Just to really motivate what I'm going to talk about today, I think it's fascinating
if we look at what Sinan just talked about.
At a high level, I could ask companies questions, really any company in the world.
Questions about their internal network, and who communications with who in a company,
really simple questions that nobody can answer.
You could say, "How much does the management team talk to the engineering team?"
Nobody knows.
You can't even answer a question like, "How much should salesperson communicate with a
customer?"
Which is crazy, right?
You think about all the detailed data we typically collect about customers external to our company,
but when it comes to inside our organizations, we forget about all of those things.
We don't remember that we actually do have data about what people do at work.
Specifically, today we have sensor data from ID badges, IOT sensors in the workplace.
We have email, chat meeting data, some really rich data about what goes on at work.
So now that we have this, we can start to do things as Sina showed, like look at networks
within the organization.
Who talks to who, all those things.
All right, and so a lot of times, I'll be having a conversation with a company, and
they'll ask me, "What is the best network if I want to try to make for the best teams
in my organization?
I want to make them work in a certain way."
Here's the best team, right?
Again, I'm from Boston.
Everyone knows where I'm from here.
That's right.
Yep, woo!
Anyway.
But so, here's the question.
First a question, how many people here work in a technology company?
A bunch, I figured a bunch.
Okay.
So then the question I'll put you is if you're running a technology company, what is the
best network to have at a technology company?
So who thinks it should be a really cohesive network, focused more on execution?
Execution?
Who thinks it should be a little bit more diverse focused on coming up with new ideas,
things like that?
A couple people.
All right, so who thinks it's neither?
All right.
So let's see, I think like literally 10% of people voted here.
Come on, we need [inaudible] guys.
The election's going to come up in a couple years, we're going to need people to actually
vote.
Okay, but the correct answer is that the surprise that there's none, and so Sina hinted at this
earlier.
But importantly, it's not just that it depends on ... It's not just that for certain companies
a certain type of network is better.
Like, if you're focused more on execution that having a cohesive network is better,
and if you're a company that's focused on innovation you should have a more diverse
network.
It's that depending on the work that you're doing any given day, any given week, any given
month, your network needs to change, and it does change.
So who I talk to this week, this month, is different from who communicate with next month.
Okay, now the difference is that the way we create these networks today is an incredibly
ad hoc way.
I'll bump into people in the hallway, and I'll start talking to them.
Maybe I have a meeting with you, and then we start emailing each other later, but that's
not directed with the same sort of thought that we put into, for example, work charts.
But of course, those actually have a lot of an impact on how we actually work.
Really what's fundamentally changed is our ability to measure these networks, right?
That today I have, for example, next generation sensor badges that look at who talks to who,
and how people talk to each other.
Again, not recording what people say, but at a millisecond level, understanding across
an organization of tens of thousands people, hundreds of thousands, what's actually going
on.
Of course, we all already have email data to day, and so then I know who emails who,
I know who chats with who, I know who's meeting with who.
That enables us to look at these changes so quickly day by day, week by week, month by
month, and actually manage to that.
It's just a very different way of managing an organization.
And this gets to what Sina hinted at earlier, that the difference between exploration and
exploitation, right?
That sometimes I need actually exploit a resource, I need to execute a project, I need to build
a thing.
Sometimes I need to come up with new ideas, I need to break out of my, do other things,
but depending on what you're working on, on a given week, one or the other network could
be different, could be more effective.
Right?
So let's get concrete here.
One of the things that we do at my company is, again, we use sensor data, digital data
to understand if people interact and collaborate at work, and one of the companies we looked
at was a multi-billion dollar IT firm, where the employees in one of the divisions were
creating these multi-million dollar data servers for their customers.
So most of these people have masters degrees, it's a very complex task.
Basically, they get a call from a sales person and he says company XYZ needs a server that
does these sorts of things.
They actually can figure these server in their software, they price it out, they send it
out to them.
What's interesting about this though, is that even though this is a very complex task, you're
able to measure their performance pretty accurately.
Essentially, look at completion time from when they're assigned a task to when they
complete it.
How long did that take?
And if it's very simple, it took them five minutes, if it was very complex, it could
take them up to eight hours.
So the question is, what do the most effective people do?
But before we answer that, if we just step back for a second.
Let's think about ... Imagine you're working on a task, right?
And imagine you have a question about how to complete it, and what you'll do is you'll
probably go try to find an expert to talk to you.
You'll find somebody like Al, right?
But the thing is that at a company, there's not just one Al, right?
There's many Als.
Actually what we found at this company, where there were four people, four informal experts,
who if you spoke to, had a pretty massive impact on how quickly you completed a task.
But this gets very interesting, because if I looked at in general, let me look at who
communicates with who, who actually talks with who, in general, across a whole month.
What you saw was that the types of networks that for that month correlated with higher
performance, was having very cohesive network.
As Sina mentioned, if the people I talk to talk a lot to each other, those people completed
tasks very quickly.
