1
00:00:00,040 --> 00:00:03,520
Welcome to the M365 FM podcast,
where we explore the people,
2
00:00:03,520 --> 00:00:06,240
ideas and technologies shaping
the future of work.
3
00:00:06,240 --> 00:00:08,240
Why cloud costs feel out of
control.
4
00:00:08,240 --> 00:00:11,080
Most companies find out they're
overspending on cloud when the
5
00:00:11,080 --> 00:00:14,000
invoice lands, and by then the
money is already gone.
6
00:00:14,000 --> 00:00:18,000
Welcome to the M365 FM podcast,
where we explore the people,
7
00:00:18,000 --> 00:00:21,039
ideas and technologies shaping
the future of work.
8
00:00:21,039 --> 00:00:23,399
FinOps sounds like another piece
of jargon, but it's really the
9
00:00:23,399 --> 00:00:26,039
missing conversation between the
people who build with technology
10
00:00:26,039 --> 00:00:27,600
and the people who pay for it.
11
00:00:27,600 --> 00:00:29,039
Cloud bills grow quietly, one
12
00:00:29,039 --> 00:00:30,399
small decision at a time, and
13
00:00:30,399 --> 00:00:32,000
nobody notices until someone
14
00:00:32,000 --> 00:00:33,200
asks who actually owns the
15
00:00:33,200 --> 00:00:33,719
number.
16
00:00:33,719 --> 00:00:34,799
By the end of this episode,
17
00:00:34,799 --> 00:00:36,060
you'll understand what FinOps
18
00:00:36,060 --> 00:00:38,079
is, how it works, where it fits
19
00:00:38,079 --> 00:00:39,520
in Microsoft three hundred sixty
20
00:00:39,520 --> 00:00:40,759
five and Azure, and the mistakes
21
00:00:40,759 --> 00:00:42,060
that trip up almost everyone who
22
00:00:42,060 --> 00:00:42,759
tries it.
23
00:00:42,759 --> 00:00:45,399
Cloud computing changed how
companies pay for technology,
24
00:00:45,399 --> 00:00:47,479
and that shift sits underneath
the whole problem.
25
00:00:47,479 --> 00:00:50,240
Years ago, you bought a server
once and spread the cost across
26
00:00:50,240 --> 00:00:52,560
several years, so you knew the
number in advance.
27
00:00:52,560 --> 00:00:54,439
Today you rent computing by the
second.
28
00:00:54,439 --> 00:00:57,079
The bill rewrites itself every
day and no single person feels
29
00:00:57,079 --> 00:00:58,240
responsible for it.
30
00:00:58,240 --> 00:01:00,950
Picture a developer who spins up
a virtual machine on a Tuesday
31
00:01:00,950 --> 00:01:02,710
afternoon to test one idea.
32
00:01:02,710 --> 00:01:04,030
A meeting pulls them away.
33
00:01:04,030 --> 00:01:05,510
The test is forgotten, and that
34
00:01:05,510 --> 00:01:06,709
machine keeps running and
35
00:01:06,709 --> 00:01:07,230
charging.
36
00:01:07,230 --> 00:01:10,109
For months, nobody chose to
waste that money, and that quiet
37
00:01:10,109 --> 00:01:12,870
accident is exactly what makes
the league so easy to miss.
38
00:01:12,870 --> 00:01:15,189
The split shows up in the
language people use.
39
00:01:15,189 --> 00:01:17,950
Finance asks for a forecast it
can trust and engineering
40
00:01:17,950 --> 00:01:20,790
answers with a diagram of
services that scale on demand.
41
00:01:20,790 --> 00:01:21,989
Neither side is wrong and
42
00:01:21,989 --> 00:01:23,069
neither side can see the whole
43
00:01:23,069 --> 00:01:23,909
picture alone.
44
00:01:23,909 --> 00:01:24,950
Which is why the bill keeps
45
00:01:24,950 --> 00:01:25,950
growing while everyone feels
46
00:01:25,950 --> 00:01:26,510
busy.
47
00:01:26,510 --> 00:01:28,069
So the money drains away in
48
00:01:28,069 --> 00:01:29,989
dozens of tiny moments in almost
49
00:01:29,989 --> 00:01:31,349
every company that touches the
50
00:01:31,349 --> 00:01:31,950
cloud.
51
00:01:31,950 --> 00:01:34,510
That leak has a name, and more
importantly, it has a practice
52
00:01:34,510 --> 00:01:35,709
built to stop it.
53
00:01:35,709 --> 00:01:37,909
What FinOps actually means.
54
00:01:37,909 --> 00:01:39,230
Start with the name because it
55
00:01:39,230 --> 00:01:40,230
gives away more than you would
56
00:01:40,230 --> 00:01:40,870
expect.
57
00:01:40,870 --> 00:01:43,670
FinOps is a mashup of finance
and DevOps, and that pairing is
58
00:01:43,670 --> 00:01:44,829
the entire idea.
59
00:01:44,829 --> 00:01:46,950
It describes the dialogue
between the builders and the
60
00:01:46,950 --> 00:01:49,790
payers, which sounds obvious
until you notice how rarely
61
00:01:49,790 --> 00:01:51,430
those two groups actually talk.
62
00:01:51,430 --> 00:01:53,629
Here's the part that surprises
most beginners.
63
00:01:53,629 --> 00:01:56,390
FinOps isn't a product you buy,
and it isn't a single job title
64
00:01:56,390 --> 00:01:57,629
you can hire for.
65
00:01:57,629 --> 00:01:58,965
The FinOps Foundation, the
66
00:01:58,965 --> 00:01:59,980
nonprofit that writes the
67
00:01:59,980 --> 00:02:01,140
playbook, calls it an
68
00:02:01,140 --> 00:02:02,379
operational framework and a
69
00:02:02,379 --> 00:02:03,219
cultural practice.
70
00:02:03,219 --> 00:02:06,219
That's a formal way of saying
it's a shared habit rather than
71
00:02:06,219 --> 00:02:07,379
a piece of software.
72
00:02:07,379 --> 00:02:09,060
The goal isn't simply to shrink
the bill.
73
00:02:09,060 --> 00:02:10,500
The foundation says it plainly
74
00:02:10,500 --> 00:02:11,740
if you think FinOps is about
75
00:02:11,740 --> 00:02:13,259
saving money, think again,
76
00:02:13,259 --> 00:02:14,180
because it's really about
77
00:02:14,180 --> 00:02:15,259
getting the most value out of
78
00:02:15,259 --> 00:02:16,020
technology.
79
00:02:16,020 --> 00:02:18,419
Sometimes the smart move is to
spend more on something that
80
00:02:18,419 --> 00:02:21,020
earns more, and the practice
gives you the numbers to make
81
00:02:21,020 --> 00:02:22,379
that call with confidence.
82
00:02:22,379 --> 00:02:25,259
You'll also hear the same idea
called cloud cost management or
83
00:02:25,259 --> 00:02:26,740
cloud financial management.
84
00:02:26,740 --> 00:02:29,699
Those labels describe a task,
while FinOps describes a culture
85
00:02:29,699 --> 00:02:31,020
that spreads across teams.
86
00:02:31,020 --> 00:02:33,419
And that difference matters more
than it sounds.
87
00:02:33,419 --> 00:02:34,780
Picture the difference in
practice.
88
00:02:34,780 --> 00:02:37,300
A team wants to run an analytics
job every night.
89
00:02:37,300 --> 00:02:38,939
The cheap option takes six hours
90
00:02:38,939 --> 00:02:40,180
on a slow machine, and the
91
00:02:40,180 --> 00:02:41,379
pricier option finishes in
92
00:02:41,379 --> 00:02:42,180
twenty minutes.
93
00:02:42,180 --> 00:02:44,780
FinOps doesn't automatically
pick the cheaper one.
94
00:02:44,780 --> 00:02:46,139
It asks what those saved hours
95
00:02:46,139 --> 00:02:47,479
are worth, then makes the trade
96
00:02:47,479 --> 00:02:48,900
off on purpose instead of by
97
00:02:48,900 --> 00:02:49,620
accident.
98
00:02:49,620 --> 00:02:50,780
That habit of deciding on
99
00:02:50,780 --> 00:02:51,740
purpose is the heart of the
100
00:02:51,740 --> 00:02:52,419
whole thing.
101
00:02:52,419 --> 00:02:54,900
It's the reason the practice
belongs to the company rather
102
00:02:54,900 --> 00:02:55,979
than to one department.
103
00:02:55,979 --> 00:02:58,759
Hold on to that idea because it
explains something strange.
104
00:02:58,759 --> 00:02:59,800
You'll see later.
105
00:02:59,800 --> 00:03:01,159
The biggest savings come from a
106
00:03:01,159 --> 00:03:02,639
decision about people rather
107
00:03:02,639 --> 00:03:03,500
than a decision about
108
00:03:03,500 --> 00:03:04,400
technology.
109
00:03:04,400 --> 00:03:05,719
The practice grew out of a real
110
00:03:05,719 --> 00:03:06,919
shift in how technology gets
111
00:03:06,919 --> 00:03:07,439
funded.
112
00:03:07,439 --> 00:03:08,719
Companies used to treat it as
113
00:03:08,719 --> 00:03:10,120
one big yearly purchase,
114
00:03:10,120 --> 00:03:11,159
something accountants call
115
00:03:11,159 --> 00:03:12,479
capital expense.
116
00:03:12,479 --> 00:03:13,879
The cloud flipped that into a
117
00:03:13,879 --> 00:03:15,000
running cost that changes
118
00:03:15,000 --> 00:03:16,560
constantly, which left finance
119
00:03:16,560 --> 00:03:17,939
teams with no fixed number to
120
00:03:17,939 --> 00:03:18,840
plan around.
121
00:03:18,840 --> 00:03:20,159
FinOps grew up to bridge that
122
00:03:20,159 --> 00:03:21,840
gap, and it has since stretched
123
00:03:21,840 --> 00:03:23,479
far past the cloud into
124
00:03:23,479 --> 00:03:24,719
software, subscriptions,
125
00:03:24,719 --> 00:03:26,819
licenses, data centers, and now
126
00:03:26,819 --> 00:03:27,520
AI.
127
00:03:27,520 --> 00:03:30,520
Today, the numbers show how far
it has spread in the FinOps
128
00:03:30,520 --> 00:03:31,919
Foundation's latest survey.
129
00:03:31,919 --> 00:03:33,400
Ninety eight percent of teams
130
00:03:33,400 --> 00:03:35,039
said they manage AI spending in
131
00:03:35,039 --> 00:03:35,599
some form.
132
00:03:35,599 --> 00:03:38,719
Ninety percent manage or plan to
manage software subscriptions.
133
00:03:38,719 --> 00:03:41,819
What began as a cloud cost club
now touches almost every kind of
134
00:03:41,819 --> 00:03:43,360
technology a company buys.
