who had
five tiers or
segments for customers. And when you build segmentation, you're not going to
apply the
same amount of time per customer. That's the whole idea behind segmentation.
And when we did an
analysis in Stakecase, we actually found out that two segments had the same
amount of time per
customer. So realistically, they weren't two segments. So out of that analysis,
we recommended
that they either merge the two segments or change something fundamentally and
how they're serving.
So that was another insight. The third, we also use it internally. And when I
started kind of
looking at the product team, I'm a big believer in in market product. That
means interacting with
customers. And we did an analysis on ourselves, like how much time we're
spending on meetings with
customers versus internal meetings. And we found out that we need to kind of in
certain areas,
we need to kind of push it up a little bit by 10 or 15%. So there's another use
case.
Awesome. That sounds like great ROI and great use cases as well for AI.
Despite the fact that we are starting to see that ROI and see those results, I
think there's
often a lot of barriers and fears that still exist when it comes to AI. One of
them being
replacing kind of that human element and with digital and automation kind of
taking so much
of that off of our plate. Curious to hear from all of our panelists how you
think about AI in terms of
coexisting with the human side and how that AI and automation can complement
instead of replace
some of those human pieces. And Ori, maybe we can start with you on this one.
So I think we all have a little bit of fear AI. I mean, I think it's justified.
From like,
hey, the robots are going to take over, are we going to have jobs in 10 years,
things like that.
But the more I spend time in it, I actually realize or think that it's really
far away.
I think we all play around with chat GPT and see that, okay, it can do so much,
but it also
can't really replace a person. But I think it can make our life easier. And I
think that there
are certain areas where I believe in kind of the 80/20. So a lot of times AI
will do 80% of that work,
but you have to do the fine touch-ups to something you're writing or something
that you summarize or
it could be that you have a task and otherwise it would take you two hours. So
I'll give you an
example. We came back from an off site in India and I had a whole bunch of
thoughts in my head
right and I didn't really feel like organizing them on a piece of paper. So I
just wrote them
kind of a stream of consciousness and then I fade it. We have our own instance
of chat GPT,
so it's secure and all that. We fed it in there and asked, can you arrange this
into kind of a
cohesive structured product thing? And it was great, right? But I still had to
kind of, so again,
so I think there's a distance, there's time before that. I think that the area
is the scaled side,
which I think is not a surprise. I think scales where we can use AI and use it
to personalize.
Today's scale is not personalized and I think that the opportunity is there.
So honestly, no fear on our side, we're pretty much AI junkies in my company
and especially my team.
If anything is building, it's building the trust. So before you start using it,
people are
a little apprehensive. I mean, is this trustworthy? Is this data usable? I mean
, and there's a lot of,
you know, your AI results are only going to be as good as your data. So there's
got to be a lot
of infrastructure work to make sure that you've rolled it out. People can
actually trust it because
there's, you know, you know what they say, there's no second chance for a first
impression. So if
anything, that was not issue because, you know, we were prepared for it, but
that was the area that
we kind of were worried about, but not really about replacing people, our
customers,
for better than for worse, they love to meet us, they love QBRs and that's not
going to change.
We're not going to have AI run these things. If anything, they're going to have
more time for
these meetings. I would agree. In overall, I would not be hearing from the,
that's AI would replace
us all or so. I would look at this situation from a different perspective by
using AI so that
it amplifies all the strengths that we have. And for example, especially for
high-touch customers,
where we have direct communication and collaboration with the customer. And I
think that AI and also
a customer success should work closely together. So that, for example, AI is
doing the data heavy
lifting while customer success managers are doing the strategic, empathetic,
collaborative job with
the customers. And then with this combination, like it's like a perfect
marriage of two sides,
and then with this combination, we can really benefit out of it. So I would
never think that it's
actually would replace some of the customer success jobs there.
It's good to hear we have confidence. We'll still have jobs. That's good.
Curious though,
as we've kind of been implementing AI, we feel confident in it overall, it
sounds like.
Doesn't mean there aren't Roblox and hurdles and maybe challenges to getting
either
it adopted effectively or delivered it in the right ways for customer success.
So curious from maybe a challenge is Roblox hiccups or learnings that everyone
has had in
leveraging or rolling out AI within your business. Curious if you could share a
little bit about
what any of those Roblox has been. And Marie, maybe we start on here at this
time.
Sure. So I can say from the corporate side, that was quite challenging to get
the approval from
all the departments to get started. Even though we know that all the AI things
are GDPR compliance
to you with all the concepts that SAP has the strict legacy, it's very, it was
quite challenging
to get the approval, but then afterwards it was done and finally everyone is
benefiting out of it.
