►
From YouTube: Analytics and Insights Discussion: Jeremy and Mek
Description
Mek Stittri (Director of Quality Engineering) and Jeremy Watson (Group Product Manager, Manage) discuss GitLab's analytics strategy and how the Insights feature (https://docs.gitlab.com/ee/user/group/insights/) will likely play an important role.
A
Hey
thanks.
Everyone
I
just
looked
on:
recording
I'm,
Jeremy,
Watson
I'm,
a
product
manager
here
get
lab
I'm
talking
with
MEC
our
director
of
quality
engineering
about
analytics
and
get
lab
and
MEC,
which
is
talking
a
little
bit
about
how
he
sees
kind
of
periscope,
which
is
a
visualization
tool
that
we
use
right
now
to
kind
of
as
a
stopgap
for
some
of
our
visualization
needs
here
in
get
lab
and
how
he
sees
kind.
That
kind
of
like
living
alongside
get
lab
kind
of
longer-term
great.
B
B
B
B
Do
this
because
it
lists
with
the
co,
it
lists
the
much
requested.
That's
the
labels
output,
the
artifact
as
the
high
level
CSP.
That's
your
best
ur.
What?
But
everything
else
may
be
diving
in
filtering
drilling
down
deeper.
We
can
make
it
as
part
of
the
anodic
split
map,
so
we
thought
I'll
give
it
back
to
Jeremy.
So.
A
We
have
women
new
product
leaders
joining
in
a
couple
weeks
to
talk
a
little
bit
about,
and
kind
of
lead
us
here
in
analytics.
Iii
know
some
of
the
background,
but
for
the
benefit
of
of
him
and
the
recording,
what
are
some
of
the
barriers
and
that
are
kind
of
keeping
us
married
to
periscope
right
now
and
forcing
us
to
use
that
as
kind
of
a
visualization
tool
at
a
reporting
tool
and
not
able
to
talk
through
this
vanilla,
yeah.
B
What
can
we
use
the
current,
for
example,
the
VSM
our
insight
for
and
solve
the
business
use
case
and
have
McDuff
food
everywhere
have
have
team
dedicated
dashboards
that
adds
value
to
the
leaders
in
that
group.
Hear
feedback
from
them
in
a
bit
a
rate
I
think
I.
Think
the
key
to
make
things
mature
is
extreme
dog,
fooding,
I.
Think
all
the
features
in
get
lab
that
I'm,
not
sure,
is
a
result
of
us
just
dog
fooding,
not
funny.
This
is
the
same
thing.
I
say
to
my
team:
I'm,
proud
of
the
quality
engineering
power.
B
It
hasn't
spent
a
dollar
for
external
party
tools.
I
think
we
use
the
flat
it.
My
name
is
hard.
We
use
people
app,
for
example,
the
test
case
management
system
right
now.
There's
other
providers,
but
hey
you
skip
lab
projects,
use
issues
just
use
our
product.
If
there's
any
additional
thing
added,
but
you
need
to
live
and
breathe
inside
get
lab
for
you
to
be
able
to
see
the
bigger
picture
and
value
versus
just
being
distracted
by
some
other
third-party
tool.
What.
A
So
one
of
the
key
we're
vesting
more,
this
idea
of
value
stream
management
and
we've
reaped
cycle
analytics
as
value
stream
analytics
and
we're
gonna
make
that
the
framework
kind
of
our
analytics
kind
of
road
pound.
What
my
understanding
is
that
no
one,
no
one,
is
really
actively
using
cycle
analytics
at
the
moment
or
value
stream
analytics
internally
at
gitlab.
A
Why
do
you
think
that
is
because
I
agree
with
you
when
you,
when
you
talk
about
dogfooding,
and
so
it
has
a
product
manager,
I
want
to
say
how
the
heck
can
we
dog
food
VSA
internally
and
start
iterating
on
this,
but
without
usage
we
can't
we
can't
we
can
start
dog
fooding.
So
how
do
we
get
to
the
first
step
and
how
do
I
make
this
feature
really
useful
for
you
and
your
team
I
think.
B
The
innovation,
the
20/80
is
accuracy
of
the
data
and
the
ability
to
look
at
a
time
span,
and
this
is
where
I'm
excited
that
insights
it's
moving
in,
because
now
we
have
a
high
level
view
like
now
makes
crunching
numbers.
