Fueling Sales AI With Conversation Intelligence
#1

Fueling Sales AI With Conversation Intelligence

If your AI strategy feels stuck, it’s probably missing the fuel that matters most: conversations. We dig into how conversation intelligence turns buyer-seller dialogue into structured data that LLMs can analyze, answer questions about, and convert into real coaching and revenue impact. Instead of treating recording as the finish line, we map the full system that connects email, calendar, mobile, and in-person meetings, then associates each interaction to the right account and opportunity so i...

Seth Marrs: Hello everyone and
welcome back to the Innovative

Revenue Leader Podcast.

So today I'm going to be talking
about conversation intelligence.

And the reason I'm talking about
this right now is next week

we're going to talk about some
of the predictions for 2026,

which are for in sales, which
are undoubtedly going to have a

ton of AI in it, a ton of
discussions around how you get

the most out of it.

And then following that, we're
going to do a four-week series

around agents and what they mean
to B2B revenue, how well how you

can leverage them, have some
guests on that can talk about

it.

But foundationally, none of this
stuff works if you're not able

to capture a conversation
between a buyer and a seller.

So anybody who's out there
that's that's trying to

implement an AI initiative, and
your tech department is saying

you can't capture and analyze
conversations, you really can't.

You can't do AI really well.

You can't leverage these tools
to their greatest and like

capabilities if you're not
capturing conversation.

That is the fuel that will
generate the majority of the

value.

It doesn't mean that it's all of
the value, but if you really

want to understand what's going
on and really drive improved

effectiveness with your sales
team, you need these types of

tools.

So I just wanted to take a step
back in this series and talk to

what this is.

So a little history for me.

I've been covering conversation
intelligence type tools for six

plus years.

I wrote the first Forester Wave
on it, which was an evaluative

report along with a colleague
named Shell Carlson.

And we wrote it on B2B revenue,
say B2B and B2C sales and

marketing.

And then I've probably done five
or six evaluative reports around

it.

I spent a year writing on this.

It's something that I've I've
held very closely to and find a

lot of value in.

It's also a category/slash
feature that a lot of people

don't understand.

So hopefully in this episode, I
can talk you through some of

that, give you a feel for some
of the things that are going on

in it.

And I think the best place to
start is a definition like what

is conversation intelligence.

So some people just think that's
the tool I use to record my

calls.

That's not if that's all you
think it is, then then this

definition to you is very much
basically a web conferencing

platform.

But conversation intelligence
basically uses AI to turn a

conversation into a form that
allows an LLM to go over the top

of it, ask questions to it, and
drive insights back, but allow

you to improve effectiveness,
understand things that you

couldn't see before.

So the art of selling was
basically the dialogue happening

between the buyer and the seller
that, when captured, can

actually be turned into a
science and used to drive

insights that most importantly
can be redirected back to the

seller to help them improve, not
redirected into your leader or

someone to be able to criticize
or give you insights around the

forecast, although those things
are possible.

The biggest value is really in
the seller getting insights that

didn't that they didn't have
before.

And it allows for you to magnify
or improve your trade like

you've never been able to do
before.

So that's what conversation has
the basic definition.

When I talk about it, that's
that's what I mean by that.

So why is it critical?

If you think about the lifeblood
of a seller, it's buyer-seller

conversations, that's where
deals are won.

If you're in a competitive deal,
the ability to engage and

properly talk to a customer
around positioning, around

influencing, all of that stuff
is where the magic happens in

sales.

These tools are a part of
understanding that.

So that's the only way that you
can do it.

Like you have to analyze those
conversations.

And you want to understand what
that looks like so you can

improve it and create that cycle
back to be able to ensure that

when you're in front of a
customer, you get the most out

of it.

And you can't take advantage,
like I said earlier, of any of

these AI tools that are looking
at conversations if you don't

have a conversation to actually
record.

So conversation intelligence is
that tool.

And whether that's embedded in
your CRM, it's embedded in your

web conferencing platform, or
you're buying a tool

specifically, a note taker or
whatever, you need that to be

able to fuel your AI
initiatives.

So that's why it's important to
talk to it today.

So just foundationally, if you
think about conversation

intelligence vendors, I usually
break them into three different

types because there's different
aspirations.

For some reason, people don't
think of web conferencing

platforms like Teams and Zoom
are a conversation intelligence

tool.

