February 6, 2020 • 35 mins
Data is the new oil. Everyone knows it, but what can an enterprise do to get more out of their data? In this in-depth conversation with Mark Skiles, of Integrity Works International, we discuss how large companies struggle with silos, break down communication barriers, and prepare their enterprise to capitalize on the emergence of data. Kyle Hamer, CEO and Growth Marketer from Hamer Marketing Group, dives deep to uncover how companies large and small can leverage a fractional CDO, and a common data catalogue to create something remarkable.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Kyle Hamer (00:05):
Oh and welcome to the summit, the podcast where we
bring you knowledge and insightsfrom industry leaders and
professionals.
My name is Kyle Haimer.
I'm your host of the podcast andowner of Haimer marketing group.
Today I'm here with my goodfriend Mark Skiles.
Mark Howard today.
Dan fine.
Thanks Kyle.
That's awesome.
Mark.
Uh, Mark and I are going to betalking about the new role of a
(00:27):
chief data officer and what'shappening in the world of
[inaudible] data
Mark Skiles (00:31):
and operations and how businesses and enterprises
can, can expand or maybeleverage data in more
intelligent ways.
But before we get into that,Mark, why don't you tell us a
little bit about yourself?
I'm sure you have kind of a,have a, a bit of a varied, an a,
a, uh, background here.
(00:51):
So started out in networking,um, worked on that for a few
years, became kind of anindependent consultant back in
the early nineties, and, uh, dida variety of projects as a part
of that.
But, uh, over time it's kind offell into the whole data world,
uh, primarily managing SQLservers, right?
You kind of a DBA, uh, and soyou work with the data a lot and
(01:12):
then you start thinking ofsomething more than just moving
these bits and bytes around.
What do they mean to people?
Right?
And so you start thinking aboutthat, uh, kinda migrated towards
the, from a technical side tomore of a business in tech,
business intelligence sphere,right?
How do you use this stuff, makeit work and it get some good
value out of it.
And so you had the mantra for awhile from data to decisions,
integrity works, uh, any of ourcompany integrity works.
(01:35):
And, uh, you know, this wholeidea of finding wherever the
data that you're looking for isat.
So we started throwing aroundthis banding about this, find
it, get it and present it rightin the BI.
In the BI world, reality is,it's not just about the
reporting, but a lot ofdifferent things around data
quality.
Uh, you know, being able to getwhat you want when you need it,
whether you're a data analyst,data scientist, kind of person
(01:57):
more on the business side or youknow, a different set of
challenges from the it side.
And that's, that's kind of wherethe CDO really came into play,
right?
That the it folks are reallyfocused on getting you the data.
Uh, they can locate thedifferent systems, but what's,
what's the most value to anorganization and what are the
business drivers for that data?
You know, how do we want tofocus its efforts?
(02:18):
And so that's really kind ofwhere the CTO came from and it's
an evolving role.
Uh, right.
Organizations have these kindsof challenges when they're
trying to push through to, to,um, this whole, um, data-driven,
you know, world that everybodythrows around.
Um, and there's just a lot ofpieces.
Moving parts to that point froma CDO is just drive from the
(02:38):
business side more.
I have a C at the C level table,uh, as opposed to being just
focused on information and kindof reporting kind of thing.
It's, it's really help us kindof navigate this whole thing and
you could get dry the businessvalue out of it.
So when you say, when you seedrive drive the business value,
one of the, one of the thingswhen you were explaining this to
me, the first time we chattedabout it was the, um, the silos
(02:59):
that information
Speaker 3 (03:00):
and data lives in.
Can you expand a little bit moreabout what you're seeing happen
in the market for, uh, reasonsthat people will need data or to
consume data information and howit's impacting enterprises that
have normally functioned ininformation silos?
Mark Skiles (03:20):
Well, one of the challenges, especially to
business users, right, is, isthey have a question.
They have to go to it to get it.
And so they can go through thewhole, you know, I get it on the
project docket and kind of workthrough the logistics.
They are, take some time, takesome money.
Um, they don't really have anyway to get their, their hands in
the pie and really ask themquestions.
So part of this is kind of beenpushed where, where the business
(03:40):
says, you know, I need it when Ineed it and I want to have to
wait six months for it.
Right?
So the silos really are thisidea that we've got a lot of
systems in our environment,enterprises, large enterprises,
but small, small companies hadthe same issue.
It really just different scale.
And that is, you know, myQuickBooks for example, is as
locked my financial data upsomewhere and it doesn't really
talk very well to my, uh, Idon't know invoicing or my, my,
(04:03):
uh, you know, my customermanagement or, or whatever it
is, right?
The, the, the, the, we've got alot of walls but not a lot of
bridges.
And so really this is around howdo you build bridges with these
different systems?
Not so much from a technicalstandpoint, but from an
organizational understanding andensuring that the things that
are really important to us, uh,get, get kind of elevated focus,
(04:23):
whether you're, you know, inpublishing.
And so it's the, the authors andthe author data there, the book
contents, whatever it is, right,that the right things get the
right focus as opposed to itsometimes being driven by the
current project on the list.
And uh, anyway, it just istrying to drive the value from
the folks who are using itright.
(04:44):
And can really say this is whatI value the most as opposed to
kind of its current prioritiesat the time.
That makes sense.
Speaker 3 (04:51):
It absolutely makes sense.
I mean, it helps betterunderstand why this role is, is
becoming so important because Ithink 20 years ago information
was there, but data wasn'tintegrated.
People weren't wanting toautomate and bring things
together, even ask largerquestions.
It just, you'd send one personout to go comb through the
(05:13):
catacombs of each, you know,each set of filing cabinets and
they came back and they were theOracle of truth.
Whatever their personal opinionwas, was the interpretation of
that information and theassessability of, of software
and technologies really forcingus to look at how we deal with
information, uh, differently.
(05:33):
But, but what, what's not fullyclear to me is, is why
organizations are saying, okay,well we need a data officer
versus making this part of likea chief information officer role
or a, um, security officer.
Like what, what is differentabout a CDO than say like a CIO?
Mark Skiles (05:55):
That's a good question.
So the, for example is a chiefinformation officer, right?
And so a lot of times we'rereally involved in collecting
the data, right?
But the focus really isn't onthe delivery and the benefit of
that data.
Um, and so if just collectingit, put in a data warehouse or
put it somewhere, doesn't giveme everything that I need.
(06:16):
Um, how do you keep your limitedresources focused on what are
doing, what really gives you themost benefit?
And so the CDO really kind ofgives you that clarity and
direction for your attention.
And it's also focused oncollecting smaller data sets as
well and govern them andmaintain them in a, in kind of a
, uh, you know, more focusedmanner.
Speaker 3 (06:32):
Well, that, that's interesting.
So a lot of organizations havelax at best.
I mean, they may have datagovernance, but data hygiene is
something totally different.
Um, but how our, how ourorganization's handling the
change or, or thinking about thechange of, of data governance
(06:52):
from silos to, okay, I have achief data officer now, or we
want to, we want to provideassessability of data across
different silos.
How's that, how that forcingthem to think differently.
Mark Skiles (07:07):
Well, one thing is really on the data quality,
right?
Because in the past you kind oflooked at just, just we have
volumes of data, we've collectedall this stuff, but you know, we
might have different reports atdifferent totals, different, you
know, the, there's conflictseemed, uh, facts if you will.
Right?
And so somebody really can offocused on, um, putting some
oversight and saying, these arethe things that organization we
(07:29):
really value the most.
And this also bleeds into thesecurity conversation, right?
Organizations really need toreally to say, um, for the, for
the most important things in mycompany, I need to have a higher
level security for, I don't needthe same level security for
every piece of data, myorganization.
But I do need to understandwhat's critical for my business.
And if something went down, forexample, um, you know, what's
(07:51):
our, um, not backup plan, butyou know, our, our, our stake in
the ground and says we can'tlive without this.
And if we do, it's going to costthem a heck of a lot money to
get it, get it back.
Right.
And so you just put some energyaround that.
I'm not talking about technicalstandpoint, just saying
identifying it first.
Um, one of the exercises if Ibleed over to, uh, to the later
conversation is that recognizethat we had to understand what
(08:15):
are the different things in, inan organization.
I'll use the example of the uh,biotech.
So you know, a seed for example,but what goes into this seed?