This
is
like
the
exec
V
like
hey.
This
is
hot
spot,
but
there's
also
the
other
drew
down
view
where
you
need
to
see
raw
data
based
on
the
futures.
You
give
it
and
like
hey.
Is
it
matching
this
right,
I
think
giving
confidence
to
users
that
that
single
number
is
correct
from
they're?
B
B
Correct
and
to
even
go
further,
maybe
we
can
provide
I
back
test
features
like
hey.
This
is
the
shark
like.
What
do
you
want
to
see?
You
add
some
filters:
okay,
the
numbers
match,
but
the
numbers
and
that
crunch
addition,
the
time
series
chart
the
numbers,
a
dynamic
table,
all
of
you.
They
match
it.
Hey
I,
have
confidence,
I'll,
use
this
and
then
moving
forward
is
how
do
you
solve
all
the
other
sharks
that
we
have
in
personal
cycle
time?
But
then
there's
like
mean
there's
like
p95:
that's
a
p99
like
no
other.
B
We
are
interested
in
the
p90
fives,
because
those
are
the
spikes
that
causes
the
number
of
the
average
to
go
up.
How
can
I
drill
down
to
that
number
find
the
NMR
find
the
people
there
that
is
going
through
this
hardship,
so
I
can
help
them
find
a
job,
that's
affecting
it?
What
can
we
do
so
like
I
think
those
are
really
I.
Think
I
mentioned
like
three
main
points,
but
I
think
from
my
point
of
view,
that
is
key
to
the
success
of
my
adoption
of
them
and
Alex.
That's.
A
Really
helpful
I'm
gonna
share
some
I
like
to
quick
walk
through,
so
we
have
a
number
of.
We
have
the
customer
advisory
board
coming
up
in
a
couple
weeks,
and
then
we
have
some.
Some
analysts
touch
points
that
we
are
gonna
share.
Our
perspective,
Ayanna
likes
I'm
gonna,
just
give
you
like
a
five
minute
walk
through
of
kind
of
my
perspective
on
this.
This
is
super
unrefined,
so,
like
please,
you
know,
caveat
I'll,
keep
that
in
mind
the
feedback
I'd
love
to
you
to
get
for.
You
is
based
off
of
what
you
just
shared.
A
Do
you
think
this
sets
us
on
the
right
path
and
be
given
the
vision?
Is
this
something
that
you
see
you
using
as
an
engineering
director
EQ
lab?
So
the
audience
for
this,
for
this
deck
is
basically
someone
who
is
unfamiliar
with
our
direction
and
for
analytics
and
wants
to
understand
kind
of
our
perspective
and
our
strategy,
and
in
my
perspective,
like
it,
takes
more
than
just
being
like
just
doing
scrum
in
one
place
to
call
yourself
like
an
agile
Enterprise
right.
A
I
could
do
you
have
an
organization,
that's
like
oh
we're
doing
we're
so
agile,
because
we're
going
to
scrum
and
the
real
picture
is
that,
no
matter
what
organizations
are
in
like
you,
it
takes
more
than
just
doing
scrum
and
one
part
of
the
value
stream
to
call
yourself
like
a
natural
organization.
We
have
that's
why
I
get
lab
is
so
powerful
right,
because
we
have
this
single
application
viewpoint.
It's
like
from
end
to
end
like
we're.
A
I
want
to
find
the
bottlenecks
and
I
want
to
be
able
to
find
groups
that
are
doing
really
well
see,
what's
through,
with
what
they're
doing
really
well
and
it
proliferate
that
the
rest
of
my
organization
and
the
thing
that
resonates
with
me
is
like
the
ideas
in
Six
Sigma,
which
are
like.
We
need
to
find
what
the
problem
is.
Maybe
you'll
measure
it
analyze
it
improve
it
and
then
put
controls
in
place
that
prevent
that
from
happening
again.
A
So
at
the
moment,
like
I'm
focused
on
two
and
three
right
now
and
then
eventually,
we're
gonna
go
to
four,
which
is
like
we're
gonna
get
smarter
about
finding
things
that
to
specific
things
you
can
do
to
improve
and
then
eventually,
in
a
few
years,
we'll
probably
get
to
number
five,
which
is
we're
gonna,
be
able
to
have
like
these
controls
so
that
you'll
be
able
to
place
in
the
product
to
prevent
things
from
happening
again,
and
you
can
just
iterate
and
continuously
improve.