135
00:03:43,360 --> 00:03:45,280
None of this requires an
engineering degree.
136
00:03:45,280 --> 00:03:48,159
The practice is built for the
person who is new to all of it.
137
00:03:48,159 --> 00:03:49,599
The first step is simply asking
138
00:03:49,599 --> 00:03:51,080
where the money goes and who is
139
00:03:51,080 --> 00:03:52,080
watching it.
140
00:03:52,080 --> 00:03:54,759
The six habits every FinOps team
shares.
141
00:03:54,759 --> 00:03:57,280
The foundation lays out six
principles, and they make the
142
00:03:57,280 --> 00:04:00,509
most sense when you read them as
habits rather than rules.
143
00:04:00,509 --> 00:04:03,229
The first is that teams have to
collaborate, which means
144
00:04:03,229 --> 00:04:06,430
finance, engineering and the
business all look at the same
145
00:04:06,430 --> 00:04:09,110
numbers instead of arguing
across a wall.
146
00:04:09,110 --> 00:04:10,189
The second habit is that
147
00:04:10,189 --> 00:04:11,710
business value drives the
148
00:04:11,710 --> 00:04:12,229
decisions.
149
00:04:12,229 --> 00:04:14,389
You aren't chasing the lowest
possible number, because a
150
00:04:14,389 --> 00:04:17,110
service that is cheap and breaks
every week isn't cheap at all.
151
00:04:17,110 --> 00:04:18,589
The third is that everyone owns
152
00:04:18,589 --> 00:04:19,829
their own usage, so the person
153
00:04:19,829 --> 00:04:20,870
who built a resource is the
154
00:04:20,870 --> 00:04:21,910
person who watches what it
155
00:04:21,910 --> 00:04:22,629
costs.
156
00:04:22,629 --> 00:04:24,389
The fourth habit is about the
data itself.
157
00:04:24,389 --> 00:04:26,910
Cost information has to be
timely, accurate and available
158
00:04:26,910 --> 00:04:27,910
to the people who need it.
159
00:04:27,910 --> 00:04:29,110
A report that shows up three
160
00:04:29,110 --> 00:04:30,310
weeks late can't change a
161
00:04:30,310 --> 00:04:31,509
decision that's already been
162
00:04:31,509 --> 00:04:32,189
made.
163
00:04:32,189 --> 00:04:34,829
The fifth is that a small
central team enables the whole
164
00:04:34,829 --> 00:04:38,189
practice, sharing best practices
and holding the standards.
165
00:04:38,189 --> 00:04:40,589
The real work stays close to the
teams doing it.
166
00:04:40,589 --> 00:04:43,069
The sixth habit is the one
people forget, and it's about
167
00:04:43,069 --> 00:04:45,310
using the flexibility you're
already paying for.
168
00:04:45,310 --> 00:04:46,470
The cloud lets you shrink when
169
00:04:46,470 --> 00:04:47,589
demand drops and grow when
170
00:04:47,589 --> 00:04:48,509
demand spikes.
171
00:04:48,509 --> 00:04:51,269
So the goal is to match what you
pay to what you actually use.
172
00:04:51,269 --> 00:04:52,709
That blend of a central core and
173
00:04:52,709 --> 00:04:54,230
local ownership is the norm, not
174
00:04:54,230 --> 00:04:55,069
the exception.
175
00:04:55,069 --> 00:04:57,660
In the Foundation's twenty
twenty six survey, eighty one
176
00:04:57,660 --> 00:05:00,319
percent of teams ran either a
centralized enablement model or
177
00:05:00,319 --> 00:05:01,300
a hub and spoke model.
178
00:05:01,300 --> 00:05:03,300
That's a formal way of
describing a small core group
179
00:05:03,300 --> 00:05:04,620
that coaches everyone else.
180
00:05:04,620 --> 00:05:07,100
That doesn't mean the central
team does the work itself.
181
00:05:07,100 --> 00:05:09,420
It means the central team draws
the map and hands out the
182
00:05:09,420 --> 00:05:11,839
compasses, because one hundred
people watching their own costs
183
00:05:11,839 --> 00:05:14,339
will always beat five people
watching everyone else's.
184
00:05:14,339 --> 00:05:15,500
The ownership habit sounds
185
00:05:15,500 --> 00:05:16,819
abstract until you watch at
186
00:05:16,819 --> 00:05:17,579
work.
187
00:05:17,579 --> 00:05:20,300
A team that tags its resources
and checks its own monthly
188
00:05:20,300 --> 00:05:23,220
number starts choosing smaller
machines on its own.
189
00:05:23,220 --> 00:05:26,060
Nobody sends a memo because the
cost finally belongs to a person
190
00:05:26,060 --> 00:05:27,300
instead of a spreadsheet.
191
00:05:27,300 --> 00:05:28,420
The collaboration habit works
192
00:05:28,420 --> 00:05:30,060
the same way when finance and
193
00:05:30,060 --> 00:05:31,819
engineering share one dashboard,
194
00:05:31,819 --> 00:05:33,040
the argument about the bill
195
00:05:33,040 --> 00:05:34,540
becomes a shared question about
196
00:05:34,540 --> 00:05:35,980
which services are earning their
197
00:05:35,980 --> 00:05:37,060
keep.
198
00:05:37,060 --> 00:05:38,220
The six habits only work
199
00:05:38,220 --> 00:05:39,620
together, and skipping one
200
00:05:39,620 --> 00:05:40,899
usually shows up as a weakness
201
00:05:40,899 --> 00:05:41,660
somewhere else.
202
00:05:41,660 --> 00:05:43,660
If the data is slow, ownership
fades.
203
00:05:43,660 --> 00:05:46,139
If ownership is missing, the
central team turns into a
204
00:05:46,139 --> 00:05:48,139
referee instead of a coach.
205
00:05:48,139 --> 00:05:50,339
How FinOps runs in three
repeating phases.
206
00:05:50,339 --> 00:05:51,579
The practice runs on a loop with
207
00:05:51,579 --> 00:05:53,019
three phases, and you cycle
208
00:05:53,019 --> 00:05:54,220
through them quickly rather than
209
00:05:54,220 --> 00:05:55,399
treating any of them as a one
210
00:05:55,399 --> 00:05:56,449
time project.
211
00:05:56,449 --> 00:05:59,290
The first phase is in form and
its job is visibility.
212
00:05:59,290 --> 00:06:00,610
You gather the cost and usage
213
00:06:00,610 --> 00:06:02,370
data, then split it up so every
214
00:06:02,370 --> 00:06:03,689
team can see its own share of
215
00:06:03,689 --> 00:06:04,170
the bill.
216
00:06:04,170 --> 00:06:05,610
The second phase is optimize,
217
00:06:05,610 --> 00:06:06,850
where you look for ways to get
218
00:06:06,850 --> 00:06:07,970
more from what you already
219
00:06:07,970 --> 00:06:08,490
spend.
220
00:06:08,490 --> 00:06:10,170
There are two flavors of savings
here.
221
00:06:10,170 --> 00:06:11,670
Usage optimization means using
222
00:06:11,670 --> 00:06:12,850
fewer resources to reach the
223
00:06:12,850 --> 00:06:14,250
same result, such as shutting
224
00:06:14,250 --> 00:06:15,410
down a machine that nobody
225
00:06:15,410 --> 00:06:15,889
needs.
226
00:06:15,889 --> 00:06:18,410
Rate optimization means paying a
better price for the resources
227
00:06:18,410 --> 00:06:19,569
you really have to keep.
228
00:06:19,569 --> 00:06:21,250
The third phase is operate, and
229
00:06:21,250 --> 00:06:22,509
this is where talk turns into
230
00:06:22,509 --> 00:06:23,209
action.
231
00:06:23,209 --> 00:06:24,730
You make the changes, measure
232
00:06:24,730 --> 00:06:25,769
what happened, and then loop
233
00:06:25,769 --> 00:06:27,689
right back to inform small,
234
00:06:27,689 --> 00:06:28,529
frequent steps.
235
00:06:28,529 --> 00:06:30,290
Beat one giant cleanup because
236
00:06:30,290 --> 00:06:31,569
the cloud changes faster than
237
00:06:31,569 --> 00:06:32,970
any annual project can keep up
238
00:06:32,970 --> 00:06:33,569
with.
239
00:06:33,569 --> 00:06:36,589
New teams usually start at the
crawl stage where they react to
240
00:06:36,589 --> 00:06:37,930
problems after they show up.
241
00:06:37,930 --> 00:06:40,490
As the habit takes hold, they
move into a walk stage and then
242
00:06:40,490 --> 00:06:41,410
a run stage.
243
00:06:41,410 --> 00:06:44,329
Their cost becomes part of how
they design things from the
244
00:06:44,329 --> 00:06:47,250
beginning, rather than something
they chase afterward.
245
00:06:47,250 --> 00:06:49,129
A whole cast of people makes
this work.
246
00:06:49,129 --> 00:06:50,490
Executives set the priorities.
247
00:06:50,490 --> 00:06:52,290
Engineers own the usage they
create.
248
00:06:52,290 --> 00:06:53,410
Finance keeps the numbers
249
00:06:53,410 --> 00:06:54,725
honest, procurement handles the
250
00:06:54,725 --> 00:06:55,759
contracts, and a FinOps
251
00:06:55,759 --> 00:06:56,800
practitioner ties the whole
252
00:06:56,800 --> 00:06:57,560
thing together.
253
00:06:57,560 --> 00:06:59,959
No single person runs FinOps
alone, which is why the word
254
00:06:59,959 --> 00:07:01,079
culture keeps coming up.
255
00:07:01,079 --> 00:07:02,399
The roles can feel like a lot at
first.
256
00:07:02,399 --> 00:07:03,800
So here is the quick version.
257
00:07:03,800 --> 00:07:05,399
Executives decide what matters.
258
00:07:05,399 --> 00:07:06,720
Engineers decide how to build
259
00:07:06,720 --> 00:07:08,560
it, finance decides how to count
260
00:07:08,560 --> 00:07:09,920
it, and procurement decides what
261
00:07:09,920 --> 00:07:10,720
to sign.
262
00:07:10,720 --> 00:07:12,040
A FinOps practitioner is the
263
00:07:12,040 --> 00:07:13,519
person who keeps all four
264
00:07:13,519 --> 00:07:14,959
talking, which is a job that's
265
00:07:14,959 --> 00:07:15,879
mostly translation.
266
00:07:15,879 --> 00:07:17,120
One small warning here because
267
00:07:17,120 --> 00:07:18,079
the loop has a condition
268
00:07:18,079 --> 00:07:18,879
attached.