Well, the germplasm is thegenetic makeup.
Well, for example, thatgermplasm is, is a, is a part of
the, the, the DNA conversationand it means different things to
different people.
(08:35):
So for the laboratory, thisgermplasm can, can be a piece of
glue in a, in a sample too,right?
But for the people in the fieldwho are planting those things,
this germplasm comes in theshape of a seed.
But the conversation around whatare the most important things in
our organizational world, whattypes of information are we do
we need?
And they wanted therelationships, those pieces of
(08:56):
information.
So the exercise was kind of thiscontext map around while you use
the term different nouns in ourworld and then you know, how
they interact with each other.
And so that was why one of thethings we learned early on is
they really need to kind of takea step back and, and, and find
out hoarder all these things.
Right.
What's the wet, where is it at?
Was kind of a secondaryquestion.
(09:17):
But this context, you know, ithas different meaning giving
different context and so thatyou kind of had this business
glossary need.
I need understand what it is,what it means and uh, and what's
the context for it.
And so we'll talk about thislater with the data catalog ID
is really kind of a way to, to,to manage that, uh,
(09:37):
understanding of what the salesthings are from a conceptual
standpoint all the way to wherethey're physically located, what
tables are they're at, where thedifferent columns with they're
at in a way to kind of put thosethings in a library form.
Um, it's kind of the geospatialapproach, right?
This whole, I want to find outwhere it's at to my data so
Speaker 3 (09:57):
that, um, you know, it's, when I find what I find
fascinating is you touched onseveral different things there.
One is, uh, data governance.
You talked about, you know,needing the ability to move
faster and different things meandifferent things to different
people.
Speaker 4 (10:16):
No.
Speaker 3 (10:18):
When you have groups that have, you know, their own
sort of internal fighting andchallenges even in their own
departments or silos, what are,what are some of the common
challenges that we seeorganizations having as they, as
they open up?
I mean, you touched on itbriefly as they open up the, uh,
the dialogue into, Hey, I needto have access to research and
(10:41):
development information or Ineed to have access to marketing
or accounting or whatever theadditional, uh, data set is what
are, what are some of the corechallenges that you see as they
begin going down that road as anenterprise or even business
ghost in that road.
Mark Skiles (10:58):
And one of the challenges we ran into and the
reason why we kind of startedwith what's the, what's the,
I'll use the term sharedlanguage.
They use that term in some kindof this domain driven design
world.
If you'd look that up, you'd seewhat I meant by that.
But, but shared language kindaunderstands this.
Every, uh, every application ordevelopment team or something
has, has language that they use.
And so you try to understandwhat the different things that
(11:21):
they are they talking about the,the objective here that we found
or the challenge I guess that wefound was that people are using
the same word.
I'll use the word inventorywithout any context.
And so the first thing we do istry to understand a, what it is
we're talking about.
And then B, what are thedifferent contexts that we're
using and try to shape those insome kind of a, of a, of a
(11:42):
language.
And we, we took that one stepfarther where we actually
modeled them conceptually.
For example, the, uh, the ideaof germplasm, the definition
from a technical, um, betweenthe field and the laboratory was
the same for 80% of theconversation, but the other 20%
was unique to the laboratory orunique to the field, right?
(12:05):
But, but that gave it context.
So, okay, when we talk aboutinventory, here are the things
that we agree on that we're all,you know, we're all speaking the
same language.
Here's the other parts that werecognize are different.
And here's what we mean by that,right?
And so we documented those kindsof things.
We, we used a conceptual modelto do it.
Now there's a variety ofdifferent tools you could do
that.
But that was kind of ourapproach.
(12:26):
Basically, you just need tounderstand what, what you have
the nouns.
We'll get back to using thatterm and then how those things
interact with each other in, inyour organization.
Right.
We're not talking technical atthis point in time.
We're not talking about howthey're implemented in your
database, but just a commonunderstanding of things that
make sense.
Speaker 3 (12:41):
Makes perfect sense.
And I think the thing that'sinteresting, what I think about,
um, writing down what we have,you know, I would imagine that
inside your catalog there has tobe some level of description
and, and for a period of time,silo a or silo B will come in
and they'll look at it, uh, aterm and there'll be thinking as
(13:03):
if they're in their own silo.
But it's really referencing morehow the other silo, um, plays
with it.
So there's the, the, the uh, thedescription component almost
becomes more valuable than eventhe, what the description of
what it is.
You know, it's, its definitionis probably more valuable in
creating common language.
(13:24):
Then the descriptor, the, the,the name itself,
Mark Skiles (13:28):
one of the tools we used is a context map.
And that was really just a bunchof boxes and squares.
They kind of took all the nouns,right?
And they have this idea of someupstream and downstream stuff,
you know, so an experiment forexample, it takes a lot of
different pieces.
A location, right?
The different timing, thedifferent chemical components.
(13:51):
And those are all kind of suckedinto this idea of what is an
experiment?
Well, if you looked at it ontheir little context map, you'd
see a box for experiment, you'dsee a box for germplasm, you'd
see a box for um, you know,chemicals, those kinds of
things.
And then you might see comingout of the experiment box some
other things that are kind of,you know, outputs from that,
things that are produced.
So you have this upstream,downstream kind of thing.
(14:12):
So one of it is the Hoyts andthen like I said, they kind of
interact with each other and wekind of put that into what we
call the context map.
It's not a unit, you know, in,in the manufacturing they have
like a, a entity relationshipkind of things.
Um, but it's basically someoneto visualize that and then begin
to, to speak to those things andpeel them apart.
Right.
And then you start buildingthese models out from there.
(14:34):
Are these definitions to useyour term around what these
things are, right.
Give them some kind of a shape.
Speaker 3 (14:41):
Well, it's, I mean, it sounds like, to me the, I
mean, I know this because we'vechatted about it, but it sounds
to me like this, this, um, thisrule of a CDO or[inaudible]
fractional CDO, uh, you, youplayed this role in a project
where it really wasorchestrating many of these
things coming together.
Can you tell me a little bitabout some of the core
(15:02):
challenges or maybe corebusiness questions that the
business is trying to ask thatthat kind of led to the
evolution of this for, for thatspecific project?
Mark Skiles (15:13):
Well, first of all, her most recent project really
wasn't a, it didn't start out askind of a CTO role.
It kind of evolved in a kind ofa semi CDO.
And it was, it was based in the,the R and D department of a
biotech company.
And so they really were goingaround this data integration
path and using API.
So they brought me in to kind ofsay, um, you know, help us
gather, uh, from a designstandpoint, the, the, the, the
(15:35):
things that we need to bebuilding these core API as they
call them, right?
At DNA integration layer.
What evolved from that initialconversation?
It was focused on building APIs.
We realized they really neededto be some common language
around um, the, the, if we'regoing to do something core and
I'll use core just to meanshared right?
Integration between the commondefinitions of something, then
(15:57):
then w we have to understandwhat it is we're talking about
and everybody had a differentdefinition of stuff.
So, so we kinda went back to thedrawing boards.
Okay, now we need to focus onthe data governance side.
We're going to inequality atdata quality out of this
equation.
We're going to have to towpeople.
They needed to be a commonunderstanding of what we're
talking about.
And so that we kind of backed upthe bus a little bit and started
(16:17):
going down this path of thiscontext mapping exercise and
some some conceptual modelingthat became very valuable cause
they, what happened, we wereable to find a tooling that took
those conceptual models and cameup automatically.
We call the auto-magically,right?
But we could automate somethings.
It pushed him up to a websiteand this, these, a web
publishing of these conceptualmodels became something that we
(16:38):
would walk into a room with anapplication development team and
it gave us something to speak towrite something visually.
I could look at that.
It became a conversation piece.
It didn't have to be exact, butit was enough for everybody to
get an understanding of what theshape was.
And you could even involve it inthe conversation.
Oh well we're missing a piecethere.
Okay.
We'll add it to the laboratoriespiece of that equation because
that's what we're talking about,right?
Throw it right in there is we'reusing the design tool and
(17:00):
display it on the screen.
Everybody's happy.
Go back and publish it again.
Right.
So here's an evolutionary kindof process.
But it was a conversation.
And really if I saw one valueout of this whole thing, it was
forcing the conversations thatpeople hadn't taken the time to
do in the past.