They are, they're lightweight in
terms of their focus, but they

have a lot of the capabilities.

Really, the only difference with
a lot of them is they don't have

a bot, which means they can't
join every call.

You can only do the recording on
your particular call.

Zoom actually does have a bot,
so you could you could use it

for both.

So Zoom kind of is set up as a
way to be a note taker as well.

Note taker is really a bot that
can come on and and really

handle any call.

So and they're more focused on
being a good note taker, so they

do things like summaries really
well and other stuff that you

would have in notes that are
beyond just all the

infrastructure of recording the
call.

So that that's another segment
of it.

And then you have another
segment that's very revenue

focused, and those are the
revenue orchestration platforms.

The vendors in that space are
really focused on how do I

tackle the sales use case.

That means their models are very
focused on sales and revenue,

that the tools that they put
around it, the integrations they

have, the data and how they mix
it together are very much

focused on revenue.

But depending on where you are,
so if you're sitting there

going, I don't have an
orchestration tool like a gong

or momentum or tension, um, all
I have is Zoom, there are things

you can do with that that would
be valuable.

Maybe can't get you to where you
want to go with some of the

bigger stuff, but it no matter
what tool you have, and I would

all but guarantee everybody has
at least one type of tool,

whether you just have a web
conferencing platform or an

orchestration tool, that's what
you need to use as your capture

engine to be able to start
deriving insights.

So next the next thing I want to
talk through is just eight

features that that you really
need to understand when you're

thinking about conversation
intelligence.

And I want to give you a flavor
of like low qual, like like low

feature level versus medium
versus high, and just talk to

that a little bit.

So I'll go through some of these
features so you have a feel when

you're thinking about these
tools, what what you should you

be looking for and what's
important.

First, and this is this is the
part that people don't look at

is how do you associate those
conversations to the right place

so you can analyze them in
context.

So, for example, if I want to
understand what's going on in an

account, my conversations all
need to be associated to that

account.

So if you don't have like a
low-level tool would not have or

would have just basic
association.

I take the domain name and I put
it onto an account, and however

many accounts have that domain,
all my interactions are kind of

copied over there.

That gives you account level.

It makes it really hard to get
to opportunity level because

there's just not technology to
get it there.

And then you get to kind of the
medium lender level vendors who

will either say, Yeah, just let
CRM do that, which usually,

unless you put a lot of work
into your CRM association is

pretty poor.

Um, but that's taking something
as simple as saying, oh, I'm

gonna take a contact, and if you
attach a contact to the if I

take a contact to put it on the
opportunity, I'm just gonna flow

all those in.

So only the ones with those
contacts will be on those

opportunities.

That's great.

The other thing you have to do
in those tools is you have to do

auto association, so you don't
let the seller assign the

contact.

If they're on a web conference,
you associate everybody in that

contact to the opportunity.

Not as it's it's not perfect,
but it's better.

And then the high-level ones
provide a fairly detailed level

of sophistication to help you
unpick the challenges that you

have in a complex organization.

Like an enterprise organization,
like if you do domain level, you

could have eight accounts with
the same, you could have

hundreds of accounts with the
same domain.

Think of like a Walmart and the
number of accounts that you

would have there.

So you need sophistication to
understand, to be able to wait

based on last email came in
versus at most active on the

opportunity on the different
accounts, meaning, you know,

there's been a lot of activity
on this account and none on

these 300.

So I know that, so I'm going to
put the opportunities there.

So I say this, and it's probably
the most boring part that nobody

looks at, but the most critical.

You can get yourself to like 60,
70% if you're just doing basic

association, get yourself to
like 80% if you're doing the

contact flow through, and then
get yourself into 90, 95% if

you're doing like the high-level
architectural stuff where you're

really aligning it.

Um think about that.

That's really important.

There isn't a perfect solution
yet.

What you're going to see over
time is these tools starting to

analyze the call and do
association based on the context

of the call.

That's not too far away.

Um, companies will just need to
build it in, but those

associations go a long way.

Because if you have an
opportunity that has very

targeted conversations that are
on that specific opportunity for

that specific opportunity,
versus a bunch of account level

conversations that may or may
not be relevant, it's very

different than if you just have
like general context.

So it allows you to do more.

It's important if you're not
doing it, you should look into

how I do better association.

What does it look like?

Push your vendors to really
understand that.

Next one is just connecting
conversations.