269
00:07:18,879 --> 00:07:22,040
The optimize phase only helps
when people can see their own
270
00:07:22,040 --> 00:07:24,639
name on the bill, and that
visibility depends on the
271
00:07:24,639 --> 00:07:27,480
tagging and allocation work from
the inform phase.
272
00:07:27,480 --> 00:07:31,040
Skip the boring setup step and
the clever savings never arrive.
273
00:07:31,040 --> 00:07:33,079
What FinOps looks like on a
normal Tuesday.
274
00:07:33,079 --> 00:07:34,480
Here is what the loop looks like
275
00:07:34,480 --> 00:07:36,160
in real life, and it's far less
276
00:07:36,160 --> 00:07:37,160
dramatic than the theory
277
00:07:37,160 --> 00:07:38,000
suggests.
278
00:07:38,000 --> 00:07:40,800
Imagine a marketing team whose
cloud spending jumps one week
279
00:07:40,800 --> 00:07:42,199
for no obvious reason.
280
00:07:42,199 --> 00:07:44,839
An automatic alert fires and
lands in the team's inbox, and
281
00:07:44,839 --> 00:07:48,040
someone finally asks the obvious
question of what changed because
282
00:07:48,040 --> 00:07:50,720
the resources were tagged, the
answer shows up in minutes.
283
00:07:50,720 --> 00:07:51,759
A test environment that was
284
00:07:51,759 --> 00:07:53,300
supposed to run for a single day
285
00:07:53,300 --> 00:07:54,819
has been quietly charging for
286
00:07:54,819 --> 00:07:56,300
two months, and a five minute
287
00:07:56,300 --> 00:07:57,420
cleanup makes the spike
288
00:07:57,420 --> 00:07:58,500
disappear.
289
00:07:58,500 --> 00:07:59,819
The money saved is small, but
290
00:07:59,819 --> 00:08:01,259
the habit that caught it is the
291
00:08:01,259 --> 00:08:02,300
real prize.
292
00:08:02,300 --> 00:08:04,660
Now the team sees its own number
every month, and that changes
293
00:08:04,660 --> 00:08:05,620
how people behave.
294
00:08:05,620 --> 00:08:08,819
Some companies go one step
further and actually bill each
295
00:08:08,819 --> 00:08:12,220
team for what it uses, a
practice called chargeback.
296
00:08:12,220 --> 00:08:13,459
Others simply show teams their
297
00:08:13,459 --> 00:08:14,879
costs in a process called
298
00:08:14,879 --> 00:08:15,860
showback.
299
00:08:15,860 --> 00:08:18,339
Showback is the gentler starting
point because it builds
300
00:08:18,339 --> 00:08:20,740
awareness before it starts
moving budgets around.
301
00:08:20,740 --> 00:08:23,839
Notice how little of that story
required a new product, a tag,
302
00:08:23,839 --> 00:08:26,660
an alert, and one curious person
did the whole job.
303
00:08:26,660 --> 00:08:29,100
And you could copy that same
pattern on a Monday morning.
304
00:08:29,100 --> 00:08:30,899
The rhythm matters more than the
tooling.
305
00:08:30,899 --> 00:08:32,279
A short monthly review, a couple
306
00:08:32,279 --> 00:08:33,860
of clear decisions, and one
307
00:08:33,860 --> 00:08:35,360
named owner for each action will
308
00:08:35,360 --> 00:08:36,659
beat a giant quarterly audit
309
00:08:36,659 --> 00:08:37,259
every time.
310
00:08:37,259 --> 00:08:39,419
The goal is a habit that
survives busy weeks.
311
00:08:39,419 --> 00:08:41,980
The alert itself is doing quiet
work underneath all of this.
312
00:08:41,980 --> 00:08:44,220
It turns a number nobody was
watching into a question
313
00:08:44,220 --> 00:08:45,500
somebody has to answer.
314
00:08:45,500 --> 00:08:47,860
That small nudge is often the
whole difference between a bill
315
00:08:47,860 --> 00:08:50,860
that stays flat and one that
doubles by the end of the year.
316
00:08:50,860 --> 00:08:53,409
Getting there doesn't need a big
meeting or a new system.
317
00:08:53,409 --> 00:08:56,490
It needs one alert, one owner,
and one honest look at the
318
00:08:56,490 --> 00:08:58,970
number each month, which is
exactly why a small team can
319
00:08:58,970 --> 00:08:59,809
start this week.
320
00:08:59,809 --> 00:09:00,769
The point of showback and
321
00:09:00,769 --> 00:09:02,090
chargeback is never to punish
322
00:09:02,090 --> 00:09:02,809
anyone.
323
00:09:02,809 --> 00:09:05,169
It's to make the cost of a
decision visible to the person
324
00:09:05,169 --> 00:09:08,370
making it so the next choice
comes out a little smarter.
325
00:09:08,370 --> 00:09:09,970
Where FinOps lives in Microsoft
326
00:09:09,970 --> 00:09:10,929
three hundred sixty five and
327
00:09:10,929 --> 00:09:11,409
Azure.
328
00:09:11,409 --> 00:09:14,129
Microsoft makes this easy to
start because the tools are
329
00:09:14,129 --> 00:09:16,610
already sitting inside the
products you use.
330
00:09:16,610 --> 00:09:17,929
The home base is Microsoft Cost
331
00:09:17,929 --> 00:09:19,169
Management, which Microsoft
332
00:09:19,169 --> 00:09:20,649
describes as a suite of FinOps
333
00:09:20,649 --> 00:09:22,129
tools that help you analyze,
334
00:09:22,129 --> 00:09:23,610
monitor, and optimize your cloud
335
00:09:23,610 --> 00:09:24,210
costs.
336
00:09:24,210 --> 00:09:25,389
You can reach it in the Azure
337
00:09:25,389 --> 00:09:26,909
portal and in the Microsoft
338
00:09:26,909 --> 00:09:28,090
three hundred sixty five Admin
339
00:09:28,090 --> 00:09:28,450
Center.
340
00:09:28,450 --> 00:09:31,169
So your cloud bill and your
software licenses finally live
341
00:09:31,169 --> 00:09:32,370
in the same place.
342
00:09:32,370 --> 00:09:35,049
Inside cost management cost
analysis lets you explore where
343
00:09:35,049 --> 00:09:37,929
the money went and budgets, let
you set a limit and get warned
344
00:09:37,929 --> 00:09:38,929
when you cross it.
345
00:09:38,929 --> 00:09:42,450
Anomaly alerts watch for unusual
spikes and scheduled alerts.
346
00:09:42,450 --> 00:09:45,129
Email you a summary on a daily
or weekly rhythm.
347
00:09:45,129 --> 00:09:46,730
Exports push the raw data into
348
00:09:46,730 --> 00:09:48,529
power BI or another system so
349
00:09:48,529 --> 00:09:49,450
you can build your own
350
00:09:49,450 --> 00:09:50,730
dashboards.
351
00:09:50,730 --> 00:09:54,159
Azure advisor is the free helper
that most beginners never open.
352
00:09:54,159 --> 00:09:56,840
Microsoft calls it a
personalized cloud consultant,
353
00:09:56,840 --> 00:09:59,840
and it studies your setup and
recommends changes every day,
354
00:09:59,840 --> 00:10:02,919
from resizing an oversized
machine to shutting down
355
00:10:02,919 --> 00:10:04,399
something that sits idle.
356
00:10:04,399 --> 00:10:05,120
It's free.
357
00:10:05,120 --> 00:10:06,279
It's already switched on, and
358
00:10:06,279 --> 00:10:07,759
it's the closest thing Azure has
359
00:10:07,759 --> 00:10:09,039
to a to do list for saving
360
00:10:09,039 --> 00:10:09,440
money.
361
00:10:09,440 --> 00:10:10,840
Then come the rate levers where
362
00:10:10,840 --> 00:10:12,559
the discounts live, and Azure
363
00:10:12,559 --> 00:10:13,840
Savings Plan can cut compute
364
00:10:13,840 --> 00:10:15,039
costs by up to sixty five
365
00:10:15,039 --> 00:10:16,200
percent from pay as you go
366
00:10:16,200 --> 00:10:17,320
prices when your usage is
367
00:10:17,320 --> 00:10:17,600
steady.
368
00:10:17,600 --> 00:10:18,740
A reservation can save up to
369
00:10:18,740 --> 00:10:19,940
seventy two percent if you
370
00:10:19,940 --> 00:10:21,399
commit to specific usage for a
371
00:10:21,399 --> 00:10:22,080
set term.
372
00:10:22,080 --> 00:10:23,279
Azure Hybrid Benefit goes
373
00:10:23,279 --> 00:10:24,759
further and lets you reuse
374
00:10:24,759 --> 00:10:26,440
Windows Server or SQL server
375
00:10:26,440 --> 00:10:28,080
licenses you already own, which
376
00:10:28,080 --> 00:10:29,480
Microsoft says can save up to
377
00:10:29,480 --> 00:10:30,679
forty percent on virtual
378
00:10:30,679 --> 00:10:31,399
machines.
379
00:10:31,399 --> 00:10:32,559
Treat every one of those numbers
380
00:10:32,559 --> 00:10:33,759
as a ceiling rather than a
381
00:10:33,759 --> 00:10:34,919
promise, because they only pay
382
00:10:34,919 --> 00:10:35,840
off when your usage is
383
00:10:35,840 --> 00:10:36,600
predictable.
384
00:10:36,600 --> 00:10:39,080
One more idea ties all of this
together, and it carries a
385
00:10:39,080 --> 00:10:41,120
slightly awkward name focus.
386
00:10:41,120 --> 00:10:44,000
The FinOps Open Cost and usage
specification is a shared format
387
00:10:44,000 --> 00:10:46,360
for billing data, and Microsoft
is a founding member.
388
00:10:46,360 --> 00:10:47,639
It means a cost report from
389
00:10:47,639 --> 00:10:48,919
Azure, from another cloud, or
390
00:10:48,919 --> 00:10:50,159
from a software vendor can
391
00:10:50,159 --> 00:10:51,274
finally be read in the same
392
00:10:51,274 --> 00:10:51,950
language.
393
00:10:51,950 --> 00:10:52,590
That's a big deal.
394
00:10:52,590 --> 00:10:54,710
The first time you have to
compare them side by side.
395
00:10:54,710 --> 00:10:55,830
And remember that software
396
00:10:55,830 --> 00:10:56,990
licensing belongs in the same
397
00:10:56,990 --> 00:10:57,990
discipline.
398
00:10:57,990 --> 00:10:59,110
A Microsoft three hundred sixty
399
00:10:59,110 --> 00:11:00,309
five copilot license runs about
400
00:11:00,309 --> 00:11:01,509
thirty dollars per user per
401
00:11:01,509 --> 00:11:02,669
month as an add on to a
402
00:11:02,669 --> 00:11:03,470
qualifying plan.