So that was the first challenge. So data protection, data privacy. Second
challenge that is still going
on and I think that we all need to work on is the adoption of those AI features
because
it is always nice to have those, but if no one is using that or if no one is
actually benefiting
out of it, then it doesn't bring any value. So we all need to push our CSMs to
use those
features. So then they see the value itself and then with these sailing points,
then they can also
spread the word of mouth and then also share it with all the other CSM that,
okay, I'm using
this cheat sheet because it saves me so much time and so on. So those are the
two challenges
we are having right now. No, the first one is done. Just the adoption. So
adoption is, I think that
would be always the quite a hard topic. So for us, I mean, it's pretty much
what I said earlier,
it was really setting up the data right and making sure, you know, we don't
cross paths
between data points that shouldn't be crossing paths. So data attribution is a
problem in many
different AI tools. And as I said earlier, the result is only going to be as
good as your data.
So that was one word block and we're so actively working on it even myself.
The other thing from a privacy standpoint, to your point Maria,
for this to be effective, we need people to open up digitally, right? Because
it's all about
getting that cumulative insight of all of our interactions with the customer.
So I've had some
issues with that. And you know, I even had one of our C levels asking me, I
want you to remove me
from that list because, you know, some of the stuff I'm discussing with
customers is I
sensitive, I don't want to, you know, I don't want to see myself in
notifications. I don't want
to see other people, you know, reading my emails basically. So that was, but I
can safely say it was,
you know, very few instances for the most part, you know, people are fine with
it and they understand
the benefit from it. So, yeah, these are the two things. Privacy a little bit
and setting up the
data, making sure it's reliable. >> Yeah. So again, I think to echo, Maria said
, I think
I see a lot of organizations challenging the legal privacy. And I see it's kind
of like a battle
between the buyer, the business, you know, folks like yourselves who want to
use AI because you
have a, you know, need for it. And then there's legal and privacy that their
job is really to
kind of secure the organization. I'm not even talking to info, I mean, that's
another topic. But
even if you have everything, GDPR and CCPA and, you know, data centers and all
that stuff.
So it really is kind of a, it's like almost like a little bit of an internal
battle where,
you know, you have to drive innovation because you're under pressure to drive
business results,
right? I mean, you're being asked to improve GRR, improve NRR without having a
bigger team,
right? So you need technology, but then when you bring technology, somebody
says, well,
that's not secure. So I think that that's kind of a little bit of the friction
that we're seeing
in the field. But we're seeing at the end of the day, more and more ways to
overcome this,
either through different levels, right? So you can, for example, on meets
example,
you can exclude certain people or you can mask certain things. So at the end of
the day,
you have to find kind of a harmony inside the organization. I think that's
really the key
because it also, it's a directive, right? If you're not going to do it, you're
kind of
going to stay behind if to use that term. Yeah. Great. Just to add to that, I
think, I mean,
to your point, there are compromises that could be made. For example, those,
you know,
notifications with sensitive information are only open to a very specific group
of people
that companies, I think everyone at the company can see these things, but it's
the people that
can actually make an impact, usually leaders. And obviously, if you were a CSM
and it's your
account, certain alerts you'd be seeing, but there are ways to do it safely and
securely,
while also, you know, getting the benefit out of it. Just to share with
everyone.
Yeah. It's awesome. So it sounds like privacy, security concerns, adoption
concerns,
I think you mentioned as well, Maria. Are there any kind of ways that you're
seeing to kind of
get over those challenges or anything you're seeing work really well to
overcome some of those
common barriers in NCS, maybe based on your experience with customers, what you
've kind of seen?
Yeah. So I think like anything nowadays, anything you're going to buy, right?
You need a business
case. And, you know, there's a framework. You know, we have a framework in
place, but I think,
you know, happy to share kind of the, I think at the end of the day, it's an RO
I question. If
to use that term, even though it's not pure ROI, but still again, you're going
to come and say,
if I implement this, I can do so-and-so cost saving, so hours, dollars, and so
on. And I can
have an impact on revenue that might take a little bit longer to show and prove
, right? Because
if you're talking about annual renewals, you know, you'll only see that impact
over a certain
period of time or things like that. And some of it's going to be quantitative.
You could actually
show, you know, from a back of the envelope calculation to an actual
statistical, you know, analysis.
And some of it's going to be more qualitative, like based on stories that we
talked about.