If you think that this is just
for web conferencing, that is a

very, very small part of it.

You need to have your email and
calendar connected.

I would argue that you need to
have your mobile phone

connected.

If you're doing BYOD, which
almost everybody is, you should

kill that and you should go to a
soft phone where you have the

number assigned by the company
and your sellers need to call

off that number, and there's an
auto record that happens off of

that, or at the very least, an
auto tracking where that call is

updated so you know that a call
happened.

All possible, very few companies
do it.

There are lots of phone
conversations that happened

between a buyer and a seller.

So it despite what people will
say about it going away, if

you're deep in a deal and you
have a good relationship, you're

likely going to pick up the
phone and call that person to

help move that deal forward.

And the other thing is the
in-person side.

It's getting more as sellers get
more comfortable with this

technology and start to realize
how they can get value from it,

they're asking for more.

Let me record a phone call in
person where I put my mobile

phone down.

Tools have that where you can do
it.

You have like a hacked version
that you could do as well, where

you can just open up your Zoom
and play that one and record it.

But ideally, you'd have
something that would be formally

set up to do it that way because
acoustics and all that stuff

would be a little bit different.

You'd want it to be built for
that purpose, but it it can work

and it and it does work.

And most of these technologies,
they have varying levels of

sophistication.

They're listening for voice to
do voice separation.

So even if there's five people
in a room, some of them can get

actually really good at
identifying those voices and at

properly associating them.

But even if you can't do that,
you'd at least have the

conversation assigned to that
account, that opportunity.

So that allows me to connect
more conversations together.

So I've talked before about how
do you associate them.

Connecting conversations lets
you have broader conversations.

If you are just analyzing off of
web conferencing, it's like

10-15% of the conversations that
are happening.

You've got to get the email and
you need to have calendaring in

so you can see what's going on
when a meeting happened and

there wasn't a recording.

So at least you know a meeting
happened.

Like those types of things.

So you've got to focus on
broadening your conversations.

So those are kind of two
infrastructure things.

Here's some basic stuff.

Like, summaries are obviously
really prevalent now.

Like that's a great tool Jen AI
can do.

It could take a summary from
anything.

Think of these in three
different ways.

So call level, dead easy.

It's one call.

You can get a summary on that.

There's some pretty easy, is
it's it's it's not aggregating

from multiple places or multiple
conversations.

So you'll see that on every
tool.

Every tool has it from web
conferencing on up.

Note takers will do a little bit
more because they understand the

importance of a customized
summary.

So they'll they spend a lot more
effort to build capabilities in

like a FAPAM that has they can
build all these templates

that'll look at it and talk
about it in different ways.

That's that's the next level
type stuff.

And then the other summaries are
when you get to object level, so

account level, opportunity
level, contact level, where you

can aggregate all this stuff
together and get really cool

summaries around what's going on
in the account, what's going on

in the in the opportunity,
what's going on with this

contact.

That sets the foundation for
some game-changing capabilities

around standard things that
sellers have had to do forever,

like account plans.

If I can do an account summary
based on just a request that

aggregates all the
conversations, I don't need I

don't need to do all that stuff
myself.

I can let the tools do it.

Another one is scorecards.

So scorecards have got have
grown in sophistication over the

last, especially in B2B.

B2C, they've been super
sophisticated for a long time,

but now they're getting a lot
more sophisticated.

So base level, like the
low-level ones, will give you

uh, you know, you go tick the
box and the seller listens to

the call and they type it in.

And even that's that's getting
less and less.

I think people have realized
sales leakers won't do that or

won't do that very well.

So even the ones that are doing
that now or providing some sort

of AI recommendation, but that's
basic.

And and another thing you have
to look at is what types of

scoring you could do.

One through ten, one through
five, yes, no, text, any of that

stuff.

But that's base level.

Like base level, you do the
scoring.

Medium level is you'll see
companies that'll do a

pre-configure.

Medic's a classic one.

Everybody can score medic
because it's just very

straightforward to score.

So building a model that can
train to look at it.

Be careful when you have someone
doing medic level deal scoring

versus macro deal scoring.

But the that custom template
that is built for you, the

reason why that's more valuable
is the companies will train it.

They train it, train it, train
it.

So when they give it to you, it
the logic has been proven out

rather than you kind of building
a summary truck, a summary

template on your own, and kind
of it doesn't have the same

training.

So that's the next level.

The highest level one is just
the automated scoring.