403
00:11:03,470 --> 00:11:06,269
Once you multiply that across a
few hundred people, it deserves
404
00:11:06,269 --> 00:11:09,350
the same scrutiny you would give
any server on the Microsoft
405
00:11:09,350 --> 00:11:10,389
three hundred sixty five side.
406
00:11:10,389 --> 00:11:12,269
The same thinking applies to
licenses.
407
00:11:12,269 --> 00:11:13,309
It's common to find people
408
00:11:13,309 --> 00:11:14,549
holding a premium license they
409
00:11:14,549 --> 00:11:15,909
never use, or two overlapping
410
00:11:15,909 --> 00:11:17,149
subscriptions doing almost the
411
00:11:17,149 --> 00:11:17,710
same job.
412
00:11:17,710 --> 00:11:18,830
Cleaning those up is often the
413
00:11:18,830 --> 00:11:20,110
fastest saving in the whole
414
00:11:20,110 --> 00:11:20,710
tenant.
415
00:11:20,710 --> 00:11:23,230
The Microsoft three hundred
sixty five admin Center shows
416
00:11:23,230 --> 00:11:25,710
who is assigned what, so the
question stops being a mystery
417
00:11:25,710 --> 00:11:26,909
and becomes a task.
418
00:11:26,909 --> 00:11:27,870
Those tools hand you the
419
00:11:27,870 --> 00:11:29,269
visibility and the discounts,
420
00:11:29,269 --> 00:11:30,509
and that part is available to
421
00:11:30,509 --> 00:11:31,110
you today.
422
00:11:31,110 --> 00:11:33,909
What no tool can hand you is the
discipline to keep using them.
423
00:11:33,909 --> 00:11:37,190
Which brings us to the mistakes
that sink almost every practice.
424
00:11:37,190 --> 00:11:39,750
The mistakes that sink a FinOps
practice.
425
00:11:39,750 --> 00:11:40,889
The first mistake is treating
426
00:11:40,889 --> 00:11:42,470
FinOps as a pure, cost cutting
427
00:11:42,470 --> 00:11:43,269
exercise.
428
00:11:43,269 --> 00:11:44,230
When savings are the only
429
00:11:44,230 --> 00:11:45,769
scoreboard, teams learn to hide
430
00:11:45,769 --> 00:11:46,950
their spending instead of
431
00:11:46,950 --> 00:11:47,950
talking about it, and the
432
00:11:47,950 --> 00:11:49,049
numbers get worse rather than
433
00:11:49,049 --> 00:11:49,669
better.
434
00:11:49,669 --> 00:11:50,705
The second is running the
435
00:11:50,705 --> 00:11:52,059
practice without ownership or
436
00:11:52,059 --> 00:11:52,460
tags.
437
00:11:52,460 --> 00:11:54,820
If nobody can say which project
caused a charge.
438
00:11:54,820 --> 00:11:57,779
Every conversation about cost
turns into a guessing game and
439
00:11:57,779 --> 00:11:59,139
the fix never sticks for long.
440
00:11:59,139 --> 00:12:01,980
The third mistake is buying a
tool and calling the job done.
441
00:12:01,980 --> 00:12:02,860
Software can show you the
442
00:12:02,860 --> 00:12:04,100
numbers, but it can't repair a
443
00:12:04,100 --> 00:12:05,700
conversation that nobody wants
444
00:12:05,700 --> 00:12:06,100
to have.
445
00:12:06,100 --> 00:12:09,059
The fourth is keeping all the
work inside one central team.
446
00:12:09,059 --> 00:12:10,460
If engineers never see their own
447
00:12:10,460 --> 00:12:11,899
costs, they never build the
448
00:12:11,899 --> 00:12:13,299
instinct to design with cost in
449
00:12:13,299 --> 00:12:14,620
mind, and the central team
450
00:12:14,620 --> 00:12:16,220
becomes a bottleneck instead of
451
00:12:16,220 --> 00:12:17,179
an enabler.
452
00:12:17,179 --> 00:12:19,860
The fifth mistake is chasing the
last few percent of savings
453
00:12:19,860 --> 00:12:21,539
while ignoring the big calls.
454
00:12:21,539 --> 00:12:24,279
The foundation's twenty twenty
six report captured this well,
455
00:12:24,279 --> 00:12:27,620
quoting a practitioner who said
the big rocks of waste are gone.
456
00:12:27,620 --> 00:12:30,179
What remains, in their words, is
a high volume of smaller
457
00:12:30,179 --> 00:12:32,379
opportunities that take far more
effort to capture.
458
00:12:32,379 --> 00:12:33,620
The sixth is trying to measure
459
00:12:33,620 --> 00:12:34,940
value without agreeing on what
460
00:12:34,940 --> 00:12:36,059
value means.
461
00:12:36,059 --> 00:12:37,179
Teams end up arguing about
462
00:12:37,179 --> 00:12:38,299
whether a tool is worth the
463
00:12:38,299 --> 00:12:40,059
money when they never define the
464
00:12:40,059 --> 00:12:41,179
outcome they were buying in the
465
00:12:41,179 --> 00:12:42,059
first place.
466
00:12:42,059 --> 00:12:44,220
Mistakes like these rarely
announce themselves.
467
00:12:44,220 --> 00:12:45,740
They show up as a slow drift.
468
00:12:45,740 --> 00:12:48,580
The dashboards look fine, the
meetings keep happening, and the
469
00:12:48,580 --> 00:12:52,000
bill still climbs because nobody
changed the behavior behind it.
470
00:12:52,000 --> 00:12:53,159
The good news is that the same
471
00:12:53,159 --> 00:12:54,600
habits from earlier fix most of
472
00:12:54,600 --> 00:12:56,000
them, which means the cure is
473
00:12:56,000 --> 00:12:56,960
usually simpler than the
474
00:12:56,960 --> 00:12:57,720
disease.
475
00:12:57,720 --> 00:13:00,759
What FinOps can't fix FinOps is
honest about its limits, and you
476
00:13:00,759 --> 00:13:01,720
should be, too.
477
00:13:01,720 --> 00:13:02,879
It can't yet tell you whether
478
00:13:02,879 --> 00:13:04,200
your AI investment is paying
479
00:13:04,200 --> 00:13:05,759
off, because the whole industry
480
00:13:05,759 --> 00:13:06,840
is still wrestling with that
481
00:13:06,840 --> 00:13:07,519
question.
482
00:13:07,519 --> 00:13:10,000
The twenty twenty six survey put
it bluntly, and it shared a
483
00:13:10,000 --> 00:13:12,559
practitioner's line that nobody
can answer whether AI is
484
00:13:12,559 --> 00:13:13,799
providing value yet.
485
00:13:13,799 --> 00:13:16,440
Measuring the money you saved by
never making a mistake is just
486
00:13:16,440 --> 00:13:18,320
as hard as another practitioner.
487
00:13:18,320 --> 00:13:21,200
In that same report, put it once
you fix a problem, it's gone.
488
00:13:21,200 --> 00:13:23,000
So how do you give a developer
489
00:13:23,000 --> 00:13:24,200
credit for a cost that never
490
00:13:24,200 --> 00:13:24,600
appeared?
491
00:13:24,600 --> 00:13:27,440
It can't mend bad architecture
or unclear ownership either,
492
00:13:27,440 --> 00:13:30,360
because those are design and
culture problems that a report
493
00:13:30,360 --> 00:13:31,679
will only expose.
494
00:13:31,679 --> 00:13:34,600
It also can't promise that the
easy wins are still waiting,
495
00:13:34,600 --> 00:13:37,240
since mature teams report that
the obvious waste has largely
496
00:13:37,240 --> 00:13:40,159
been picked over, and it can't
turn a commitment discount into
497
00:13:40,159 --> 00:13:42,960
free money, a reservation or a
savings plan only helps when
498
00:13:42,960 --> 00:13:45,960
your usage is steady, so buying
one for a workload you plan to
499
00:13:45,960 --> 00:13:48,080
shut down can lose more than it
saves the cost.
500
00:13:48,080 --> 00:13:50,990
Data itself also arrives after
the fact, which is worth sitting
501
00:13:50,990 --> 00:13:51,870
with for a second.
502
00:13:51,870 --> 00:13:52,590
You're always reading
503
00:13:52,590 --> 00:13:54,169
yesterday's numbers, so FinOps
504
00:13:54,169 --> 00:13:55,629
is a practice of steering, not a
505
00:13:55,629 --> 00:13:55,870
switch.
506
00:13:55,870 --> 00:13:57,149
You flip once and forget.
507
00:13:57,149 --> 00:13:59,629
The harder truth is that FinOps
asks for patience.
508
00:13:59,629 --> 00:14:02,190
The first pass usually finds a
few obvious wins, and the second
509
00:14:02,190 --> 00:14:04,789
pass finds fewer, which is
normal rather than a sign that
510
00:14:04,789 --> 00:14:06,190
the practice stopped working.
511
00:14:06,190 --> 00:14:07,750
None of that makes the practice
weak.
512
00:14:07,750 --> 00:14:09,110
It just means FinOps is a way of
513
00:14:09,110 --> 00:14:10,169
asking better questions about
514
00:14:10,169 --> 00:14:11,710
technology spending, not a
515
00:14:11,710 --> 00:14:12,909
machine that prints savings on
516
00:14:12,909 --> 00:14:13,269
demand.
517
00:14:13,269 --> 00:14:16,029
Knowing what it can't do is what
keeps you from expecting magic,
518
00:14:16,029 --> 00:14:19,190
and then giving up when the
magic never shows.
519
00:14:19,190 --> 00:14:20,830
What to remember about FinOps.
520
00:14:20,830 --> 00:14:22,950
So here is the short version to
carry with you.
521
00:14:22,950 --> 00:14:24,470
FinOps is a shared habit of
522
00:14:24,470 --> 00:14:26,549
watching what technology costs
523
00:14:26,549 --> 00:14:28,350
and deciding together what it is
524
00:14:28,350 --> 00:14:28,870
worth.
525
00:14:28,870 --> 00:14:31,789
It runs on a simple loop of
inform, optimize, and operate.
526
00:14:31,789 --> 00:14:33,289
Microsoft hands you the starting
527
00:14:33,289 --> 00:14:34,750
tools inside the products you
528
00:14:34,750 --> 00:14:35,629
already use.
529
00:14:35,629 --> 00:14:38,029
Now for one question, and I
really do want your answer.
530
00:14:38,029 --> 00:14:40,590
When was the last time someone
on your team asked who owns the
531
00:14:40,590 --> 00:14:43,669
cloud bill if this made the
topic feel less intimidating?
532
00:14:43,669 --> 00:14:47,669
Subscribe to M365 FM to master
the tools and the mindset
533
00:14:47,669 --> 00:14:49,350
shaping the future of work.