Essentially, people with the most cohesive task who would complete ... With cohesive
networks, would complete tasks 50% faster than people in the least cohesive networks.
Pretty strong effect!
And you would see these informal experts who were scattered throughout the network, and
there wasn't really a corelation between their position and performance.
But then if we just said, let's look at who communicates with who when they're working
on a task, and then it looked very different.
Basically, you saw these informal experts at the center of a network, and that's how
you identify them, right?
They had the same job titles of everybody else, but you saw that these people, one of
these four people, over the course of a month almost everybody talked to them at some point
when they were working on a task.
When you talk to one of those people, you completed that task in a third of the time
that it normally took you.
This is a massive effect.
But here's what's really interesting.
These informal experts had statistically average performance, individually, and this is because
we're measuring performance on an individual level.
I care about your output, so if I look at how quickly these informal experts complete
a task, they looked identical to everybody else, and a big reason was they were spending
on average about four hours a day helping their coworkers.
Now, this is a problem, because ... Here's the thing, I want to this behavior, if I run
an organization.
Right?
I want them to help other people in the company.
If I make 20 people I work with 10% more effective, and I spend six hours of my time talking to
them, I mean, that should be my job.
But if you can't quantify that, if I can't see that I need this network change in a very
specific period of time, it's an interesting trade off, right?
That I want in general, I want cohesive networks, but then specifically for these in-task networks,
I want people to have very different types of behaviors.
You're talking about, I need to change a whole bunch of things, and in particular, this company
tried to change incentives.
So what they did is ... Originally again, they paid people based on individual performance.
They said if you complete tasks more quickly, we pay you more, and the challenge here again
is that if I'm an informal expert, I will get paid the same as everybody else, but what
they actually did is they changed incentives.
They said, "Well, we're actually going to have a group performance target, and if we
hit that performance target, everybody gets a bonus."
So you're turning incentivized people to help the group out.
In addition to what they did, which I really like, is they had ... Every employee at the
end of the day would ... They go around in a circle, and they would call out one person
who helped them that day, really for the informal benefits there, because frankly today, that's
why these informal experts were engaging in this behavior.
It's because they wanted to help out their coworkers, and what we can do today is with
the proliferation of data about what's happening at work, is I can quantify even in a one-minute
level, what's happening.
Are these interactions happening?
Are they positive, are they negative for the organization?
And I think that importantly, when we take this view, we start to understand how ... When
we think about productivity, both in an individual and a corporate level, again, we like to think
about, especially in the US, the amount of stuff you produce.
The number emails you answer, the number of reports you write, and again, I think that's
completely the wrong way to look at it.
I mean, of course you would at the end of the day need some work to get done, but importantly
the way that really we improve the output of the group, that's really what we care about.
In order to do that, we can't just focus on getting the right answer for today, because
what works today could not work next month, or next year.
So really to make sure that we understand these patterns, we see them changing over
time, and we're able to look at what actually maps to outcomes, that's what's really critical.
So ... It's funny, because Sina showed literally the exact same slide.
I think what a lot of the work that we've done is about looking over very large groups
of people, and trying to see the trends, the network patterns that correlate to outcomes,
and then using that to again, plan changes.
But the point is that's not a one-time thing.
Like, a lot of companies, they'll maybe hire some consultants, they'll try to develop a
new training program, let's say, and they'll roll that out, and they'll think that works
for all time.
But of course, you know, five or ten years later, they realize that didn't work, or it
doesn't matter anymore today.
And I think that because now we have this dynamic information about what goes on in
the organization, we can start to just admit that we actually first of all, don't know
what works today, but also we can admit that these things change really rapidly.
So by seeing how our networks are changing day by day, week by week, month by month,
and relating that outcomes, we can actually start to understand how we can evolve our
organization along with a network.
So, that's all I got, thanks very much.
For more infomation >> The Flash | Inside: When Harry Met Harry… | The CW - Duration: 1:46. 
For more infomation >> Max heroes vs Max Troops - Duration: 5:23. 
For more infomation >> Fight breaks out between homeless man, Dunkin' Donuts manager - Duration: 0:52. 
For more infomation >> Vi presento Bonnie [SUB ES] - Duration: 3:28. 
For more infomation >> عــــــــاجل: الملك محمد السادس ينهي زيارته الرسمية لقطر بإنجزات كبيرة - Duration: 1:42. 
For more infomation >> ¿QUÉ SIGNIFICA OK? | DrawXpress (el verdadero origen de la palabra es INCREÍBLE) - Duration: 3:13. 
For more infomation >> شكون سيفط ليا هاد الكارطونة، و علاش ؟ ! - Duration: 10:02. 
For more infomation >> Пасхалки в The Evil Within [Easter Eggs] - Duration: 11:00. 
For more infomation >> D-Day Dice: 2nd Edition Gameplay Runthrough - Duration: 23:16. 
Không có nhận xét nào:
Đăng nhận xét