We
have
this
cool.
A
You
know
framework
of
how
we
think
about
about
DevOps,
but
if
you
just
lay
this
out
in
the
in
a
value
chain,
you
get
something
like
this,
which
is
you
know
right
now.
Value
stream,
analytics
cycle
analytics
talks
about
this
in
a
different
way
and
I
think
that
we
need
to
put
this
DevOps
framework
directly
in
the
product
where
you
start
with
plan.
A
So
you
know,
one
thing
that
we
can
do
is
like
the
value
stream
analytics
tries
to
do
is
just
show
the
time
spent
in
each
stage,
which
is
like
the
the
the
eventually
I'd
love
to
see
this
broken
down
into
like
waste
and
time
and
actual
like
active
time.
But
right
now
we're
just
thinking
about
this
in
terms
of
just
general
time
in
the
stage.
So
the
change
that
I
want
to
see
us
move
towards
is
I
want
to
see
this
directly
in
the
actual
product.
A
A
What
does
that
mean
and
I
want
to
be
able
to
allow
people
to
drill
deeper,
which
is
for
each
one
of
those
stages?
I
should
be
able
to
click
into
you,
know,
planning
or
create
or
verify,
and
then
to
see
like.
Oh,
this
went
up.
Why
tell
me
more
because
any
improvement
made
anywhere
except
besides,
the
bottleneck
is
an
illusion
like
that.
Is
your
constraint
in
your
development
process
like
you,
if
you,
if
you
have
a
bottleneck,
the
thing
that
you
need
to
focus
on
is
finding
it
and
then
alleviating,
and
so
like.
A
That's
where,
like
I
see
so
right
now
like
for
planning
analytics,
that's
where
I
see
like
insights,
for
instance,
like
one
thing
that,
like
I
I,
think
is
one
of
the
killer.
Features
that
like
insights,
allows
us
to
do
is
defect.
Tracking
people
want
to
be
able
to
see
like
mean
time
to
time
to
resolve,
I
want
to
be
able
to
track
issues
and,
like
my
work
flow
across
the
instance,
it
insights-
you
can
do
all
of
this
right
now.
A
I
think
that
there's
things
that
we
need
to
do
in
the
future
around
like
automating
issue
workflows,
so
that
you
know
you
have.
You
know,
guarantees
that,
like
this,
these
these
charts
are
trustworthy
and
that
we
we
know
that
people
are
adhering
to
them.
I
know
a
lot
of
the
work
that
your
team
does
is
trying
to
enforce,
like
the
use
of
these
thank
issue
that
I
love
labels
so
that
people.
A
So,
though
we
can
trust
the
data,
I
think
there's
more,
we
could
do
there
to
make
sure
that
you
know
we
have
like
triage
functionality
in
the
product
so
that
don't
have
to
question
your
data
or
like
whether
or
not
people
are
labeling
things
York
right
way
and
you're
able
to
see
like
the
P
95
and
the
P
99
s
based
off
of
this,
which
is
like
okay
I,
have
plan.
I
can
see
kind
of
like
how
people
are
scheduling
issues
like
our
capacity.
Let
me
double
click
into
that
and
I.
B
I
think
this
is
great
I.
One
thing
I
want
to
call
out
is
I
love
sci-fi.
We
can
go
back
to
that
for
a
little
bit
sure
I
think
what
you're
looking
at
this
is
a
Fox
5,
&
6
you're.
Looking
at
this
at
the
whole
stage,
but
alcide
see
the
pattern
for
inside
amount
explore
you
go
to
spy
5
I,
think
like
5
sci-fi,
it's
so
measure
and
allies.
B
The
one
that
you
had
the
Six
Sigma
perfect,
thank
you
yeah,
so
I
think
analysis,
self,
also
kind
of
fits
this
pattern
because
we
have
measure
and
allies
and
it
Maps.
What
I
said
earlier
to
begin
at
the
column
where
we
need
to
measure
and
I
think
measure
is,
is
the
raw
data
like
the
x
huge
shark
analyze
is
too
high
value
top-down
X
X
summary
like
what
is
the
hotspot?