00:00:00,040 --> 00:00:03,520
Welcome to the M365 FM podcast,
where we explore the people,
2
00:00:03,520 --> 00:00:06,240
ideas and technologies shaping
the future of work.
3
00:00:06,240 --> 00:00:08,240
Why cloud costs feel out of
control.
4
00:00:08,240 --> 00:00:11,080
Most companies find out they're
overspending on cloud when the
5
00:00:11,080 --> 00:00:14,000
invoice lands, and by then the
money is already gone.
6
00:00:14,000 --> 00:00:18,000
Welcome to the M365 FM podcast,
where we explore the people,
7
00:00:18,000 --> 00:00:21,039
ideas and technologies shaping
the future of work.
8
00:00:21,039 --> 00:00:23,399
FinOps sounds like another piece
of jargon, but it's really the
9
00:00:23,399 --> 00:00:26,039
missing conversation between the
people who build with technology
10
00:00:26,039 --> 00:00:27,600
and the people who pay for it.
11
00:00:27,600 --> 00:00:29,039
Cloud bills grow quietly, one
12
00:00:29,039 --> 00:00:30,399
small decision at a time, and
13
00:00:30,399 --> 00:00:32,000
nobody notices until someone
14
00:00:32,000 --> 00:00:33,200
asks who actually owns the
15
00:00:33,200 --> 00:00:33,719
number.
16
00:00:33,719 --> 00:00:34,799
By the end of this episode,
17
00:00:34,799 --> 00:00:36,060
you'll understand what FinOps
18
00:00:36,060 --> 00:00:38,079
is, how it works, where it fits
19
00:00:38,079 --> 00:00:39,520
in Microsoft three hundred sixty
20
00:00:39,520 --> 00:00:40,759
five and Azure, and the mistakes
21
00:00:40,759 --> 00:00:42,060
that trip up almost everyone who
22
00:00:42,060 --> 00:00:42,759
tries it.
23
00:00:42,759 --> 00:00:45,399
Cloud computing changed how
companies pay for technology,
24
00:00:45,399 --> 00:00:47,479
and that shift sits underneath
the whole problem.
25
00:00:47,479 --> 00:00:50,240
Years ago, you bought a server
once and spread the cost across
26
00:00:50,240 --> 00:00:52,560
several years, so you knew the
number in advance.
27
00:00:52,560 --> 00:00:54,439
Today you rent computing by the
second.
28
00:00:54,439 --> 00:00:57,079
The bill rewrites itself every
day and no single person feels
29
00:00:57,079 --> 00:00:58,240
responsible for it.
30
00:00:58,240 --> 00:01:00,950
Picture a developer who spins up
a virtual machine on a Tuesday
31
00:01:00,950 --> 00:01:02,710
afternoon to test one idea.
32
00:01:02,710 --> 00:01:04,030
A meeting pulls them away.
33
00:01:04,030 --> 00:01:05,510
The test is forgotten, and that
34
00:01:05,510 --> 00:01:06,709
machine keeps running and
35
00:01:06,709 --> 00:01:07,230
charging.
36
00:01:07,230 --> 00:01:10,109
For months, nobody chose to
waste that money, and that quiet
37
00:01:10,109 --> 00:01:12,870
accident is exactly what makes
the league so easy to miss.
38
00:01:12,870 --> 00:01:15,189
The split shows up in the
language people use.
39
00:01:15,189 --> 00:01:17,950
Finance asks for a forecast it
can trust and engineering
40
00:01:17,950 --> 00:01:20,790
answers with a diagram of
services that scale on demand.
41
00:01:20,790 --> 00:01:21,989
Neither side is wrong and
42
00:01:21,989 --> 00:01:23,069
neither side can see the whole
43
00:01:23,069 --> 00:01:23,909
picture alone.
44
00:01:23,909 --> 00:01:24,950
Which is why the bill keeps
45
00:01:24,950 --> 00:01:25,950
growing while everyone feels
46
00:01:25,950 --> 00:01:26,510
busy.
47
00:01:26,510 --> 00:01:28,069
So the money drains away in
48
00:01:28,069 --> 00:01:29,989
dozens of tiny moments in almost
49
00:01:29,989 --> 00:01:31,349
every company that touches the
50
00:01:31,349 --> 00:01:31,950
cloud.
51
00:01:31,950 --> 00:01:34,510
That leak has a name, and more
importantly, it has a practice
52
00:01:34,510 --> 00:01:35,709
built to stop it.
53
00:01:35,709 --> 00:01:37,909
What FinOps actually means.
54
00:01:37,909 --> 00:01:39,230
Start with the name because it
55
00:01:39,230 --> 00:01:40,230
gives away more than you would
56
00:01:40,230 --> 00:01:40,870
expect.
57
00:01:40,870 --> 00:01:43,670
FinOps is a mashup of finance
and DevOps, and that pairing is
58
00:01:43,670 --> 00:01:44,829
the entire idea.
59
00:01:44,829 --> 00:01:46,950
It describes the dialogue
between the builders and the
60
00:01:46,950 --> 00:01:49,790
payers, which sounds obvious
until you notice how rarely
61
00:01:49,790 --> 00:01:51,430
those two groups actually talk.
62
00:01:51,430 --> 00:01:53,629
Here's the part that surprises
most beginners.
63
00:01:53,629 --> 00:01:56,390
FinOps isn't a product you buy,
and it isn't a single job title
64
00:01:56,390 --> 00:01:57,629
you can hire for.
65
00:01:57,629 --> 00:01:58,965
The FinOps Foundation, the
66
00:01:58,965 --> 00:01:59,980
nonprofit that writes the
67
00:01:59,980 --> 00:02:01,140
playbook, calls it an
68
00:02:01,140 --> 00:02:02,379
operational framework and a
69
00:02:02,379 --> 00:02:03,219
cultural practice.
70
00:02:03,219 --> 00:02:06,219
That's a formal way of saying
it's a shared habit rather than
71
00:02:06,219 --> 00:02:07,379
a piece of software.
72
00:02:07,379 --> 00:02:09,060
The goal isn't simply to shrink
the bill.
73
00:02:09,060 --> 00:02:10,500
The foundation says it plainly
74
00:02:10,500 --> 00:02:11,740
if you think FinOps is about
75
00:02:11,740 --> 00:02:13,259
saving money, think again,
76
00:02:13,259 --> 00:02:14,180
because it's really about
77
00:02:14,180 --> 00:02:15,259
getting the most value out of
78
00:02:15,259 --> 00:02:16,020
technology.
79
00:02:16,020 --> 00:02:18,419
Sometimes the smart move is to
spend more on something that
80
00:02:18,419 --> 00:02:21,020
earns more, and the practice
gives you the numbers to make
81
00:02:21,020 --> 00:02:22,379
that call with confidence.
82
00:02:22,379 --> 00:02:25,259
You'll also hear the same idea
called cloud cost management or
83
00:02:25,259 --> 00:02:26,740
cloud financial management.
84
00:02:26,740 --> 00:02:29,699
Those labels describe a task,
while FinOps describes a culture
85
00:02:29,699 --> 00:02:31,020
that spreads across teams.
86
00:02:31,020 --> 00:02:33,419
And that difference matters more
than it sounds.
87
00:02:33,419 --> 00:02:34,780
Picture the difference in
practice.
88
00:02:34,780 --> 00:02:37,300
A team wants to run an analytics
job every night.
89
00:02:37,300 --> 00:02:38,939
The cheap option takes six hours
90
00:02:38,939 --> 00:02:40,180
on a slow machine, and the
91
00:02:40,180 --> 00:02:41,379
pricier option finishes in
92
00:02:41,379 --> 00:02:42,180
twenty minutes.
93
00:02:42,180 --> 00:02:44,780
FinOps doesn't automatically
pick the cheaper one.
94
00:02:44,780 --> 00:02:46,139
It asks what those saved hours
95
00:02:46,139 --> 00:02:47,479
are worth, then makes the trade
96
00:02:47,479 --> 00:02:48,900
off on purpose instead of by
97
00:02:48,900 --> 00:02:49,620
accident.
98
00:02:49,620 --> 00:02:50,780
That habit of deciding on
99
00:02:50,780 --> 00:02:51,740
purpose is the heart of the
100
00:02:51,740 --> 00:02:52,419
whole thing.
101
00:02:52,419 --> 00:02:54,900
It's the reason the practice
belongs to the company rather
102
00:02:54,900 --> 00:02:55,979
than to one department.
103
00:02:55,979 --> 00:02:58,759
Hold on to that idea because it
explains something strange.
104
00:02:58,759 --> 00:02:59,800
You'll see later.
105
00:02:59,800 --> 00:03:01,159
The biggest savings come from a
106
00:03:01,159 --> 00:03:02,639
decision about people rather
107
00:03:02,639 --> 00:03:03,500
than a decision about
108
00:03:03,500 --> 00:03:04,400
technology.
109
00:03:04,400 --> 00:03:05,719
The practice grew out of a real
110
00:03:05,719 --> 00:03:06,919
shift in how technology gets
111
00:03:06,919 --> 00:03:07,439
funded.
112
00:03:07,439 --> 00:03:08,719
Companies used to treat it as
113
00:03:08,719 --> 00:03:10,120
one big yearly purchase,
114
00:03:10,120 --> 00:03:11,159
something accountants call
115
00:03:11,159 --> 00:03:12,479
capital expense.
116
00:03:12,479 --> 00:03:13,879
The cloud flipped that into a
117
00:03:13,879 --> 00:03:15,000
running cost that changes
118
00:03:15,000 --> 00:03:16,560
constantly, which left finance
119
00:03:16,560 --> 00:03:17,939
teams with no fixed number to
120
00:03:17,939 --> 00:03:18,840
plan around.
121
00:03:18,840 --> 00:03:20,159
FinOps grew up to bridge that
122
00:03:20,159 --> 00:03:21,840
gap, and it has since stretched
123
00:03:21,840 --> 00:03:23,479
far past the cloud into
124
00:03:23,479 --> 00:03:24,719
software, subscriptions,
125
00:03:24,719 --> 00:03:26,819
licenses, data centers, and now
126
00:03:26,819 --> 00:03:27,520
AI.
127
00:03:27,520 --> 00:03:30,520
Today, the numbers show how far
it has spread in the FinOps
128
00:03:30,520 --> 00:03:31,919
Foundation's latest survey.
129
00:03:31,919 --> 00:03:33,400
Ninety eight percent of teams
130
00:03:33,400 --> 00:03:35,039
said they manage AI spending in
131
00:03:35,039 --> 00:03:35,599
some form.
132
00:03:35,599 --> 00:03:38,719
Ninety percent manage or plan to
manage software subscriptions.
133
00:03:38,719 --> 00:03:41,819
What began as a cloud cost club
now touches almost every kind of
134
00:03:41,819 --> 00:03:43,360
technology a company buys.