You build it, you train it using
prompting or design, any sort of

different things, and it
automatically scores for you.

That is the market where it's
going.

It's getting a lot, lot better.

You can do a ton of stuff.

It's really on you at the level
that you can really train this

to get accurate scores.

So now you're stopping having to
do full call reviews.

You just listen to it and then
understand what the score is,

and you coach off the score, and
you coach off the really good

tools, will take you right to
the snippet and give you

examples that you could look at.

So you're coaching based on
those things.

Another thing is really
straightforward asking questions

to conversations.

So there's varying levels, but
you there's a lot of times you

have a question you want to ask,
you call a person.

I would recommend if you have
these tools, like a gong's

really popular for this, like
just ask questions to it.

Every tool will allow you to ask
a question to a call.

So from web conferencing,
note-kaking up, you can ask a

question to a call and get an
answer.

The next level would be I can
ask a question to an object

where I go to the account and I
ask a question to an account,

ask a question to an
opportunity, ask a question to a

to a contact, and you get an
answer back based on all the

aggregated conversations that
have happened over a period of

time.

The really sophisticated ones
are doing allowing you to ask a

question to your entire data
set.

So I'm asking a question, a
general question, and it

searches the system searches
through, pulls back a response

based on all the objects, all
the data points, that type of

stuff, or a certain set of those
to get a very sophisticated

answer that's looking across
multiple channels, tools, data

sets, all that.

Those are the different levels
there.

Another thing that's really,
really valuable is triggers.

So your ability to set something
up so that I can identify when

it happens on a call.

So you have like the low-level
one would be keywords and

phrases.

That's been around forever.

If a word comes up, flag it.

If a phrase comes up, flag it.

So that's that's fairly
remedial.

But the many of the like web
conferencing, those types of

tools won't have this.

But having just base, this that
works great for like competitor

mentions and those types of
things.

Low level, but that's that
that's what that trigger would

look like.

Medium is the same thing that I
talked about before with with

scorecards.

You do pre-trained triggers.

These companies have figured out
the ones that are most

prevalent.

They build them for you, they're
really good.

You just put them in place and
maybe tune them to your

business.

And then the high-level ones are
doing prompt created triggers,

like where I can go in, build it
out, and ask for exactly what I

want.

So I can get surgical around my
objection for my company and I

want to see it.

And then when it when it sees
it, it flags it and it can

alert, it keeps track of it.

So you can do trending, all of
those types of things.

So really sophisticated, really
valuable.

The next thing is finding
conversations.

This is fairly remedial.

Everyone's gonna have folders
and shares and snippets that you

could do all the way through web
conferencing.

Then you have the medium level
where you take your keywords and

you put them in and you say
these are the things I want you

to look for, and it just kind of
tracks those things so you can

easily find them, similar to
what you see in others, where

you have like a folder and then
you have a keyword, so it allows

you, like you see this in in
like note apps, like ever, like

Evernote or or in any of those,
or OneNote.

And then they're the high-level
ones are the ones that do AI

generate.

They're looking for trends in,
and they start surfacing those

trends up when they see them.

So you'll get really interesting
like triggers or tags that you

may have not have thought of,
but it's coming up over and over

and allows you to do things
around insights that you would

have never done because you
didn't know that was happening,

you didn't know to ask it was
happening.

And then the next thing is
around reporting.

When you think about reporting,
this is this is critically

important for you to understand
what's going on.

The low-level version will do
behavioral stuff.

Everyone does behavioral, your
talk time, your monologues, your

questions, that type of stuff.

Straightforward, every tool will
do that.

The the next level is topic
level.

So when you get to the topics
where it's identifying those

topics and then allowing you to
understand how often they're

happening and reporting them at
the individual person, and so

you can track them.

And then the more the most
sophisticated ones are doing

trending, where I can take all
those triggers and things that

are happening and show you how
they're trending over time.

So as you're training,
maneuvering things around,

making adjustments, you can
watch those numbers go up and

down.

So, really critical to a leader
that's trying to understand the

impact of the things they're
doing.

So you want to have the
trending, you want to have that

visibility.

So those are eight features that
are really critical.

So if you're thinking about like
how do I maximize my ability to

capture conversations and do it
the right way, take those eight

things and evaluate around it
what your current situation has.

So the last thing I want to do
here is just talk about four

things that are going to be
disrupted based on what I just

talked about here.