And just a common awareness ofsome of the challenges we were
facing and trying to figure outa way to, to, to address them.
So, um, I, I really see myselfas kind of an early, uh, chief
(17:25):
data officer.
We didn't have a lot of thingsfigured out.
We were kind of walking throughthe process and, and, and, you
know, they love the things we dodifferently, uh, doing it again,
but we learned a lot.
And so that now I had placedmyself in a, in a, you know, a
phase to CDO.
So
Speaker 3 (17:43):
that's a, that's really cool.
So like when, when, when I sayit's really cool, I, I'm
thinking about, you're talkingabout the, the common language
and the sorting things out andyou know, creating
visualizations.
It's like, it was also mostlyyou were building, uh, the
building the role as you werebuilding visibility for the
company.
And I can't imagine what kind ofimpact that was having for teams
(18:06):
around.
You talked a little bit aboutthe impact of getting access to
information or creating commonlanguage with your, with your
catalog.
What, what that meant or whatkind of the aha, like what were
some of the big revelations forthis company?
Mark Skiles (18:22):
Well, first of all, just to clarify, this company
doesn't actually have a datacatalog as of yet.
They recognize a need for it andthey asked me to do a lot of
research on it and so I cameback with some, some
recommendations which haven'tbeen implemented yet.
And that's more of a timingthing, uh, as opposed to
anything else.
But, uh, in terms of elevatingin the, in the value, um, what
happened was as, as uppermanagement began to see these
(18:45):
things and people were usingthem in kind of a practical way,
the data governance came, all ofa sudden got escalated to, you
know, tending a lot of meetings.
People wanted to know, you know,not just what you're working on,
but how does this affect, causethey were going through a merger
and acquisition and that's oneof the times this is really,
really important, right?
You're bringing in a new companyand you need to know that.
Do you mean the same thing whenyou're talking inventory?
(19:08):
Right?
Uh, and so that those really gotelevated into the upper
management levels and because itwas conceptual, they can kind of
grip on, get a grip on thosethings as opposed to, you know,
talking about the bits and bytesor the table definitions or that
kind of a thing.
Right?
They could begin to understandhow these concepts fit together.
And so it became a very criticaland highly valued piece of the
(19:30):
upper management's, um,evolution too.
Because they could get a grip,they could, they could chew on
those conceptual things and geta picture of them, um, as
opposed to showing them tabletable definitions.
Right?
We constantly harping on thatthis is not a table definition.
It does not map to all thecolumns in a table.
Right?
So just to understand this as a,this is a business conceptual
(19:52):
model, not a data, not a, not adatabase table structure.
Speaker 3 (20:01):
I can, I'm, I'm chuckling cause it's like, I
can't imagine, depending on thetype of executive, they were
probably some or even seniorleadership that you would be
talking to if there were somethat were really, really
relieved that you weren'ttalking about database structure
and there are others that werereally annoyed that you were
talking about concepts thatweren't tied into tables.
Mark Skiles (20:19):
Well usually that came in like your data modelers
or your, your DBHs or your, youknow, even your data warehouse
guys.
A lot of times they're sofocused on what the table
structure was or something.
They had a hard time pullingthemselves out of that.
Uh, so, you know, theconversation was great, but
everybody had to do a littlelearning.
I'm okay.
When we walk in the room withMark Skiles with David Bowen,
we're not talking about, youknow, table structures and uh,
(20:43):
that was a learning process foreverybody.
Speaker 3 (20:45):
As you, as you kind of go through this process or
this project, it sounds like alot of the, the silos had to be
torn down.
Did you take a, a warehouse anda cubing approach or a, you
know, is this a dealing withdata lakes?
What does this, what does thisreally mean for how data is not
necessarily interpreted, but howdata is accessed and
Speaker 5 (21:09):
Mmm.
Speaker 3 (21:10):
Cattle catalogs probably you already got the
word most the how, how it'saccessed, how it's, uh,
Mark Skiles (21:17):
well, one of, one of the things if this, if this,
I'll use a higher level model atthe moment just for that
conceptual idea is thatdifferent systems kind of, um,
and, and there's differentlevels of abstraction, you know,
and database world, you havethis conceptual, logical,
physical kind of kind of amoment within data models.
Right?
Um, and so how would you take adata from system a using this
(21:40):
conceptual model and somehowbring it into the data warehouse
or system B?
Right.
So they created something calleda, uh, I think we call it a data
pump, remember what it was, butit basically used, took this
conception model, took fromsystem a transactionally, turned
it into this XML file kind ofthing.
And think could be re-read onthe back end and sucked off the,
the, the data pump, right.
(22:00):
In a way that, that uh, theycould kind of rearrange it on
the other end of the way theywanted to, but they had this
common definition of things andsaid, okay, this is the way it's
, uh, uh, we can read itdynamically, right.
Cause it was XML based or, or uh, uh, it could be Jason Bass,
however you defined it.
But the data pump I think was aXML and a CSS stuff.
(22:21):
Um, so anyway, I don't know ifthat answered your question as
far as how they were doing it,but it's, it, it wasn't so much
a, um, you have this element ofdata mapping.
But the other thing was justautomation around taking these
conceptual models and, and usingthem as a common, like a, uh,
almost a schema kind of thing.
And those of you in your dataworld would kind of understand
the scheme of conversation.
(22:42):
That's really the shape of anygiven, given data coming in.
And they use it in the Jasonworld a lot where some things
are coming in off the internetor you know, some sensors
sending me data in a certainshape, right?
They would call it that dataschema, right?
So, um, the different ways tokind of move from this, this
conceptual thing to somethingthat's, that has more shape,
(23:03):
right?
And so we had tried to do someautomation things to leverage
those models and then you couldplug them in and they can
automate, uh, the moment of frompoint a to point B, given the
use of those, those commondefinitions, that's probably a
little too nebulous for you.
I'm not sure how much detailyou're asking for.
I'm not trying to avoid yourquestion, but I just hesitated,
(23:24):
jump too much into technicaldetails.
So
Speaker 3 (23:25):
yeah, I'm glad.
I'm glad you didn't go anydeeper.
Why I might've like come out theother side, Mark.
But that being said from a, it,it's fascinating to me because I
hear about, you know, and I haveenough, I have enough knowledge
to be dangerous, you know,usually like the, to, to um, to
have the, to understand what'sbeing talked about.
(23:47):
But not everybody fullyappreciates, you know, how
nebulous and fine tuned detailedlike how much information and
structure is buying theinformation that they need.
Mark Skiles (23:56):
Yeah.
Speaker 3 (23:56):
[inaudible] do you think, do you think that like do
you think in this, like theproject that you set up for, for
this company and, and the roleof the evolution of this data
officer and how companies arethinking about it, do you think
that there'll be a,
Mark Skiles (24:13):
the[inaudible]
Speaker 3 (24:13):
point in which information will be free flowing
and we won't be dealing with,not necessarily governance per
se, cause you want to protectyour IP but deal like, will we
be in a spot where marketing canaccess what research and
development is doing and isthree years down the road and
seeing how that may beinfluenced by what they're
seeing happen in the markettoday?
Or is it well we get informationflowing that direction or is it
(24:38):
gonna really stay continued insilos?
It's just, you know, maybebusiness adjacent, there's value
in doing this.
Mark Skiles (24:45):
Well that's a good question.
I think it's going to have to[inaudible].
And what I mean by that is, youknow, you had this, this idea of
um, Oh, what's the term theyuse, can have a um, uh, a person
who could do their own analysis.
Right?
The, the, the um, let me justforgot the term off the top of
my head, but the point isaccessibility is going to be a
(25:06):
big deal because I want to askquestions in, in more real time
as opposed to being hung upwaiting on it.
So to do that you can have togive, uh, the marketing
department or the accountingdepartment or something better
access to their data causereally it's theirs, right?
It's just been kind of locked upin systems.
So the answer to your questionis yes, it's gonna evolve that
way.
The timeframe will be a littlequestion Mark.
(25:27):
I think companies are going tohave to invest in that.
And it does take a little timeand money, right?
This, this catalog ideas.
But one of the solutions to thatis not the only approach, but
this geospatial guide kind ofserves as the library if you
will, from the data definitions.
Oftentimes like the dataglossary, right?
You add the context into theequation so people could look
up, you know, where's, where'sgermplasm at?