135
00:03:43,360 --> 00:03:45,280
None of this requires an
engineering degree.
136
00:03:45,280 --> 00:03:48,159
The practice is built for the
person who is new to all of it.
137
00:03:48,159 --> 00:03:49,599
The first step is simply asking
138
00:03:49,599 --> 00:03:51,080
where the money goes and who is
139
00:03:51,080 --> 00:03:52,080
watching it.
140
00:03:52,080 --> 00:03:54,759
The six habits every FinOps team
shares.
141
00:03:54,759 --> 00:03:57,280
The foundation lays out six
principles, and they make the
142
00:03:57,280 --> 00:04:00,509
most sense when you read them as
habits rather than rules.
143
00:04:00,509 --> 00:04:03,229
The first is that teams have to
collaborate, which means
144
00:04:03,229 --> 00:04:06,430
finance, engineering and the
business all look at the same
145
00:04:06,430 --> 00:04:09,110
numbers instead of arguing
across a wall.
146
00:04:09,110 --> 00:04:10,189
The second habit is that
147
00:04:10,189 --> 00:04:11,710
business value drives the
148
00:04:11,710 --> 00:04:12,229
decisions.
149
00:04:12,229 --> 00:04:14,389
You aren't chasing the lowest
possible number, because a
150
00:04:14,389 --> 00:04:17,110
service that is cheap and breaks
every week isn't cheap at all.
151
00:04:17,110 --> 00:04:18,589
The third is that everyone owns
152
00:04:18,589 --> 00:04:19,829
their own usage, so the person
153
00:04:19,829 --> 00:04:20,870
who built a resource is the
154
00:04:20,870 --> 00:04:21,910
person who watches what it
155
00:04:21,910 --> 00:04:22,629
costs.
156
00:04:22,629 --> 00:04:24,389
The fourth habit is about the
data itself.
157
00:04:24,389 --> 00:04:26,910
Cost information has to be
timely, accurate and available
158
00:04:26,910 --> 00:04:27,910
to the people who need it.
159
00:04:27,910 --> 00:04:29,110
A report that shows up three
160
00:04:29,110 --> 00:04:30,310
weeks late can't change a
161
00:04:30,310 --> 00:04:31,509
decision that's already been
162
00:04:31,509 --> 00:04:32,189
made.
163
00:04:32,189 --> 00:04:34,829
The fifth is that a small
central team enables the whole
164
00:04:34,829 --> 00:04:38,189
practice, sharing best practices
and holding the standards.
165
00:04:38,189 --> 00:04:40,589
The real work stays close to the
teams doing it.
166
00:04:40,589 --> 00:04:43,069
The sixth habit is the one
people forget, and it's about
167
00:04:43,069 --> 00:04:45,310
using the flexibility you're
already paying for.
168
00:04:45,310 --> 00:04:46,470
The cloud lets you shrink when
169
00:04:46,470 --> 00:04:47,589
demand drops and grow when
170
00:04:47,589 --> 00:04:48,509
demand spikes.
171
00:04:48,509 --> 00:04:51,269
So the goal is to match what you
pay to what you actually use.
172
00:04:51,269 --> 00:04:52,709
That blend of a central core and
173
00:04:52,709 --> 00:04:54,230
local ownership is the norm, not
174
00:04:54,230 --> 00:04:55,069
the exception.
175
00:04:55,069 --> 00:04:57,660
In the Foundation's twenty
twenty six survey, eighty one
176
00:04:57,660 --> 00:05:00,319
percent of teams ran either a
centralized enablement model or
177
00:05:00,319 --> 00:05:01,300
a hub and spoke model.
178
00:05:01,300 --> 00:05:03,300
That's a formal way of
describing a small core group
179
00:05:03,300 --> 00:05:04,620
that coaches everyone else.
180
00:05:04,620 --> 00:05:07,100
That doesn't mean the central
team does the work itself.
181
00:05:07,100 --> 00:05:09,420
It means the central team draws
the map and hands out the
182
00:05:09,420 --> 00:05:11,839
compasses, because one hundred
people watching their own costs
183
00:05:11,839 --> 00:05:14,339
will always beat five people
watching everyone else's.
184
00:05:14,339 --> 00:05:15,500
The ownership habit sounds
185
00:05:15,500 --> 00:05:16,819
abstract until you watch at
186
00:05:16,819 --> 00:05:17,579
work.
187
00:05:17,579 --> 00:05:20,300
A team that tags its resources
and checks its own monthly
188
00:05:20,300 --> 00:05:23,220
number starts choosing smaller
machines on its own.
189
00:05:23,220 --> 00:05:26,060
Nobody sends a memo because the
cost finally belongs to a person
190
00:05:26,060 --> 00:05:27,300
instead of a spreadsheet.
191
00:05:27,300 --> 00:05:28,420
The collaboration habit works
192
00:05:28,420 --> 00:05:30,060
the same way when finance and
193
00:05:30,060 --> 00:05:31,819
engineering share one dashboard,
194
00:05:31,819 --> 00:05:33,040
the argument about the bill
195
00:05:33,040 --> 00:05:34,540
becomes a shared question about
196
00:05:34,540 --> 00:05:35,980
which services are earning their
197
00:05:35,980 --> 00:05:37,060
keep.
198
00:05:37,060 --> 00:05:38,220
The six habits only work
199
00:05:38,220 --> 00:05:39,620
together, and skipping one
200
00:05:39,620 --> 00:05:40,899
usually shows up as a weakness
201
00:05:40,899 --> 00:05:41,660
somewhere else.
202
00:05:41,660 --> 00:05:43,660
If the data is slow, ownership
fades.
203
00:05:43,660 --> 00:05:46,139
If ownership is missing, the
central team turns into a
204
00:05:46,139 --> 00:05:48,139
referee instead of a coach.
205
00:05:48,139 --> 00:05:50,339
How FinOps runs in three
repeating phases.
206
00:05:50,339 --> 00:05:51,579
The practice runs on a loop with
207
00:05:51,579 --> 00:05:53,019
three phases, and you cycle
208
00:05:53,019 --> 00:05:54,220
through them quickly rather than
209
00:05:54,220 --> 00:05:55,399
treating any of them as a one
210
00:05:55,399 --> 00:05:56,449
time project.
211
00:05:56,449 --> 00:05:59,290
The first phase is in form and
its job is visibility.
212
00:05:59,290 --> 00:06:00,610
You gather the cost and usage
213
00:06:00,610 --> 00:06:02,370
data, then split it up so every
214
00:06:02,370 --> 00:06:03,689
team can see its own share of
215
00:06:03,689 --> 00:06:04,170
the bill.
216
00:06:04,170 --> 00:06:05,610
The second phase is optimize,
217
00:06:05,610 --> 00:06:06,850
where you look for ways to get
218
00:06:06,850 --> 00:06:07,970
more from what you already
219
00:06:07,970 --> 00:06:08,490
spend.
220
00:06:08,490 --> 00:06:10,170
There are two flavors of savings
here.
221
00:06:10,170 --> 00:06:11,670
Usage optimization means using
222
00:06:11,670 --> 00:06:12,850
fewer resources to reach the
223
00:06:12,850 --> 00:06:14,250
same result, such as shutting
224
00:06:14,250 --> 00:06:15,410
down a machine that nobody
225
00:06:15,410 --> 00:06:15,889
needs.
226
00:06:15,889 --> 00:06:18,410
Rate optimization means paying a
better price for the resources
227
00:06:18,410 --> 00:06:19,569
you really have to keep.
228
00:06:19,569 --> 00:06:21,250
The third phase is operate, and
229
00:06:21,250 --> 00:06:22,509
this is where talk turns into
230
00:06:22,509 --> 00:06:23,209
action.
231
00:06:23,209 --> 00:06:24,730
You make the changes, measure
232
00:06:24,730 --> 00:06:25,769
what happened, and then loop
233
00:06:25,769 --> 00:06:27,689
right back to inform small,
234
00:06:27,689 --> 00:06:28,529
frequent steps.
235
00:06:28,529 --> 00:06:30,290
Beat one giant cleanup because
236
00:06:30,290 --> 00:06:31,569
the cloud changes faster than
237
00:06:31,569 --> 00:06:32,970
any annual project can keep up
238
00:06:32,970 --> 00:06:33,569
with.
239
00:06:33,569 --> 00:06:36,589
New teams usually start at the
crawl stage where they react to
240
00:06:36,589 --> 00:06:37,930
problems after they show up.
241
00:06:37,930 --> 00:06:40,490
As the habit takes hold, they
move into a walk stage and then
242
00:06:40,490 --> 00:06:41,410
a run stage.
243
00:06:41,410 --> 00:06:44,329
Their cost becomes part of how
they design things from the
244
00:06:44,329 --> 00:06:47,250
beginning, rather than something
they chase afterward.
245
00:06:47,250 --> 00:06:49,129
A whole cast of people makes
this work.
246
00:06:49,129 --> 00:06:50,490
Executives set the priorities.
247
00:06:50,490 --> 00:06:52,290
Engineers own the usage they
create.
248
00:06:52,290 --> 00:06:53,410
Finance keeps the numbers
249
00:06:53,410 --> 00:06:54,725
honest, procurement handles the
250
00:06:54,725 --> 00:06:55,759
contracts, and a FinOps
251
00:06:55,759 --> 00:06:56,800
practitioner ties the whole
252
00:06:56,800 --> 00:06:57,560
thing together.
253
00:06:57,560 --> 00:06:59,959
No single person runs FinOps
alone, which is why the word
254
00:06:59,959 --> 00:07:01,079
culture keeps coming up.
255
00:07:01,079 --> 00:07:02,399
The roles can feel like a lot at
first.
256
00:07:02,399 --> 00:07:03,800
So here is the quick version.
257
00:07:03,800 --> 00:07:05,399
Executives decide what matters.
258
00:07:05,399 --> 00:07:06,720
Engineers decide how to build
259
00:07:06,720 --> 00:07:08,560
it, finance decides how to count
260
00:07:08,560 --> 00:07:09,920
it, and procurement decides what
261
00:07:09,920 --> 00:07:10,720
to sign.
262
00:07:10,720 --> 00:07:12,040
A FinOps practitioner is the
263
00:07:12,040 --> 00:07:13,519
person who keeps all four
264
00:07:13,519 --> 00:07:14,959
talking, which is a job that's
265
00:07:14,959 --> 00:07:15,879
mostly translation.
266
00:07:15,879 --> 00:07:17,120
One small warning here because
267
00:07:17,120 --> 00:07:18,079
the loop has a condition
268
00:07:18,079 --> 00:07:18,879
attached.