They're already being disrupted
in a fairly large way.

And you should think about it
when you're looking at this

technology, or you're just
looking about how do I make a

difference with AI in your
company.

First one, the biggest one, is
auto-update of fields.

If you're not currently today
looking for a way to have your

sellers stop inputting data into
CRM, that is a mistake because

the technology is getting to the
point where you could do some

some really, really cool stuff.

Companies like win.ai have been
have been working on this for

years and are getting pretty
good at it.

Gonk just released something or
with data extractor around this

where you can just have the
conversation analyze and apply

it to a field.

At first, this has been around
forever where you can just take

text and then throw the text
into a field in CRM, but now

it's getting sophisticated
enough that you could identify

it and use a list of values.

You can identify it and say this
means yes or no, and then you

can update everything.

So if you're looking for the a
gigantic way to get efficiency

and then a better understanding
of deals, you should be in you

should be investigating this
heavily and working like crazy

to get your sellers out of data
entry.

Just think of things like stage
progression.

Anyone like stage progression
has been garbage for years, it

doesn't work, sellers don't
update it.

This would allow you to actually
use stage progression because

it'd be happened automatically
based on the conversation.

And it'll also allow you to do
stage regression where when you

see things not happen, you could
draw it back.

So it'd make that staging really
valuable and also really

valuable for partnership with
sales and marketing, because

that's a important staging is
really important in that

relationship.

So they this is probably the
most like early stage of the

ones I'll talk about, but
absolutely critical, and you

should absolutely be
experimenting with it.

Next one is deal visibility.

This one's been around for a
long time, but your ability to

understand the conversations and
have it map it to your

qualification details to how how
you look at the questioning to

understand and say, Oh, what's
my like you can use medic, you

can use anything like that,
where you're trying to

understand if you've qualified
the customer.

That's been around for a while.

It's fairly, it's probably the
it's probably the most advanced

in terms of people using it, but
you should be using this to

determine your whether your
sales team has done the

qualification, has done the work
that they they needed to do and

have it as part of your
forecasting.

That's probably the nuance for
me is a lot of people weren't

using it for forecasting and now
a lot of a lot are starting to

realize wait, I can get a lot of
value from pulling conversations

into under better understand
where my deals are and which

ones they're going to close.

The third one's account planning
if you're still having your

seller like spend a week and a
half updating a spreadsheet or a

PowerPoint or something to be
able to create their account

plans you are wasting a ton of
time and your account plans

aren't sustainable.

In the past you'd have like big
organizations with big high

level accounts that are that you
would do that for sellers would

do it, take forever to do it,
pitch it, and then never touch

it again because they can't they
don't have the time to with

conversation intelligence and
these tools you can just set the

structure up and allow that to
be on demand.

So you can do it for every
account you put click a button

and it pulls up your account
plan then the seller just

provides their input and then
they get inputs back.

Because if you set the account
plan up it should be insightful

for the seller too.

That's changing the entire game
on how account clients and to me

it it should spark a renaissance
in use.

It may not spark a renaissance
in vendors but it'll be a key

feature in any tool that is
working to uh support an

enterprise organization because
any big enterprise will have

these types of account plans or
at least aspire to have them and

out and have them for for real.

And then the the the final one
is this pre-call prep.

If you don't use this for
pre-call prep then you're

wasting a lot of your seller's
time.

Actually you're probably not
wasting your seller's time your

seller's probably not doing it.

And then if you're going up
against a seller where you have

a button that can click and
it'll go pre-call prep for you

and bring everything in in the
conversation and say what's

going on think about if you're a
seller that has that going up

against a seller who doesn't
who's just kind of trying to

figure it all out and doesn't
know all this stuff and can't go

in and get it you're gonna
you're gonna struggle to

represent yourself at the same
level as those who have a tool

like this that that would allow
you to do it.

So those are the four areas to
to just to recap from the reason

why I'm talking conversation
intelligence now and these

important features is kind of
set us up for what's coming in

the prediction side and what's
coming on the agent side because

you're gonna see a lot of
different stuff and things that

are going to be really cool and
where things are going.

But foundationally the easiest
thing to do right now is and the

must do if you want any of this
stuff to actually be real is you

have to you have to find a way
to get your conversations

captured and to have a plan to
understand how to analyze use

the capabilities that already
exist today.

I hope this was was useful and
we'll talk to you next week