This.
B
High
impact
to
your
business,
you
need
to
just
understand
these
10
numbers
and
then
improve
is
like
how
do
we
continue
to
iterate
and
improve
on
the
data
drilling
down?
Assuming
when
you
get
when
you
get
the
number
you
can
do
down
from
analyze
to
measure
and
from
measure
to
even
down
to
the
issue
level
like
what
is
that
issue
that
is
like
causing
this
spiking
time?
B
A
Yeah
I
agree:
I
think
that
that
really
resonates
with
me
as
well
and
I
think
that
we're
focused
on
number
two
and
then
number
three
right
now
and
like
if
you
go
deeper
into
like
create,
for
instance,
this
is
like
I
want.
This
is
the
the
you
can
see
like
we're,
starting
to
think
about
this
a
little
bit
with
you
know,
I
want
again
I
want
to
be
able
double-click
in
to
create
okay,
our
you
know
our
developers
are
less
productive
over
time.
Why
is
that?
A
A
B
I'm
actually
happy
that
you
brought
up
plan
analytics
because
that's
actually
eighty
percent
of
what
the
current
in
size
curve
bility
is
it's
like
time
series
data
issue
and
labels
behind
of
yeah.
We
have
a
community
to
render
mush
requests
with
labels,
but
that's
probably
the
facet
of
create
and,
as
we
mature
and
I
was
expecting,
I
think
the
vision
will
be
clear
where
this
dissing
Alex
and
this
chart
is
like
under
it's
like
plan
ish
and
this
chart
is
like
create
ish
and
then
your
overall
and
now
this
umbrella
comes
over
like
hey.
B
The
other
thing
I
want
to
mention
before
time
is
this.
This
vision,
that's
really
close
to
what
the
engineering
opportunity
was
doing
started
doing
last
year,
which
is
there's
this
to
two
areas.
One
is
labeling,
you
should
see
accuracy
of
the
data,
that's
the
input.
The
output
is
insights,
which
the
first
prototype
was
the
quality
dashboard,
so
input,
output,
input,
output
and
I
think
we
kind
of
have
a
clear
line
of
sight
into
what
the
output
piece
looks
like
with,
like
everything
being
in
the
same
direction,
strategic-wise
with
analytics,
but
that's
not
that's
the
output.
B
A
B
Next
iteration
that
we
have
is
try
of
serverless,
so
we're
starting
to
combine
things
so
that
we
don't
have
to
wait
for
the
next
day.
It's
it's
based
on
events
right
now.
If
you
add
a
desk
link,
it
would
just
automatically
add
a
customer
label
can
be
planned
to
be
called
that
going
further.
Why
don't
we.
A
B
Have-
and
this
is
me
like
in
looking
at
the
big
picture-
my
it
might
not
be
realistic,
but
at
the
label
level,
what
can
we
just
define
rules
there's
instead,
instead
of
having
BOTS
and
automation,
yeah,
no
files,
complex
configurations,
the
rule
to
try
ask
is
to
just
be
map
directly
when
you
create
the
label.
I
create
this
label.
I
want
every
sake.
Second
week
of
the
month,
you
run
this
rule.
This
is
a
condition
if
it
matches
this
condition.
B
Apply
this
label
to
this
issue,
apply
this
label
to
Amar's,
save
it
at
the
back
end.
We
integrate
with
serverless
and
just
have
that
spin
up
resource
and
run
that
rule
when
you
create
so
I
think
that's
like
converging
and
usability,
but
I
think
that's
where
we
should
be
headed
in
terms
of
them.
Adding
value
to
the
product.
I
think.
A
So
I
know
that
we've
introduced
like
customizable
cycle,
analytic
stages,
I'm,
not
convinced,
that's
the
right
way
to
go
because
of
the
fact
that
it
might
need
to
be
at
the
issue
or
level
so
that
you
can
define
this
logic.
There
can
be
guaranteed
that
you
understand
how
your
logic
works.
Maybe
it's
not
being
applied,
but
you
know
your
logic
is
wrong,
but
at
least
you
know
that
what
the
rules
are
and
you
can
trust
the
data
that
that's
coming
out
so
cool.
That's
that's
really
helpful.