269
00:07:18,879 --> 00:07:22,040
The optimize phase only helps
when people can see their own
270
00:07:22,040 --> 00:07:24,639
name on the bill, and that
visibility depends on the
271
00:07:24,639 --> 00:07:27,480
tagging and allocation work from
the inform phase.
272
00:07:27,480 --> 00:07:31,040
Skip the boring setup step and
the clever savings never arrive.
273
00:07:31,040 --> 00:07:33,079
What FinOps looks like on a
normal Tuesday.
274
00:07:33,079 --> 00:07:34,480
Here is what the loop looks like
275
00:07:34,480 --> 00:07:36,160
in real life, and it's far less
276
00:07:36,160 --> 00:07:37,160
dramatic than the theory
277
00:07:37,160 --> 00:07:38,000
suggests.
278
00:07:38,000 --> 00:07:40,800
Imagine a marketing team whose
cloud spending jumps one week
279
00:07:40,800 --> 00:07:42,199
for no obvious reason.
280
00:07:42,199 --> 00:07:44,839
An automatic alert fires and
lands in the team's inbox, and
281
00:07:44,839 --> 00:07:48,040
someone finally asks the obvious
question of what changed because
282
00:07:48,040 --> 00:07:50,720
the resources were tagged, the
answer shows up in minutes.
283
00:07:50,720 --> 00:07:51,759
A test environment that was
284
00:07:51,759 --> 00:07:53,300
supposed to run for a single day
285
00:07:53,300 --> 00:07:54,819
has been quietly charging for
286
00:07:54,819 --> 00:07:56,300
two months, and a five minute
287
00:07:56,300 --> 00:07:57,420
cleanup makes the spike
288
00:07:57,420 --> 00:07:58,500
disappear.
289
00:07:58,500 --> 00:07:59,819
The money saved is small, but
290
00:07:59,819 --> 00:08:01,259
the habit that caught it is the
291
00:08:01,259 --> 00:08:02,300
real prize.
292
00:08:02,300 --> 00:08:04,660
Now the team sees its own number
every month, and that changes
293
00:08:04,660 --> 00:08:05,620
how people behave.
294
00:08:05,620 --> 00:08:08,819
Some companies go one step
further and actually bill each
295
00:08:08,819 --> 00:08:12,220
team for what it uses, a
practice called chargeback.
296
00:08:12,220 --> 00:08:13,459
Others simply show teams their
297
00:08:13,459 --> 00:08:14,879
costs in a process called
298
00:08:14,879 --> 00:08:15,860
showback.
299
00:08:15,860 --> 00:08:18,339
Showback is the gentler starting
point because it builds
300
00:08:18,339 --> 00:08:20,740
awareness before it starts
moving budgets around.
301
00:08:20,740 --> 00:08:23,839
Notice how little of that story
required a new product, a tag,
302
00:08:23,839 --> 00:08:26,660
an alert, and one curious person
did the whole job.
303
00:08:26,660 --> 00:08:29,100
And you could copy that same
pattern on a Monday morning.
304
00:08:29,100 --> 00:08:30,899
The rhythm matters more than the
tooling.
305
00:08:30,899 --> 00:08:32,279
A short monthly review, a couple
306
00:08:32,279 --> 00:08:33,860
of clear decisions, and one
307
00:08:33,860 --> 00:08:35,360
named owner for each action will
308
00:08:35,360 --> 00:08:36,659
beat a giant quarterly audit
309
00:08:36,659 --> 00:08:37,259
every time.
310
00:08:37,259 --> 00:08:39,419
The goal is a habit that
survives busy weeks.
311
00:08:39,419 --> 00:08:41,980
The alert itself is doing quiet
work underneath all of this.
312
00:08:41,980 --> 00:08:44,220
It turns a number nobody was
watching into a question
313
00:08:44,220 --> 00:08:45,500
somebody has to answer.
314
00:08:45,500 --> 00:08:47,860
That small nudge is often the
whole difference between a bill
315
00:08:47,860 --> 00:08:50,860
that stays flat and one that
doubles by the end of the year.
316
00:08:50,860 --> 00:08:53,409
Getting there doesn't need a big
meeting or a new system.
317
00:08:53,409 --> 00:08:56,490
It needs one alert, one owner,
and one honest look at the
318
00:08:56,490 --> 00:08:58,970
number each month, which is
exactly why a small team can
319
00:08:58,970 --> 00:08:59,809
start this week.
320
00:08:59,809 --> 00:09:00,769
The point of showback and
321
00:09:00,769 --> 00:09:02,090
chargeback is never to punish
322
00:09:02,090 --> 00:09:02,809
anyone.
323
00:09:02,809 --> 00:09:05,169
It's to make the cost of a
decision visible to the person
324
00:09:05,169 --> 00:09:08,370
making it so the next choice
comes out a little smarter.
325
00:09:08,370 --> 00:09:09,970
Where FinOps lives in Microsoft
326
00:09:09,970 --> 00:09:10,929
three hundred sixty five and
327
00:09:10,929 --> 00:09:11,409
Azure.
328
00:09:11,409 --> 00:09:14,129
Microsoft makes this easy to
start because the tools are
329
00:09:14,129 --> 00:09:16,610
already sitting inside the
products you use.
330
00:09:16,610 --> 00:09:17,929
The home base is Microsoft Cost
331
00:09:17,929 --> 00:09:19,169
Management, which Microsoft
332
00:09:19,169 --> 00:09:20,649
describes as a suite of FinOps
333
00:09:20,649 --> 00:09:22,129
tools that help you analyze,
334
00:09:22,129 --> 00:09:23,610
monitor, and optimize your cloud
335
00:09:23,610 --> 00:09:24,210
costs.
336
00:09:24,210 --> 00:09:25,389
You can reach it in the Azure
337
00:09:25,389 --> 00:09:26,909
portal and in the Microsoft
338
00:09:26,909 --> 00:09:28,090
three hundred sixty five Admin
339
00:09:28,090 --> 00:09:28,450
Center.
340
00:09:28,450 --> 00:09:31,169
So your cloud bill and your
software licenses finally live
341
00:09:31,169 --> 00:09:32,370
in the same place.
342
00:09:32,370 --> 00:09:35,049
Inside cost management cost
analysis lets you explore where
343
00:09:35,049 --> 00:09:37,929
the money went and budgets, let
you set a limit and get warned
344
00:09:37,929 --> 00:09:38,929
when you cross it.
345
00:09:38,929 --> 00:09:42,450
Anomaly alerts watch for unusual
spikes and scheduled alerts.
346
00:09:42,450 --> 00:09:45,129
Email you a summary on a daily
or weekly rhythm.
347
00:09:45,129 --> 00:09:46,730
Exports push the raw data into
348
00:09:46,730 --> 00:09:48,529
power BI or another system so
349
00:09:48,529 --> 00:09:49,450
you can build your own
350
00:09:49,450 --> 00:09:50,730
dashboards.
351
00:09:50,730 --> 00:09:54,159
Azure advisor is the free helper
that most beginners never open.
352
00:09:54,159 --> 00:09:56,840
Microsoft calls it a
personalized cloud consultant,
353
00:09:56,840 --> 00:09:59,840
and it studies your setup and
recommends changes every day,
354
00:09:59,840 --> 00:10:02,919
from resizing an oversized
machine to shutting down
355
00:10:02,919 --> 00:10:04,399
something that sits idle.
356
00:10:04,399 --> 00:10:05,120
It's free.
357
00:10:05,120 --> 00:10:06,279
It's already switched on, and
358
00:10:06,279 --> 00:10:07,759
it's the closest thing Azure has
359
00:10:07,759 --> 00:10:09,039
to a to do list for saving
360
00:10:09,039 --> 00:10:09,440
money.
361
00:10:09,440 --> 00:10:10,840
Then come the rate levers where
362
00:10:10,840 --> 00:10:12,559
the discounts live, and Azure
363
00:10:12,559 --> 00:10:13,840
Savings Plan can cut compute
364
00:10:13,840 --> 00:10:15,039
costs by up to sixty five
365
00:10:15,039 --> 00:10:16,200
percent from pay as you go
366
00:10:16,200 --> 00:10:17,320
prices when your usage is
367
00:10:17,320 --> 00:10:17,600
steady.
368
00:10:17,600 --> 00:10:18,740
A reservation can save up to
369
00:10:18,740 --> 00:10:19,940
seventy two percent if you
370
00:10:19,940 --> 00:10:21,399
commit to specific usage for a
371
00:10:21,399 --> 00:10:22,080
set term.
372
00:10:22,080 --> 00:10:23,279
Azure Hybrid Benefit goes
373
00:10:23,279 --> 00:10:24,759
further and lets you reuse
374
00:10:24,759 --> 00:10:26,440
Windows Server or SQL server
375
00:10:26,440 --> 00:10:28,080
licenses you already own, which
376
00:10:28,080 --> 00:10:29,480
Microsoft says can save up to
377
00:10:29,480 --> 00:10:30,679
forty percent on virtual
378
00:10:30,679 --> 00:10:31,399
machines.
379
00:10:31,399 --> 00:10:32,559
Treat every one of those numbers
380
00:10:32,559 --> 00:10:33,759
as a ceiling rather than a
381
00:10:33,759 --> 00:10:34,919
promise, because they only pay
382
00:10:34,919 --> 00:10:35,840
off when your usage is
383
00:10:35,840 --> 00:10:36,600
predictable.
384
00:10:36,600 --> 00:10:39,080
One more idea ties all of this
together, and it carries a
385
00:10:39,080 --> 00:10:41,120
slightly awkward name focus.
386
00:10:41,120 --> 00:10:44,000
The FinOps Open Cost and usage
specification is a shared format
387
00:10:44,000 --> 00:10:46,360
for billing data, and Microsoft
is a founding member.
388
00:10:46,360 --> 00:10:47,639
It means a cost report from
389
00:10:47,639 --> 00:10:48,919
Azure, from another cloud, or
390
00:10:48,919 --> 00:10:50,159
from a software vendor can
391
00:10:50,159 --> 00:10:51,274
finally be read in the same
392
00:10:51,274 --> 00:10:51,950
language.
393
00:10:51,950 --> 00:10:52,590
That's a big deal.
394
00:10:52,590 --> 00:10:54,710
The first time you have to
compare them side by side.
395
00:10:54,710 --> 00:10:55,830
And remember that software
396
00:10:55,830 --> 00:10:56,990
licensing belongs in the same
397
00:10:56,990 --> 00:10:57,990
discipline.
398
00:10:57,990 --> 00:10:59,110
A Microsoft three hundred sixty
399
00:10:59,110 --> 00:11:00,309
five copilot license runs about
400
00:11:00,309 --> 00:11:01,509
thirty dollars per user per
401
00:11:01,509 --> 00:11:02,669
month as an add on to a
402
00:11:02,669 --> 00:11:03,470
qualifying plan.