So I think that those are, you know, key areas that, you know, you need to be
prepared in this,
you know, day and age. Even smaller organizations are today being asked by CFOs
and CEOs and so on
to justify a purchase of a technology. So, yeah, I think that's one area that I
think,
you know, and we have experience on that and happy to share, you know, around
that.
And you mentioned some of those levers. I mean, right, taking CS, sorry, C,
sweet executives, CS, automatic, I'm in there. I'm not taking them out kind of
of some of those
conditions. Any other, you know, to to you to recommend to kind of overcome
some of those barriers?
So first of all, you know, once to be perfectly honest, once some of our
business leaders saw
the outcomes of this, they were just astonished, they couldn't get enough of it
. So sometimes it's
just, you know, put them in front of the data, in front of the insights and you
blow them away.
And then they're willing to pay the price, you know, both financially and from
a privacy
perspective, so to speak. Otherwise, it's really about being very strict with
what you share with
whom and we're very strict about it. Because it is, you know, a little bit
intrusive, I think in a
fair, balanced way, in my opinion, that's why I'm using it. But we do want to
respect, you know,
people's privacy and so on and so forth. So to be very strict about it, I've
had, you know, people
add, we have this Slack modifications group for these extremely negative
messages and other stuff.
We've had people add, you know, people that were not permitted to this, like
immediately,
you know, you remove them and you have to please don't add anyone without my
permission. There's a
lot of, you know, sensitive information running in this channel. Please be
aware of that. But
yeah, other than that, I mean, the value, the value to the business was so
great that other
people were willing to make that compromise. On our challenge that we have an
adoption, we usually
go through the both ways. So bottom up and top down approach. For the bottom up
, we usually try to
find the advocates of the features or of certain things. And they are usually
also those CSMs.
They're actually then either leading some sessions where they share their
knowledge or they are also
sharing how they're using Gainside, for example, on a daily basis. And then
with that, they are also
doing some kind of advocating on those AI features that we are using right now.
And talking about from the point of view of the top down approach, we usually,
so are, for example,
directors, customer success directors, they're also collaborating a lot with
the other lines of
businesses. And they also see that other lines of businesses are using AI. And
it was the recent
example. Yeah, that's one of the directors was just coming over and saying that
, hey,
those guys, they have in their Gainside instance cheat sheet, and we do not
have it. Why and how
come? Can we implement it? And then, yeah, we were just about to open it to all
the audience of
Gainside and then we were just like, okay, you need just need to be patient.
Why wait a sec?
But in overall, once you have this patient from also from the directors, and
you see that they
are going to be using it, then everyone is have this kind of a spark within
themselves. And then
they can also leverage these features. Awesome. Well, we have so many good
questions. I'm seeing
scrolling on the slide over here. So I want to make sure we have some time to
get to those. I'm
going to ask one more question of our panelists first. Thinking ahead and
looking to the future
of AI, what are you most excited about for AI to bring to the table for
customer success?
Sure. So, from my perspective, I wish I would have more data consistency, and I
wish that AI
would be able to bring this data into the right buckets and then can process it
more effectively.
But if we're talking more realistically, I would love to see more inputs on the
cross-sell and upsell opportunities. Yes, we do have a lot of call to actions.
We do have a lot
of automations behind that recognize that. Okay, there isn't a potential upsell
or cross-sell,
but still, I wish we would have more of that. I wish we would have more
proactive approach rather
than reactive, as it's quite important to be very sensible to all the changes
the customer has.
And then I really hope that with the features that Gainside and Stakey's have,
I wish that we can also implement it as fast as possible to them can also
benefit out of it.
I think I said my piece on stage earlier today, but to me, it's all about
taking the digital
journey to the next level. So, if you will, introduce a Netflix or a Spotify-
like experience,
the customers in B2B. So, it's not just about learning and creating the
visibility,
but also proactively approaching the customer's in-app. I mean, if you did that
, you might also
be interested in that. Oh, I see that, according to the role, you should
probably be interested in
that this is some of the best things other people in your category have done
with this feature.