(25:47):
Well first of all, what's itmean?
Okay.
Given the context of lab, that'swhat I'm talking about and we
look specifically at labscontext, but where's that labs
are unpleasant located, whatkind of systems is that?
Well, the data catalog kind ofgive you access to that and
especially in a machine learningcatalog was kind of in the next
evolution.
A Gartner calls is the newblack.
Those, those data catalogs willactually um, scrolls, want more
(26:11):
right there.
They'll, they'll, they'll trollthrough your environment, all
yours, your, your different SQLnon sequel, you know, different
kinds of data sources and, andkind of pull those out and say,
well these are what's out there.
You tell me what's good or not.
And somebody that the dataowners, data stewards cause some
crowdsourcing.
We would go out and say, okay,this kid is, has an 80% data
quality.
(26:32):
If you really want stuff, lookat this column.
Right?
And they can kind of pull thosepieces together and get to see
even in a big data environment,these are the, the the, the
piece of information in this bigpool of information that you
want to on, right?
We've been using this for ourtesting or we've been using this
set of production environment.
Don't waste a lot of time andother stuff.
This is it.
As the data catalog gives you aplace to put all that stuff you
(26:54):
can, you can interact with otherpeople who are using it and it
becomes, you know, over time itjust evolves.
It's a tooling, nothing morethan that, but it gives you a
way to kind of put all thesepieces together in a way that
people could find what they'relooking for.
And then many of them now willgive you a query mechanism or
you can do some discoveryqueries on the data.
Is that the right data?
Am I looking for that?
Does they have the pieces ofinformation that I need and uh,
(27:16):
I can pull them together, createsome joints, you know, that's
more of a technical term, right.
But, but piece them together,mash them up and then I start
asking some questions in a moredynamic fashion and having to
wait on on it.
So it really kind of, it's a,it's a leveler if you will.
The data catalog is and aelation and Colibra two
companies that we were reallyresearching as far as this
(27:36):
machine learning data catalog.
You want to look them up causethey just kind of bubbled at the
top for the things that we weredoing.
One Collibra, it's more focusedon the data governance side, a
little bit more techie from thatperspective in relationship.
Really more focused on the userinterface and how you interact
with the data as opposed to youknow, high level really
governing this thing from aenforcement and a bunch of
rules.
So long wouldn't answer say yes,it's got to go that way.
(27:59):
We've got to make it moreaccessible.
It will happen.
It's not.
If it's more of a, when threeyears that may be a bit
aggressive only because of theinvestments take an organization
to get there and you know,depending on the industry you're
in, insurance tends to runpretty slowly too to some more
of the rapid environments.
Tech obviously serves on thefront side of those things.
(28:22):
I was just going to ask you, youknow, is this a three year, five
year, 10 year?
Like what, what, what do we,what can we expect to see in the
coming years?
He answered that a little bit,but what it feels like, what it
feels like to me, I hear yousaying is, is very similar to
what we saw Google do when itbegan indexing the internet and
websites, you know, in thenineties and they, you know,
(28:47):
they are now a behemoth of ask anatural language, natural
language question in LP, ininside a browser, in tailoring
the results to where you know,what they think you're focused
on and what might be mostimportant.
Can you see a world where eveninside my company, I'm having
(29:10):
the same type of experienceswhere I might have with Google?
Absolutely.
That's probably a great analogy,right?
I just, I just, uh, crawlthrough all of my internal
stuff, right.
And, uh, and create somemechanism that I can, I can
query it.
Right.
Um, so you know that, that'sprobably a great comparison.
(29:33):
Um, the other thing that's kindof happening the industry and I,
I think when I, you and I spokeearlier, we talked about this a
little bit, but to see idea ofhow the API economy, and I don't
know if you, um, kind of bringsinto question, so companies
instead of just holding ontothemselves are gonna figure out
ways to expose it externally and, and monetize it.
Right?
So if, if there's a way topublicly expose, I dunno, even
(29:55):
IP data may be out there forprice, right?
But whatever they, they, theywant to say, okay, we're going
to kind of let you guys haveaccess to our stuff in exchange
for, you know, whatever it is.
And API APIs become kind of themechanisms to do that.
People can ask questions ofthese API APIs under the covers
using other, you know,programmatically, right.
(30:15):
Uh, or I can do it from abrowser, but just being as a way
to kind of expose, uh, data in adifferent way and monetize in
different ways.
So I see the API economy reallykind of becoming more and more,
um, flagrant, if you will.
Speaker 3 (30:29):
Yeah.
Well, and it speaks to astartup.
I was part of a mentoring groupfor several years ago.
You know, when we first talkedabout the idea of this data or
API economy, you know, initiallyit was like, well, yeah, it's
everywhere.
APS are everywhere where wewould would, this isn't anything
new.
But as, as you were talking justnow, there's a component to it
(30:51):
that really rings true becausethis particular, this particular
business model was looking at,uh, activities for a, what
lawyers, how they would, howthey would settle a case, how
long it would take to settle acase on a specific type of
claim.
(31:11):
Who the, who the people werethat were involved with it from
a, from a legal standpoint.
And then, um, taking thatinformation and feeding it back
to an insurance company to helpthem settle faster or for the
right amount, right.
To take the, you know, somebodygot injured in a work accident
and they had to wait three yearsto get their$200,000 or whatever
(31:34):
the number is, they could, theycould settle much quicker.
Well, when you think aboutinformation that sits inside of
organizations, I mean, if Icould just figure out a way to
access that, this thing overhere that would make my life and
everybody else's life so mucheasier.
I mean, I think you're ontosomething with that.
I mean it, and I'm on the techside, we think about it a lot,
(31:56):
but I don't know as you thinkabout it as much when you get
into, um, services basedindustries or you know, finance
and insurance, they're alwayslooking for ways to leverage
information to make more money.
Kyle Hamer (32:08):
Yeah.
Speaker 3 (32:09):
But I don't know, is there fully appreciating or
fully vested into the, Hey, if Ishare information, I could even
make more and more money.
Like, um, it's a, it's a reallyinteresting and kind of a new
concept.
Mark Skiles (32:22):
Well, the whole idea of information sharing,
really scared of it.
That's one of the things thatkind of bumped into a lot of
times because even developmentteams aren't used to sharing,
they think of themselves assiloed.
Right.
I've just, this data is onlyrelevant to my application.
I don't really care whateverybody else thinks about it.
Right.
I'll get into the data warehousethat'll become a single source
Speaker 3 (32:40):
of truth or, or however they, but reality is at
an application standpoint, someof these applications needed to
talk to each othertransactionally in a different
way.
And so they had to startthinking about it in a, in a, in
a, how do we share things?
Well, right.
That was just a, knew that thatwas a totally cultural barrier
that we had to kind of workthrough.
It's not done yet.
That's a conversation.
(33:00):
That's not, yeah.
Well and I completely, I cancompletely appreciate that we're
working with another, uh, we'regoing on another project right
now where we have threedisparate systems all looking at
singular behavior and we'd liketo get it aggregated into a
single view because it's, it'schallenging to say, well, this
one tracks where you, where youcame from, that one tracks how
(33:22):
you behave and this one, this ishow we support and sell and
serve into that.
So it's, it's, uh, the, the, themore technology we build, the, I
think the higher demand there isgoing to be for that, you know,
uh, information exchange orinner polarity between two
applications beyond just here'sthe API, connect them together
(33:43):
using a Zapier where your owncustom code.
It's making it meaningful,making it so that I can get
access to the information.
Yep.
Cool.
And that gets us back to thedata to decisions thing, right?
To turn it from just data, Raulpieces of bits and bites of
things into something useful fordecision making.
So, so ultimately it has to, ifit's not, it's not useful to me
(34:03):
in that fashion.
Uh, then I spend a lot of moneyand get not getting the dollar
value that I need to out of it.
Well, you're, you're spot on andyou've been consistent for years
with, from data to decisions.
Uh, integrity works.
I think that will continue and Ilove, I love hearing about your
project.
Well, thanks.
Thank you.
Enjoyed the conversation.
(34:24):
Thanks for the invite.
Thanks for sharing today, Mark.
Those for those that, uh, wantto get ahold of Mark, his email
address will be on the website,a Haimer marketing.com or you
can
Kyle Hamer (34:35):
searching for Mark Skiles on LinkedIn, although
I'll tell you he's not on thereas much as he is his email.