403
00:11:03,470 --> 00:11:06,269
Once you multiply that across a
few hundred people, it deserves
404
00:11:06,269 --> 00:11:09,350
the same scrutiny you would give
any server on the Microsoft
405
00:11:09,350 --> 00:11:10,389
three hundred sixty five side.
406
00:11:10,389 --> 00:11:12,269
The same thinking applies to
licenses.
407
00:11:12,269 --> 00:11:13,309
It's common to find people
408
00:11:13,309 --> 00:11:14,549
holding a premium license they
409
00:11:14,549 --> 00:11:15,909
never use, or two overlapping
410
00:11:15,909 --> 00:11:17,149
subscriptions doing almost the
411
00:11:17,149 --> 00:11:17,710
same job.
412
00:11:17,710 --> 00:11:18,830
Cleaning those up is often the
413
00:11:18,830 --> 00:11:20,110
fastest saving in the whole
414
00:11:20,110 --> 00:11:20,710
tenant.
415
00:11:20,710 --> 00:11:23,230
The Microsoft three hundred
sixty five admin Center shows
416
00:11:23,230 --> 00:11:25,710
who is assigned what, so the
question stops being a mystery
417
00:11:25,710 --> 00:11:26,909
and becomes a task.
418
00:11:26,909 --> 00:11:27,870
Those tools hand you the
419
00:11:27,870 --> 00:11:29,269
visibility and the discounts,
420
00:11:29,269 --> 00:11:30,509
and that part is available to
421
00:11:30,509 --> 00:11:31,110
you today.
422
00:11:31,110 --> 00:11:33,909
What no tool can hand you is the
discipline to keep using them.
423
00:11:33,909 --> 00:11:37,190
Which brings us to the mistakes
that sink almost every practice.
424
00:11:37,190 --> 00:11:39,750
The mistakes that sink a FinOps
practice.
425
00:11:39,750 --> 00:11:40,889
The first mistake is treating
426
00:11:40,889 --> 00:11:42,470
FinOps as a pure, cost cutting
427
00:11:42,470 --> 00:11:43,269
exercise.
428
00:11:43,269 --> 00:11:44,230
When savings are the only
429
00:11:44,230 --> 00:11:45,769
scoreboard, teams learn to hide
430
00:11:45,769 --> 00:11:46,950
their spending instead of
431
00:11:46,950 --> 00:11:47,950
talking about it, and the
432
00:11:47,950 --> 00:11:49,049
numbers get worse rather than
433
00:11:49,049 --> 00:11:49,669
better.
434
00:11:49,669 --> 00:11:50,705
The second is running the
435
00:11:50,705 --> 00:11:52,059
practice without ownership or
436
00:11:52,059 --> 00:11:52,460
tags.
437
00:11:52,460 --> 00:11:54,820
If nobody can say which project
caused a charge.
438
00:11:54,820 --> 00:11:57,779
Every conversation about cost
turns into a guessing game and
439
00:11:57,779 --> 00:11:59,139
the fix never sticks for long.
440
00:11:59,139 --> 00:12:01,980
The third mistake is buying a
tool and calling the job done.
441
00:12:01,980 --> 00:12:02,860
Software can show you the
442
00:12:02,860 --> 00:12:04,100
numbers, but it can't repair a
443
00:12:04,100 --> 00:12:05,700
conversation that nobody wants
444
00:12:05,700 --> 00:12:06,100
to have.
445
00:12:06,100 --> 00:12:09,059
The fourth is keeping all the
work inside one central team.
446
00:12:09,059 --> 00:12:10,460
If engineers never see their own
447
00:12:10,460 --> 00:12:11,899
costs, they never build the
448
00:12:11,899 --> 00:12:13,299
instinct to design with cost in
449
00:12:13,299 --> 00:12:14,620
mind, and the central team
450
00:12:14,620 --> 00:12:16,220
becomes a bottleneck instead of
451
00:12:16,220 --> 00:12:17,179
an enabler.
452
00:12:17,179 --> 00:12:19,860
The fifth mistake is chasing the
last few percent of savings
453
00:12:19,860 --> 00:12:21,539
while ignoring the big calls.
454
00:12:21,539 --> 00:12:24,279
The foundation's twenty twenty
six report captured this well,
455
00:12:24,279 --> 00:12:27,620
quoting a practitioner who said
the big rocks of waste are gone.
456
00:12:27,620 --> 00:12:30,179
What remains, in their words, is
a high volume of smaller
457
00:12:30,179 --> 00:12:32,379
opportunities that take far more
effort to capture.
458
00:12:32,379 --> 00:12:33,620
The sixth is trying to measure
459
00:12:33,620 --> 00:12:34,940
value without agreeing on what
460
00:12:34,940 --> 00:12:36,059
value means.
461
00:12:36,059 --> 00:12:37,179
Teams end up arguing about
462
00:12:37,179 --> 00:12:38,299
whether a tool is worth the
463
00:12:38,299 --> 00:12:40,059
money when they never define the
464
00:12:40,059 --> 00:12:41,179
outcome they were buying in the
465
00:12:41,179 --> 00:12:42,059
first place.
466
00:12:42,059 --> 00:12:44,220
Mistakes like these rarely
announce themselves.
467
00:12:44,220 --> 00:12:45,740
They show up as a slow drift.
468
00:12:45,740 --> 00:12:48,580
The dashboards look fine, the
meetings keep happening, and the
469
00:12:48,580 --> 00:12:52,000
bill still climbs because nobody
changed the behavior behind it.
470
00:12:52,000 --> 00:12:53,159
The good news is that the same
471
00:12:53,159 --> 00:12:54,600
habits from earlier fix most of
472
00:12:54,600 --> 00:12:56,000
them, which means the cure is
473
00:12:56,000 --> 00:12:56,960
usually simpler than the
474
00:12:56,960 --> 00:12:57,720
disease.
475
00:12:57,720 --> 00:13:00,759
What FinOps can't fix FinOps is
honest about its limits, and you
476
00:13:00,759 --> 00:13:01,720
should be, too.
477
00:13:01,720 --> 00:13:02,879
It can't yet tell you whether
478
00:13:02,879 --> 00:13:04,200
your AI investment is paying
479
00:13:04,200 --> 00:13:05,759
off, because the whole industry
480
00:13:05,759 --> 00:13:06,840
is still wrestling with that
481
00:13:06,840 --> 00:13:07,519
question.
482
00:13:07,519 --> 00:13:10,000
The twenty twenty six survey put
it bluntly, and it shared a
483
00:13:10,000 --> 00:13:12,559
practitioner's line that nobody
can answer whether AI is
484
00:13:12,559 --> 00:13:13,799
providing value yet.
485
00:13:13,799 --> 00:13:16,440
Measuring the money you saved by
never making a mistake is just
486
00:13:16,440 --> 00:13:18,320
as hard as another practitioner.
487
00:13:18,320 --> 00:13:21,200
In that same report, put it once
you fix a problem, it's gone.
488
00:13:21,200 --> 00:13:23,000
So how do you give a developer
489
00:13:23,000 --> 00:13:24,200
credit for a cost that never
490
00:13:24,200 --> 00:13:24,600
appeared?
491
00:13:24,600 --> 00:13:27,440
It can't mend bad architecture
or unclear ownership either,
492
00:13:27,440 --> 00:13:30,360
because those are design and
culture problems that a report
493
00:13:30,360 --> 00:13:31,679
will only expose.
494
00:13:31,679 --> 00:13:34,600
It also can't promise that the
easy wins are still waiting,
495
00:13:34,600 --> 00:13:37,240
since mature teams report that
the obvious waste has largely
496
00:13:37,240 --> 00:13:40,159
been picked over, and it can't
turn a commitment discount into
497
00:13:40,159 --> 00:13:42,960
free money, a reservation or a
savings plan only helps when
498
00:13:42,960 --> 00:13:45,960
your usage is steady, so buying
one for a workload you plan to
499
00:13:45,960 --> 00:13:48,080
shut down can lose more than it
saves the cost.
500
00:13:48,080 --> 00:13:50,990
Data itself also arrives after
the fact, which is worth sitting
501
00:13:50,990 --> 00:13:51,870
with for a second.
502
00:13:51,870 --> 00:13:52,590
You're always reading
503
00:13:52,590 --> 00:13:54,169
yesterday's numbers, so FinOps
504
00:13:54,169 --> 00:13:55,629
is a practice of steering, not a
505
00:13:55,629 --> 00:13:55,870
switch.
506
00:13:55,870 --> 00:13:57,149
You flip once and forget.
507
00:13:57,149 --> 00:13:59,629
The harder truth is that FinOps
asks for patience.
508
00:13:59,629 --> 00:14:02,190
The first pass usually finds a
few obvious wins, and the second
509
00:14:02,190 --> 00:14:04,789
pass finds fewer, which is
normal rather than a sign that
510
00:14:04,789 --> 00:14:06,190
the practice stopped working.
511
00:14:06,190 --> 00:14:07,750
None of that makes the practice
weak.
512
00:14:07,750 --> 00:14:09,110
It just means FinOps is a way of
513
00:14:09,110 --> 00:14:10,169
asking better questions about
514
00:14:10,169 --> 00:14:11,710
technology spending, not a
515
00:14:11,710 --> 00:14:12,909
machine that prints savings on
516
00:14:12,909 --> 00:14:13,269
demand.
517
00:14:13,269 --> 00:14:16,029
Knowing what it can't do is what
keeps you from expecting magic,
518
00:14:16,029 --> 00:14:19,190
and then giving up when the
magic never shows.
519
00:14:19,190 --> 00:14:20,830
What to remember about FinOps.
520
00:14:20,830 --> 00:14:22,950
So here is the short version to
carry with you.
521
00:14:22,950 --> 00:14:24,470
FinOps is a shared habit of
522
00:14:24,470 --> 00:14:26,549
watching what technology costs
523
00:14:26,549 --> 00:14:28,350
and deciding together what it is
524
00:14:28,350 --> 00:14:28,870
worth.
525
00:14:28,870 --> 00:14:31,789
It runs on a simple loop of
inform, optimize, and operate.
526
00:14:31,789 --> 00:14:33,289
Microsoft hands you the starting
527
00:14:33,289 --> 00:14:34,750
tools inside the products you
528
00:14:34,750 --> 00:14:35,629
already use.
529
00:14:35,629 --> 00:14:38,029
Now for one question, and I
really do want your answer.
530
00:14:38,029 --> 00:14:40,590
When was the last time someone
on your team asked who owns the
531
00:14:40,590 --> 00:14:43,669
cloud bill if this made the
topic feel less intimidating?
532
00:14:43,669 --> 00:14:47,669
Subscribe to M365 FM to master
the tools and the mindset
533
00:14:47,669 --> 00:14:49,350
shaping the future of work.