And then, you know, you know, fusing all these data points together about the
health of the
customer, about usage drops, about sentiment and topics that come up on calls
to create
that proactive messaging, going back to the customer to put them back on track,
and to make sure they actually, you know, are doing the right things to meet
their KPIs,
and their goals, which they set when they bought the tool, right? They had to
make a business case
to their leadership to get a product. So, let's remember what these things were
, and let AI
kind of help you navigate towards those goals. This is a very, you know, kind
of a, I don't want
to say far-fetched because I think we're going to get through it a few years,
but that's what I,
that's what I see AI doing in the future. Yeah, so I think the agentic approach
, which I think a
lot of folks kind of have been reading about, yeah, I'm 100% on board of that,
and happy to share
more thoughts on that. I actually think to take one step back, I think a lot of
organizations
know what needs to be done, right? You kind of have a real good understanding
of your business,
and you have like really good processes in place, but half of the time we just
don't know if they're
actually fully, you know, occurring, you know, happening. And we do like a
churn analysis or a
win-lawness analysis. I would say 80% of the time we found that we find out
that some stage just
didn't happen, right? Or something that I'm not talking about those extreme,
you know, out-of-the-box
situations, right? So I think one promise of AI is to kind of do this constant,
I would say like
almost like a real-time audit of what's happening. Like is the organization
working the way it should
be? You know, I kind of, if I think about a car, right, there's all these
computers now in cars,
and they kind of, every time you know your tire pressure is out, or this is out
, you know,
they kind of alert you and they let you know. So, and there's a reason, right?
Because somebody
said that the tire pressure needs to be, I don't know, whatever it is, 40 psi,
right? So the same
thing in your organization, you have a list of things that need to happen, you
know, you guys are
experts, but you just don't know half of the time if they're happening, if
people, you know, are
following through or if they have other challenges. So I'd say that is, you
know, one core area. The
second is looking, you know, today it's very much like a question and answer. I
write a question,
you know, in LLMs, and I get an answer, but wouldn't it be nice if, you know,
they kind of anticipated
my question, right? If the data already came and showed me that, you know, that
there's, you know,
a skew in the efficiency, then maybe it starts to kind of do the second or
third level of quaring
for me. So it saves me time. So there's that aspect of that, and we are working
on those things.
It connects to an agentic flow where the idea here is that different parts of
the platform are
asking questions and talking to each other to kind of almost mimic like how a
person thinks.
But I would say that those are, I think, things that are not, you know, we're
not talking about
robots in the future. These are things that could happen very soon and could
really drastically
change the operational view of an organization. Awesome. Well, I'm very excited
for the future
with AI. All of these great use cases included, but I think there's so much we
can anticipate from
it. I do know we have a couple of really good questions that have come in on
our slide of if
we can throw those up, perhaps, and maybe go through some of those with the
panel on the time we have
left. First question, Maria, I'll ask you this one. If you could do one thing
with AI with your
CS team and you don't have staircase, what would it be? So I would say we do
not use take as yet,
but we do have some features within Gainside. And I really, I'm a big fan of
the AMP-S analysis
and also AMP-S, yeah, customer satisfaction. So then Gainside has a very nice
feature that
allows you to bucket all the customers and to see if they are really in-fence
or if they are
the ones that need support and then based on, like, based on this revenue
metrics, you can also
identify what kind of customers you would like to focus on. And as well as if I
'm coming back to
the AMP-S, it also gives such a nice opportunity to see what kind of things and
what kind of features
or main things customers are struggling with and what do they appreciate. So
then you can focus more
on the conversation and on leading the collaboration with the customer on that.
So yeah, the AMP-S analysis and all the analytical stuff is my passion there
and I really like that.
This next one, Amit, I'll ask you this question. Since it sounds like you use
AI pretty heavily
internally, but what are some of your top tips to prepare CS teams to use AI
most effectively?
So first of all, you need to get them hyped on AI and show them, you know, how
much work you're
going to save them if they adopt AI and this is the best seller, right? And as
long as there's
the wheel, there's a way, right? That's what they always say. So this is the
first thing.
Sell it to your teams. And then after that, you know, I heard it on the other
session that we
were listening in on. AI is not a silver bullet yet. Maybe at some point it
will be. So, you know,
CSMs needs to understand that this is going to enhance and augment them. It's
not going to replace
them. And so, you know, they have to be mindful. Again, I'm repeating it like a
broken record,
but it's only going to be as good as their data. So if they don't keep data
hygiene,
unfortunately, I mean, there's still going to be data hygiene tasks, then
things could go sideways.
So these are the two things like understand what you're getting, but also what
you have to give
in return for this to be effective and to keep serving.
Awesome. This next one, oh, the question's changed. Okay, there we go.
Collecting all the data on customer interactions who are in your organization,
who they had contact with, is GDPR like a topic with customers that comes up?
Or maybe you could answer this one. It's always. I would say there's never a
case where it isn't.