Um, but, uh, we do appreciateyour time today, Mark.
And for those that arelistening, tune in next week
when we have another episode forsummit.
Interesting topics, businessinsights.
Oh and welcome to the summit, the podcast where we
bring you knowledge and insightsfrom industry leaders and
professionals.
My name is Kyle Haimer.
I'm your host of the podcast andowner of Haimer marketing group.
Today I'm here with my goodfriend Mark Skiles.
Mark Howard today.
Dan fine.
Thanks Kyle.
That's awesome.
Mark.
Uh, Mark and I are going to betalking about the new role of a
(00:27):
chief data officer and what'shappening in the world of
[inaudible] data
Mark Skiles (00:31):
and operations and how businesses and enterprises
can, can expand or maybeleverage data in more
intelligent ways.
But before we get into that,Mark, why don't you tell us a
little bit about yourself?
I'm sure you have kind of a,have a, a bit of a varied, an a,
a, uh, background here.
(00:51):
So started out in networking,um, worked on that for a few
years, became kind of anindependent consultant back in
the early nineties, and, uh, dida variety of projects as a part
of that.
But, uh, over time it's kind offell into the whole data world,
uh, primarily managing SQLservers, right?
You kind of a DBA, uh, and soyou work with the data a lot and
(01:12):
then you start thinking ofsomething more than just moving
these bits and bytes around.
What do they mean to people?
Right?
And so you start thinking aboutthat, uh, kinda migrated towards
the, from a technical side tomore of a business in tech,
business intelligence sphere,right?
How do you use this stuff, makeit work and it get some good
value out of it.
And so you had the mantra for awhile from data to decisions,
integrity works, uh, any of ourcompany integrity works.
(01:35):
And, uh, you know, this wholeidea of finding wherever the
data that you're looking for isat.
So we started throwing aroundthis banding about this, find
it, get it and present it rightin the BI.
In the BI world, reality is,it's not just about the
reporting, but a lot ofdifferent things around data
quality.
Uh, you know, being able to getwhat you want when you need it,
whether you're a data analyst,data scientist, kind of person
(01:57):
more on the business side or youknow, a different set of
challenges from the it side.
And that's, that's kind of wherethe CDO really came into play,
right?
That the it folks are reallyfocused on getting you the data.
Uh, they can locate thedifferent systems, but what's,
what's the most value to anorganization and what are the
business drivers for that data?
You know, how do we want tofocus its efforts?
(02:18):
And so that's really kind ofwhere the CTO came from and it's
an evolving role.
Uh, right.
Organizations have these kindsof challenges when they're
trying to push through to, to,um, this whole, um, data-driven,
you know, world that everybodythrows around.
Um, and there's just a lot ofpieces.
Moving parts to that point froma CDO is just drive from the
(02:38):
business side more.
I have a C at the C level table,uh, as opposed to being just
focused on information and kindof reporting kind of thing.
It's, it's really help us kindof navigate this whole thing and
you could get dry the businessvalue out of it.
So when you say, when you seedrive drive the business value,
one of the, one of the thingswhen you were explaining this to
me, the first time we chattedabout it was the, um, the silos
(02:59):
that information
Speaker 3 (03:00):
and data lives in.
Can you expand a little bit moreabout what you're seeing happen
in the market for, uh, reasonsthat people will need data or to
consume data information and howit's impacting enterprises that
have normally functioned ininformation silos?
Mark Skiles (03:20):
Well, one of the challenges, especially to
business users, right, is, isthey have a question.
They have to go to it to get it.
And so they can go through thewhole, you know, I get it on the
project docket and kind of workthrough the logistics.
They are, take some time, takesome money.
Um, they don't really have anyway to get their, their hands in
the pie and really ask themquestions.
So part of this is kind of beenpushed where, where the business
(03:40):
says, you know, I need it when Ineed it and I want to have to
wait six months for it.
Right?
So the silos really are thisidea that we've got a lot of
systems in our environment,enterprises, large enterprises,
but small, small companies hadthe same issue.
It really just different scale.
And that is, you know, myQuickBooks for example, is as
locked my financial data upsomewhere and it doesn't really
talk very well to my, uh, Idon't know invoicing or my, my,
(04:03):
uh, you know, my customermanagement or, or whatever it
is, right?
The, the, the, the, we've got alot of walls but not a lot of
bridges.
And so really this is around howdo you build bridges with these
different systems?
Not so much from a technicalstandpoint, but from an
organizational understanding andensuring that the things that
are really important to us, uh,get, get kind of elevated focus,
(04:23):
whether you're, you know, inpublishing.
And so it's the, the authors andthe author data there, the book
contents, whatever it is, right,that the right things get the
right focus as opposed to itsometimes being driven by the
current project on the list.
And uh, anyway, it just istrying to drive the value from
the folks who are using itright.
(04:44):
And can really say this is whatI value the most as opposed to
kind of its current prioritiesat the time.
That makes sense.
Speaker 3 (04:51):
It absolutely makes sense.
I mean, it helps betterunderstand why this role is, is
becoming so important because Ithink 20 years ago information
was there, but data wasn'tintegrated.
People weren't wanting toautomate and bring things
together, even ask largerquestions.
It just, you'd send one personout to go comb through the
(05:13):
catacombs of each, you know,each set of filing cabinets and
they came back and they were theOracle of truth.
Whatever their personal opinionwas, was the interpretation of
that information and theassessability of, of software
and technologies really forcingus to look at how we deal with
information, uh, differently.
(05:33):
But, but what, what's not fullyclear to me is, is why
organizations are saying, okay,well we need a data officer
versus making this part of likea chief information officer role
or a, um, security officer.
Like what, what is differentabout a CDO than say like a CIO?
Mark Skiles (05:55):
That's a good question.
So the, for example is a chiefinformation officer, right?
And so a lot of times we'rereally involved in collecting
the data, right?
But the focus really isn't onthe delivery and the benefit of
that data.
Um, and so if just collectingit, put in a data warehouse or
put it somewhere, doesn't giveme everything that I need.
(06:16):
Um, how do you keep your limitedresources focused on what are
doing, what really gives you themost benefit?
And so the CDO really kind ofgives you that clarity and
direction for your attention.
And it's also focused oncollecting smaller data sets as
well and govern them andmaintain them in a, in kind of a
, uh, you know, more focusedmanner.
Speaker 3 (06:32):
Well, that, that's interesting.
So a lot of organizations havelax at best.
I mean, they may have datagovernance, but data hygiene is
something totally different.
Um, but how our, how ourorganization's handling the
change or, or thinking about thechange of, of data governance
(06:52):
from silos to, okay, I have achief data officer now, or we
want to, we want to provideassessability of data across
different silos.
How's that, how that forcingthem to think differently.
Mark Skiles (07:07):
Well, one thing is really on the data quality,
right?
Because in the past you kind oflooked at just, just we have
volumes of data, we've collectedall this stuff, but you know, we
might have different reports atdifferent totals, different, you
know, the, there's conflictseemed, uh, facts if you will.
Right?
And so somebody really can offocused on, um, putting some
oversight and saying, these arethe things that organization we
(07:29):
really value the most.
And this also bleeds into thesecurity conversation, right?
Organizations really need toreally to say, um, for the, for
the most important things in mycompany, I need to have a higher
level security for, I don't needthe same level security for
every piece of data, myorganization.
But I do need to understandwhat's critical for my business.
And if something went down, forexample, um, you know, what's
(07:51):
our, um, not backup plan, butyou know, our, our, our stake in
the ground and says we can'tlive without this.
And if we do, it's going to costthem a heck of a lot money to
get it, get it back.
Right.
And so you just put some energyaround that.
I'm not talking about technicalstandpoint, just saying
identifying it first.
Um, one of the exercises if Ibleed over to, uh, to the later
conversation is that recognizethat we had to understand what
(08:15):
are the different things in, inan organization.
I'll use the example of the uh,biotech.
So you know, a seed for example,but what goes into this seed?
Well, the germplasm is thegenetic makeup.
Well, for example, thatgermplasm is, is a, is a part of
the, the, the DNA conversationand it means different things to
different people.
(08:35):
So for the laboratory, thisgermplasm can, can be a piece of
glue in a, in a sample too,right?
But for the people in the fieldwho are planting those things,
this germplasm comes in theshape of a seed.
But the conversation around whatare the most important things in
our organizational world, whattypes of information are we do
we need?
And they wanted therelationships, those pieces of
(08:56):
information.
So the exercise was kind of thiscontext map around while you use
the term different nouns in ourworld and then you know, how
they interact with each other.
And so that was why one of thethings we learned early on is
they really need to kind of takea step back and, and, and find
out hoarder all these things.
Right.
What's the wet, where is it at?
Was kind of a secondaryquestion.
(09:17):
But this context, you know, ithas different meaning giving
different context and so thatyou kind of had this business
glossary need.
I need understand what it is,what it means and uh, and what's
the context for it.
And so we'll talk about thislater with the data catalog ID
is really kind of a way to, to,to manage that, uh,
(09:37):
understanding of what the salesthings are from a conceptual
standpoint all the way to wherethey're physically located, what
tables are they're at, where thedifferent columns with they're
at in a way to kind of put thosethings in a library form.
Um, it's kind of the geospatialapproach, right?
This whole, I want to find outwhere it's at to my data so
Speaker 3 (09:57):
that, um, you know, it's, when I find what I find
fascinating is you touched onseveral different things there.
One is, uh, data governance.
You talked about, you know,needing the ability to move
faster and different things meandifferent things to different
people.
Speaker 4 (10:16):
No.
Speaker 3 (10:18):
When you have groups that have, you know, their own
sort of internal fighting andchallenges even in their own
departments or silos, what are,what are some of the common
challenges that we seeorganizations having as they, as
they open up?
I mean, you touched on itbriefly as they open up the, uh,
the dialogue into, Hey, I needto have access to research and
(10:41):
development information or Ineed to have access to marketing
or accounting or whatever theadditional, uh, data set is what
are, what are some of the corechallenges that you see as they
begin going down that road as anenterprise or even business
ghost in that road.
Mark Skiles (10:58):
And one of the challenges we ran into and the
reason why we kind of startedwith what's the, what's the,
I'll use the term sharedlanguage.
They use that term in some kindof this domain driven design
world.
If you'd look that up, you'd seewhat I meant by that.
But, but shared language kindaunderstands this.
Every, uh, every application ordevelopment team or something
has, has language that they use.
And so you try to understandwhat the different things that
(11:21):
they are they talking about the,the objective here that we found
or the challenge I guess that wefound was that people are using
the same word.
I'll use the word inventorywithout any context.
And so the first thing we do istry to understand a, what it is
we're talking about.
And then B, what are thedifferent contexts that we're
using and try to shape those insome kind of a, of a, of a
(11:42):
language.
And we, we took that one stepfarther where we actually
modeled them conceptually.
For example, the, uh, the ideaof germplasm, the definition
from a technical, um, betweenthe field and the laboratory was
the same for 80% of theconversation, but the other 20%
was unique to the laboratory orunique to the field, right?
(12:05):
But, but that gave it context.
So, okay, when we talk aboutinventory, here are the things
that we agree on that we're all,you know, we're all speaking the
same language.
Here's the other parts that werecognize are different.
And here's what we mean by that,right?
And so we documented those kindsof things.
We, we used a conceptual modelto do it.
Now there's a variety ofdifferent tools you could do
that.
But that was kind of ourapproach.
(12:26):
Basically, you just need tounderstand what, what you have
the nouns.
We'll get back to using thatterm and then how those things
interact with each other in, inyour organization.
Right.
We're not talking technical atthis point in time.
We're not talking about howthey're implemented in your
database, but just a commonunderstanding of things that
make sense.
Speaker 3 (12:41):
Makes perfect sense.
And I think the thing that'sinteresting, what I think about,
um, writing down what we have,you know, I would imagine that
inside your catalog there has tobe some level of description
and, and for a period of time,silo a or silo B will come in
and they'll look at it, uh, aterm and there'll be thinking as
(13:03):
if they're in their own silo.
But it's really referencing morehow the other silo, um, plays
with it.
So there's the, the, the uh, thedescription component almost
becomes more valuable than eventhe, what the description of
what it is.
You know, it's, its definitionis probably more valuable in
creating common language.
(13:24):
Then the descriptor, the, the,the name itself,
Mark Skiles (13:28):
one of the tools we used is a context map.
And that was really just a bunchof boxes and squares.
They kind of took all the nouns,right?
And they have this idea of someupstream and downstream stuff,
you know, so an experiment forexample, it takes a lot of
different pieces.
A location, right?
The different timing, thedifferent chemical components.
(13:51):
And those are all kind of suckedinto this idea of what is an
experiment?
Well, if you looked at it ontheir little context map, you'd
see a box for experiment, you'dsee a box for germplasm, you'd
see a box for um, you know,chemicals, those kinds of
things.
And then you might see comingout of the experiment box some
other things that are kind of,you know, outputs from that,
things that are produced.
So you have this upstream,downstream kind of thing.
(14:12):
So one of it is the Hoyts andthen like I said, they kind of
interact with each other and wekind of put that into what we
call the context map.
It's not a unit, you know, in,in the manufacturing they have
like a, a entity relationshipkind of things.
Um, but it's basically someoneto visualize that and then begin
to, to speak to those things andpeel them apart.
Right.
And then you start buildingthese models out from there.
(14:34):
Are these definitions to useyour term around what these
things are, right.
Give them some kind of a shape.
Speaker 3 (14:41):
Well, it's, I mean, it sounds like, to me the, I
mean, I know this because we'vechatted about it, but it sounds
to me like this, this, um, thisrule of a CDO or[inaudible]
fractional CDO, uh, you, youplayed this role in a project
where it really wasorchestrating many of these
things coming together.
Can you tell me a little bitabout some of the core
(15:02):
challenges or maybe corebusiness questions that the
business is trying to ask thatthat kind of led to the
evolution of this for, for thatspecific project?
Mark Skiles (15:13):
Well, first of all, her most recent project really
wasn't a, it didn't start out askind of a CTO role.
It kind of evolved in a kind ofa semi CDO.
And it was, it was based in the,the R and D department of a
biotech company.
And so they really were goingaround this data integration
path and using API.
So they brought me in to kind ofsay, um, you know, help us
gather, uh, from a designstandpoint, the, the, the, the
(15:35):
things that we need to bebuilding these core API as they
call them, right?
At DNA integration layer.
What evolved from that initialconversation?
It was focused on building APIs.
We realized they really neededto be some common language
around um, the, the, if we'regoing to do something core and
I'll use core just to meanshared right?
Integration between the commondefinitions of something, then
(15:57):
then w we have to understandwhat it is we're talking about
and everybody had a differentdefinition of stuff.
So, so we kinda went back to thedrawing boards.
Okay, now we need to focus onthe data governance side.
We're going to inequality atdata quality out of this
equation.
We're going to have to towpeople.
They needed to be a commonunderstanding of what we're
talking about.
And so that we kind of backed upthe bus a little bit and started
(16:17):
going down this path of thiscontext mapping exercise and
some some conceptual modelingthat became very valuable cause
they, what happened, we wereable to find a tooling that took
those conceptual models and cameup automatically.
We call the auto-magically,right?
But we could automate somethings.
It pushed him up to a websiteand this, these, a web
publishing of these conceptualmodels became something that we
(16:38):
would walk into a room with anapplication development team and
it gave us something to speak towrite something visually.
I could look at that.
It became a conversation piece.
It didn't have to be exact, butit was enough for everybody to
get an understanding of what theshape was.
And you could even involve it inthe conversation.
Oh well we're missing a piecethere.
Okay.
We'll add it to the laboratoriespiece of that equation because
that's what we're talking about,right?
Throw it right in there is we'reusing the design tool and
(17:00):
display it on the screen.
Everybody's happy.
Go back and publish it again.
Right.
So here's an evolutionary kindof process.
But it was a conversation.
And really if I saw one valueout of this whole thing, it was
forcing the conversations thatpeople hadn't taken the time to
do in the past.
And just a common awareness ofsome of the challenges we were
facing and trying to figure outa way to, to, to address them.
So, um, I, I really see myselfas kind of an early, uh, chief
(17:25):
data officer.
We didn't have a lot of thingsfigured out.