And it doesn't matter if it's, you know, staircase or a gain site or whoever
you're talking to,
you should definitely make sure that, you know, you have first of all those
credentials, right?
You know, from SOC to GDPR, and if you're selling in the US, CCPA and all that
stuff.
But usually that's kind of the base level. By the way, there is, it really
depends on the
organization's perspective. From a GDPR perspective, there's no problem with
that analysis.
It's, you know, a lot of times people think GDPR is one thing, but really, it's
not. It doesn't
cover this case, by the way. I think there are other cases. For example, in
Germany, there's more
sensitivity to looking at a certain employee level and maybe in France too. And
then, you know,
and if you want to, you know, by the way, GDPR doesn't necessarily force you to
host data in EU,
but most companies want it there. So there's certain, you know, overall, I
would say, you know,
not restrictions, but, you know, organizational philosophy around data and
privacy. Remember
that there is a trade-off, be, you know, you need an insight. For an insight,
you need data,
right? So at the end of the day, we can't create an insight out of the air. We
do have to be very
conscious about privacy, security, and all that. But there has to be a trade-
off between the two.
And once again, we are, in like many organizations, we are GDPR compliant and,
you know, can still
do the analysis of interactions. Awesome. And there's, I think, another follow-
up
staircase question is, I'll ask you as well, Ori. I'm going to jump down to, it
sounds like
staircase is very reactive, providing info after the negative email, after the
negative sentiment.
How can it be used to be more proactive and prevent that email from ever
happening?
That's a great question. So when we started Staircase, we did a lot of, like,
on the predictions
side of things. And I always give this example of the weather, right? Like, how
much in the
future can we predict weather, typically, at week to 10 days? So I'd be very
careful of anyone
who's promising you predictions way into the future, especially in the world
that we live
in customer success, where there are so many things that are changing. And
therefore, you know,
a predictive capability way into the future is very challenging. Where you can
use Staircase in
a proactive approach, many times, yes, there is a negative email, but it might
have been the third
or the fourth, right? So wouldn't it be nice to see the first one, right? So we
've done many
analysis in retrospect, historical data, and you find that, just like in the
demo that I did on
stage, you find that churn occurred, and you actually can see it on a timeline,
and you'll see there was
an extremely negative, and then two weeks before that was another extremely
negative,
and two weeks before that was another extremely negative. So yeah, the third
extremely negative
doesn't really help anybody, right? Because by that time, the customer is gone.
But after the
first one, if it was escalated correctly in the organization, if it was dealt
with, and so on and
so forth, you could have, you know, been proactive around that. So it's almost
like zero reaction time
versus prediction. I think this next one's also a technical question. So sorry,
I'm going to send
it your way. How can we combine internal AI engines with Gainsite AI?
Yeah, so there are a couple of ways to do it. One is by sharing the output of,
if you have an AI
engine that's doing something today that you think is useful, you can basically
share the output of
that data, which can be fed into Gainsite or into Staircase, and then can be
used, for example,
like, for example, I'll give an example. So let's say you're doing some kind of
analysis on usage
data through an AI engine, or it's part of your health score, something like
that. You can then take
that output and, you know, put it, and feed it into Gainsite and use that in
another more robust
health score. That's just one example. We don't, at this moment, allow kind of
engines to interact
with each other, and I'm not really sure, you know, how we would do that or why
we would do that.
But if there's a specific use case, then I'm happy to talk offline after this.
And we don't have,
like, a connectivity or shared data, that means with engines for a lot of
reasons that we talked
about. But I would say that usually there's a pipeline in machine learning, so
there's an output
of one model that feeds into another model that feeds into another. So that
could be the way to do
that through the data layer. Awesome. There was a, I'm actually going to skip
down to the how-to
video question. So this one's for you and me. You mentioned using a how-to
video creation tool.
Do you want to share a little bit more about that? Of course. It's a great tool
. It's called
Guide G-U-I-D-D-E. And it basically tracks your key strokes and mouse clicks
within your app,
and creates a video or, like, a slide deck with automatic narration, like
describing what you've
done. And then you can also edit the text and, you know, do whatever you want
with it. So you
create videos in seconds for customers. And also talk about ROI. We actually
use it for our academy.
So instead of using these very specific enablement platforms, we just create
these videos and then
push them in app. So yeah, it's called Guide and it's a great tool. Awesome.
Sounds like a good one.
I think I'm going to ask you one more question at me, and then we'll get you
Maria for the next one.