We were kind of walking throughthe process and, and, and, you
know, they love the things we dodifferently, uh, doing it again,
but we learned a lot.
And so that now I had placedmyself in a, in a, you know, a
phase to CDO.
So
Speaker 3 (17:43):
that's a, that's really cool.
So like when, when, when I sayit's really cool, I, I'm
thinking about, you're talkingabout the, the common language
and the sorting things out andyou know, creating
visualizations.
It's like, it was also mostlyyou were building, uh, the
building the role as you werebuilding visibility for the
company.
And I can't imagine what kind ofimpact that was having for teams
(18:06):
around.
You talked a little bit aboutthe impact of getting access to
information or creating commonlanguage with your, with your
catalog.
What, what that meant or whatkind of the aha, like what were
some of the big revelations forthis company?
Mark Skiles (18:22):
Well, first of all, just to clarify, this company
doesn't actually have a datacatalog as of yet.
They recognize a need for it andthey asked me to do a lot of
research on it and so I cameback with some, some
recommendations which haven'tbeen implemented yet.
And that's more of a timingthing, uh, as opposed to
anything else.
But, uh, in terms of elevatingin the, in the value, um, what
happened was as, as uppermanagement began to see these
(18:45):
things and people were usingthem in kind of a practical way,
the data governance came, all ofa sudden got escalated to, you
know, tending a lot of meetings.
People wanted to know, you know,not just what you're working on,
but how does this affect, causethey were going through a merger
and acquisition and that's oneof the times this is really,
really important, right?
You're bringing in a new companyand you need to know that.
Do you mean the same thing whenyou're talking inventory?
(19:08):
Right?
Uh, and so that those really gotelevated into the upper
management levels and because itwas conceptual, they can kind of
grip on, get a grip on thosethings as opposed to, you know,
talking about the bits and bytesor the table definitions or that
kind of a thing.
Right?
They could begin to understandhow these concepts fit together.
And so it became a very criticaland highly valued piece of the
(19:30):
upper management's, um,evolution too.
Because they could get a grip,they could, they could chew on
those conceptual things and geta picture of them, um, as
opposed to showing them tabletable definitions.
Right?
We constantly harping on thatthis is not a table definition.
It does not map to all thecolumns in a table.
Right?
So just to understand this as a,this is a business conceptual
(19:52):
model, not a data, not a, not adatabase table structure.
Speaker 3 (20:01):
I can, I'm, I'm chuckling cause it's like, I
can't imagine, depending on thetype of executive, they were
probably some or even seniorleadership that you would be
talking to if there were somethat were really, really
relieved that you weren'ttalking about database structure
and there are others that werereally annoyed that you were
talking about concepts thatweren't tied into tables.
Mark Skiles (20:19):
Well usually that came in like your data modelers
or your, your DBHs or your, youknow, even your data warehouse
guys.
A lot of times they're sofocused on what the table
structure was or something.
They had a hard time pullingthemselves out of that.
Uh, so, you know, theconversation was great, but
everybody had to do a littlelearning.
I'm okay.
When we walk in the room withMark Skiles with David Bowen,
we're not talking about, youknow, table structures and uh,
(20:43):
that was a learning process foreverybody.
Speaker 3 (20:45):
As you, as you kind of go through this process or
this project, it sounds like alot of the, the silos had to be
torn down.
Did you take a, a warehouse anda cubing approach or a, you
know, is this a dealing withdata lakes?
What does this, what does thisreally mean for how data is not
necessarily interpreted, but howdata is accessed and
Speaker 5 (21:09):
Mmm.
Speaker 3 (21:10):
Cattle catalogs probably you already got the
word most the how, how it'saccessed, how it's, uh,
Mark Skiles (21:17):
well, one of, one of the things if this, if this,
I'll use a higher level model atthe moment just for that
conceptual idea is thatdifferent systems kind of, um,
and, and there's differentlevels of abstraction, you know,
and database world, you havethis conceptual, logical,
physical kind of kind of amoment within data models.
Right?
Um, and so how would you take adata from system a using this
(21:40):
conceptual model and somehowbring it into the data warehouse
or system B?
Right.
So they created something calleda, uh, I think we call it a data
pump, remember what it was, butit basically used, took this
conception model, took fromsystem a transactionally, turned
it into this XML file kind ofthing.
And think could be re-read onthe back end and sucked off the,
the, the data pump, right.
(22:00):
In a way that, that uh, theycould kind of rearrange it on
the other end of the way theywanted to, but they had this
common definition of things andsaid, okay, this is the way it's
, uh, uh, we can read itdynamically, right.
Cause it was XML based or, or uh, uh, it could be Jason Bass,
however you defined it.
But the data pump I think was aXML and a CSS stuff.
(22:21):
Um, so anyway, I don't know ifthat answered your question as
far as how they were doing it,but it's, it, it wasn't so much
a, um, you have this element ofdata mapping.
But the other thing was justautomation around taking these
conceptual models and, and usingthem as a common, like a, uh,
almost a schema kind of thing.
And those of you in your dataworld would kind of understand
the scheme of conversation.
(22:42):
That's really the shape of anygiven, given data coming in.
And they use it in the Jasonworld a lot where some things
are coming in off the internetor you know, some sensors
sending me data in a certainshape, right?
They would call it that dataschema, right?
So, um, the different ways tokind of move from this, this
conceptual thing to somethingthat's, that has more shape,
(23:03):
right?
And so we had tried to do someautomation things to leverage
those models and then you couldplug them in and they can
automate, uh, the moment of frompoint a to point B, given the
use of those, those commondefinitions, that's probably a
little too nebulous for you.
I'm not sure how much detailyou're asking for.
I'm not trying to avoid yourquestion, but I just hesitated,
(23:24):
jump too much into technicaldetails.
So
Speaker 3 (23:25):
yeah, I'm glad.
I'm glad you didn't go anydeeper.
Why I might've like come out theother side, Mark.
But that being said from a, it,it's fascinating to me because I
hear about, you know, and I haveenough, I have enough knowledge
to be dangerous, you know,usually like the, to, to um, to
have the, to understand what'sbeing talked about.
(23:47):
But not everybody fullyappreciates, you know, how
nebulous and fine tuned detailedlike how much information and
structure is buying theinformation that they need.
Mark Skiles (23:56):
Yeah.
Speaker 3 (23:56):
[inaudible] do you think, do you think that like do
you think in this, like theproject that you set up for, for
this company and, and the roleof the evolution of this data
officer and how companies arethinking about it, do you think
that there'll be a,
Mark Skiles (24:13):
the[inaudible]
Speaker 3 (24:13):
point in which information will be free flowing
and we won't be dealing with,not necessarily governance per
se, cause you want to protectyour IP but deal like, will we
be in a spot where marketing canaccess what research and
development is doing and isthree years down the road and
seeing how that may beinfluenced by what they're
seeing happen in the markettoday?
Or is it well we get informationflowing that direction or is it
(24:38):
gonna really stay continued insilos?
It's just, you know, maybebusiness adjacent, there's value
in doing this.
Mark Skiles (24:45):
Well that's a good question.
I think it's going to have to[inaudible].
And what I mean by that is, youknow, you had this, this idea of
um, Oh, what's the term theyuse, can have a um, uh, a person
who could do their own analysis.
Right?
The, the, the um, let me justforgot the term off the top of
my head, but the point isaccessibility is going to be a
(25:06):
big deal because I want to askquestions in, in more real time
as opposed to being hung upwaiting on it.
So to do that you can have togive, uh, the marketing
department or the accountingdepartment or something better
access to their data causereally it's theirs, right?
It's just been kind of locked upin systems.
So the answer to your questionis yes, it's gonna evolve that
way.
The timeframe will be a littlequestion Mark.
(25:27):
I think companies are going tohave to invest in that.
And it does take a little timeand money, right?
This, this catalog ideas.
But one of the solutions to thatis not the only approach, but
this geospatial guide kind ofserves as the library if you
will, from the data definitions.
Oftentimes like the dataglossary, right?
You add the context into theequation so people could look
up, you know, where's, where'sgermplasm at?
(25:47):
Well first of all, what's itmean?
Okay.
Given the context of lab, that'swhat I'm talking about and we
look specifically at labscontext, but where's that labs
are unpleasant located, whatkind of systems is that?