But I'm going to skip down to you recommending, where do you recommend to start
when you're cleaning
your data to prepare for AI? You've mentioned that a couple times. That's a
great, that's a great
question. I'm sure Ory knows what I'm going to say. But make sure that if you
're selling your product
through a partner ecosystem, and many of us do, especially in B2B, you have to
be very careful
about it, because that could mix data and cross-reference data in a way that it
shouldn't do.
So that's the first thing I would do. And that's the first thing we do in the
circuit. We basically,
you know, removed all these kind of joint domains or joint partners. And it
proved itself, because
anyway, we mostly wanted to cover the customer interaction. So as long as the
customer was
participating in an interaction, we would get the data, it would feed the
engine, we would get the
alerts. So that's the first thing I would do. And the other thing is your
contact data or people
data. This is a tough one, because it's a lot of work. In the industry I come
from, there's
a huge turnover. So it's a lot of work updating the stakeholders, who's the
champion, who's the
executive sponsor, so on and so forth. But it does pay off. So that's the two
things. Partner
ecosystem and contact people or just users. >> Awesome. Maria, I'll ask you
this next one.
For the second question here, when you're thinking about using AI for meeting
summaries, I mentioned
using cheat sheet as well, right, to summarize accounts. Do you typically find
that the AI gets
things wrong? Is it typically right? How do you make sure it is effective for
you and your needs?
>> So the most important thing is that it also mentioned already by everyone
here,
is that it is very important that CSMs are feeding the data in a timeline on a
regular basis. So then
AI can identify and also can summarize all the things in a better way. But for
most of the things
that we've used so far, I haven't seen that there are some big discrepancies.
Yes, we are making sure
that CSMs are aware that they need to cross check if they want to dive into the
numbers.
They can cross check because, again, it pulls from all the timeline activities
and all the emails
that were fed already to gain side. So they need to be aware that there might
be discrepancies,
but still in overall, for most of the things, it is a very big help for both
them and also
directors who are also sometimes interacting with the customers. >> Awesome.
>> Oriel, I'll ask you this next one. I'm going to skip to, I think it's a
second on the screen.
You mentioned business cases earlier. What advice would you give to someone who
sees the value of
something like staircase but needs to position it internally to the finance
folks, the people with
the purse strings? >> Yes. Again, I use this framework all the time. On one
hand, there's cost
cutting. On the other hand, there's revenue enhancements. On the cost cutting
side, just think about
right now, just like, for example, feeding. So with staircase, you can minimize
the amount of
quote-unquote manual note-taking. So I think every digital interaction is
captured by staircase,
but if you went and had lunch or dinner with a customer, obviously staircase
doesn't know about
that. So we do, those notes should be diligently written by the CSM, but
typically, I would say
it's probably 80/20, 80% is digital, and you already removed hours and hours
from a CSM, and we can do
that calculation together. We've done that. The other, the same, I think was
mentioned here by
summarization, prep for meeting. So you can build a framework. I would kind of
look at a CSM as a
day in the life of the CSM and look how much time they're spending on various
tasks and see what would
be the impact of AI on those tasks. So prepping for meeting, let's say it takes
an hour, you know,
when in old school, maybe now it takes 10 minutes, you know, logging data on a
daily basis, so maybe
you spend two hours a day, maybe with AI, you now spend 30 minutes a day, and
so on and so forth.
So that would be the cost cutting side. On the revenue side, we have several
statistical models
based, and we can share them, based on historical examples and existing
customers,
about the propensity of saving an account. And just to kind of give you the
framework,
you know, churn has a lot of reasons, right? So I would say about a third of it
happens
due to, let's say, commercial, right, pricing, issues like that. A third
probably happens of
things that, you know, are market driven acquisitions. I don't know, companies
went out of business,
things like that. And a third is probably related to service and product, right
? And so that third
that's commercial and that third that's product and service, that's an area
where we can definitely
work with. So okay, so you say 60% of my churn, so I'm just going to run some
numbers. Say you have
a million dollars of churn or 10 million dollars in it. So 600,000 or 60% of
that is already something
you can work with. Now let's say you want to be very conservative out of that
60%, let's say
only about a third of that you can save. Again, so if it's $600,000 or $6
million, it's $200,000
or $2 million, that compared to the cost of the software.
Awesome. Well I think we are out of time. We have a couple good questions left,
so feel free to
find any of our panelists and ask those directly afterwards. So I want to give
everyone a huge round
of applause for all of your great insights. And I think we have a short break
between this and our
next session. So again, thank you all so much for sharing all of your great
insights, learnings,
and disinformation about AI in general. But thank you all so much. And thank
you so much everyone.