Well, the data catalog kind ofgive you access to that and
especially in a machine learningcatalog was kind of in the next
evolution.
A Gartner calls is the newblack.
Those, those data catalogs willactually um, scrolls, want more
(26:11):
right there.
They'll, they'll, they'll trollthrough your environment, all
yours, your, your different SQLnon sequel, you know, different
kinds of data sources and, andkind of pull those out and say,
well these are what's out there.
You tell me what's good or not.
And somebody that the dataowners, data stewards cause some
crowdsourcing.
We would go out and say, okay,this kid is, has an 80% data
quality.
(26:32):
If you really want stuff, lookat this column.
Right?
And they can kind of pull thosepieces together and get to see
even in a big data environment,these are the, the the, the
piece of information in this bigpool of information that you
want to on, right?
We've been using this for ourtesting or we've been using this
set of production environment.
Don't waste a lot of time andother stuff.
This is it.
As the data catalog gives you aplace to put all that stuff you
(26:54):
can, you can interact with otherpeople who are using it and it
becomes, you know, over time itjust evolves.
It's a tooling, nothing morethan that, but it gives you a
way to kind of put all thesepieces together in a way that
people could find what they'relooking for.
And then many of them now willgive you a query mechanism or
you can do some discoveryqueries on the data.
Is that the right data?
Am I looking for that?
Does they have the pieces ofinformation that I need and uh,
(27:16):
I can pull them together, createsome joints, you know, that's
more of a technical term, right.
But, but piece them together,mash them up and then I start
asking some questions in a moredynamic fashion and having to
wait on on it.
So it really kind of, it's a,it's a leveler if you will.
The data catalog is and aelation and Colibra two
companies that we were reallyresearching as far as this
(27:36):
machine learning data catalog.
You want to look them up causethey just kind of bubbled at the
top for the things that we weredoing.
One Collibra, it's more focusedon the data governance side, a
little bit more techie from thatperspective in relationship.
Really more focused on the userinterface and how you interact
with the data as opposed to youknow, high level really
governing this thing from aenforcement and a bunch of
rules.
So long wouldn't answer say yes,it's got to go that way.
(27:59):
We've got to make it moreaccessible.
It will happen.
It's not.
If it's more of a, when threeyears that may be a bit
aggressive only because of theinvestments take an organization
to get there and you know,depending on the industry you're
in, insurance tends to runpretty slowly too to some more
of the rapid environments.
Tech obviously serves on thefront side of those things.
(28:22):
I was just going to ask you, youknow, is this a three year, five
year, 10 year?
Like what, what, what do we,what can we expect to see in the
coming years?
He answered that a little bit,but what it feels like, what it
feels like to me, I hear yousaying is, is very similar to
what we saw Google do when itbegan indexing the internet and
websites, you know, in thenineties and they, you know,
(28:47):
they are now a behemoth of ask anatural language, natural
language question in LP, ininside a browser, in tailoring
the results to where you know,what they think you're focused
on and what might be mostimportant.
Can you see a world where eveninside my company, I'm having
(29:10):
the same type of experienceswhere I might have with Google?
Absolutely.
That's probably a great analogy,right?
I just, I just, uh, crawlthrough all of my internal
stuff, right.
And, uh, and create somemechanism that I can, I can
query it.
Right.
Um, so you know that, that'sprobably a great comparison.
(29:33):
Um, the other thing that's kindof happening the industry and I,
I think when I, you and I spokeearlier, we talked about this a
little bit, but to see idea ofhow the API economy, and I don't
know if you, um, kind of bringsinto question, so companies
instead of just holding ontothemselves are gonna figure out
ways to expose it externally and, and monetize it.
Right?
So if, if there's a way topublicly expose, I dunno, even
(29:55):
IP data may be out there forprice, right?
But whatever they, they, theywant to say, okay, we're going
to kind of let you guys haveaccess to our stuff in exchange
for, you know, whatever it is.
And API APIs become kind of themechanisms to do that.
People can ask questions ofthese API APIs under the covers
using other, you know,programmatically, right.
(30:15):
Uh, or I can do it from abrowser, but just being as a way
to kind of expose, uh, data in adifferent way and monetize in
different ways.
So I see the API economy reallykind of becoming more and more,
um, flagrant, if you will.
Speaker 3 (30:29):
Yeah.
Well, and it speaks to astartup.
I was part of a mentoring groupfor several years ago.
You know, when we first talkedabout the idea of this data or
API economy, you know, initiallyit was like, well, yeah, it's
everywhere.
APS are everywhere where wewould would, this isn't anything
new.
But as, as you were talking justnow, there's a component to it
(30:51):
that really rings true becausethis particular, this particular
business model was looking at,uh, activities for a, what
lawyers, how they would, howthey would settle a case, how
long it would take to settle acase on a specific type of
claim.
(31:11):
Who the, who the people werethat were involved with it from
a, from a legal standpoint.
And then, um, taking thatinformation and feeding it back
to an insurance company to helpthem settle faster or for the
right amount, right.
To take the, you know, somebodygot injured in a work accident
and they had to wait three yearsto get their$200,000 or whatever
(31:34):
the number is, they could, theycould settle much quicker.
Well, when you think aboutinformation that sits inside of
organizations, I mean, if Icould just figure out a way to
access that, this thing overhere that would make my life and
everybody else's life so mucheasier.
I mean, I think you're ontosomething with that.
I mean it, and I'm on the techside, we think about it a lot,
(31:56):
but I don't know as you thinkabout it as much when you get
into, um, services basedindustries or you know, finance
and insurance, they're alwayslooking for ways to leverage
information to make more money.
Kyle Hamer (32:08):
Yeah.
Speaker 3 (32:09):
But I don't know, is there fully appreciating or
fully vested into the, Hey, if Ishare information, I could even
make more and more money.
Like, um, it's a, it's a reallyinteresting and kind of a new
concept.
Mark Skiles (32:22):
Well, the whole idea of information sharing,
really scared of it.
That's one of the things thatkind of bumped into a lot of
times because even developmentteams aren't used to sharing,
they think of themselves assiloed.
Right.
I've just, this data is onlyrelevant to my application.
I don't really care whateverybody else thinks about it.
Right.
I'll get into the data warehousethat'll become a single source
Speaker 3 (32:40):
of truth or, or however they, but reality is at
an application standpoint, someof these applications needed to
talk to each othertransactionally in a different
way.
And so they had to startthinking about it in a, in a, in
a, how do we share things?
Well, right.
That was just a, knew that thatwas a totally cultural barrier
that we had to kind of workthrough.
It's not done yet.
That's a conversation.
(33:00):
That's not, yeah.
Well and I completely, I cancompletely appreciate that we're
working with another, uh, we'regoing on another project right
now where we have threedisparate systems all looking at
singular behavior and we'd liketo get it aggregated into a
single view because it's, it'schallenging to say, well, this
one tracks where you, where youcame from, that one tracks how
(33:22):
you behave and this one, this ishow we support and sell and
serve into that.
So it's, it's, uh, the, the, themore technology we build, the, I
think the higher demand there isgoing to be for that, you know,
uh, information exchange orinner polarity between two
applications beyond just here'sthe API, connect them together
(33:43):
using a Zapier where your owncustom code.
It's making it meaningful,making it so that I can get
access to the information.
Yep.
Cool.
And that gets us back to thedata to decisions thing, right?
To turn it from just data, Raulpieces of bits and bites of
things into something useful fordecision making.
So, so ultimately it has to, ifit's not, it's not useful to me
(34:03):
in that fashion.
Uh, then I spend a lot of moneyand get not getting the dollar
value that I need to out of it.
Well, you're, you're spot on andyou've been consistent for years
with, from data to decisions.
Uh, integrity works.
I think that will continue and Ilove, I love hearing about your
project.
Well, thanks.
Thank you.
Enjoyed the conversation.
(34:24):
Thanks for the invite.
Thanks for sharing today, Mark.
Those for those that, uh, wantto get ahold of Mark, his email
address will be on the website,a Haimer marketing.com or you
can
Kyle Hamer (34:35):
searching for Mark Skiles on LinkedIn, although
I'll tell you he's not on thereas much as he is his email.
Um, but, uh, we do appreciateyour time today, Mark.
And for those that arelistening, tune in next week
when we have another episode forsummit.
Interesting topics, businessinsights.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.