November 20, 2023 • 23 mins
My guest this time is Adam Sharman from data consultants to the manufacturing sector https://dsifer.com/ - who “combine extensive experience in manufacturing, primary and food processing with world leading data analysis to help business leaders embrace the future of productivity”.
We had an excellent chat about the uses of data in manufacturing businesses and, as so often, we came back to the secret sauce really being change management and understanding that at the end of the day data is about people.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Welcome to another episode of the Data Revolution podcast.
(00:19):
Today, my guest is Adam Sharman, who's from a company called Decipher, who are data specialists for manufacturing.
And he's based in New Zealand. Welcome, Adam.
Thanks so much for having me. Okay, that's a pleasure to be here.
Yeah, I was really interested when we were chatting about some of the data challenges that you face in manufacturing, because, you know, it's an area that not many people think about data being connected to.
(00:43):
So tell us a bit about your organization and what they do with data.
Yeah, fantastic. Well, thanks so much for having me. Yeah, it's a really interesting space.
So our organization focuses on all things data for manufacturers.
So that's everything from setting up data architecture, visualization of business intelligence, and then analytics and insight off the back of it.
(01:08):
And we're really supporting advanced manufacturing, so digital manufacturing.
There's been a real shift in the last five years to use data, particularly as manufacturers start to gather more data from PLCs, from their machines, and try and connect that to business systems.
We're finding a lot of manufacturers are starting to understand the power of data, but they're probably not using it effectively yet for a connected decision ecosystem.
(01:38):
And that's really what we want to be trying to support is not just manufacturing data, but actually how does that connect to the overall business and how do you use that then to drive decision making at the business level.
So many years ago I used to work for GE, you know, and I had some involvement with manufacturing.
And a lot of people don't understand how it used to be sort of big gray smokestacks, but now it's advanced manufacturing. Can you talk a bit about that?
(02:08):
Yeah, absolutely. So there are still those big gray smokestacks about the place.
But actually what we're seeing is even in those more traditional heavy industries, the digitization of the manufacturing environment really focuses on automation of process, automation of system.
And that could be to reduce things like labor consumption.
(02:32):
So you're moving away from factory lines that have hundreds of people on it doing repetitive tasks.
And that can all be automated through robotics.
But then using the data that comes from those machines to actually drive accuracy, productivity and reduce things like health and safety risk, quality risk, and ultimately increase productivity.
(02:55):
And what we're seeing is that a lot of organizations are starting to realize that they cannot just replicate what they currently do or used to do more efficiently, but they can actually use it as a strategic advantage.
So it might be that they can start to rethink how they position themselves in the market based on things like increased what we call diphots are delivered in full on time.
(03:21):
So making what you say you will when you say you will improve quality and improved product.
So all of these digital tools are actually taking data from the manufacturing environments are from the tools themselves and then putting them into optimization models.
So optimization of your production planning, for example, inventory planning and then feeding them back into the machine so that machines can be using things like video monitoring for quality, video monitoring for quality.
(03:50):
So it sounds to me like, you know, they're taking all of those old fashioned Japanese production ideals like Kanban and stuff and sort of ramping them up on steroids.
So what are some of the things that the organizations are finding interesting in their data?
Yeah, so it really is seeing a vast array of maturity when it comes to how people are using the data. So sometimes it's a case of, you know, we've got the spreadsheet that we've built that looks at things like on product time or overall equipment effectiveness.
(04:27):
And then they're replicating that and that's sort of you know, standard reporting.
But then you can actually start to use some of that historical data to get into prediction models. So particularly things around predictive maintenance, for example.
So, you know, if you're running a machine for a certain number of hours or certain type of job, you can tell when that machine will need to be maintained or when it's likely to fail.
(04:52):
And the prediction models will take historical data for that machine and then information from the manufacturers of that machine, how you're running it, what the maintenance history is of it.
And then can actually predict when that machine will need to be maintained.
And that reduces things like unplanned downtime, which can be very costly for particularly for the organizations.
(05:16):
So every sort of half an hour, every minute even of unplanned downtime can be thousands if not millions of dollars.
Yeah, I remember those.
So really what I'm interested in though is that sounds like pure play old fashioned machine learning.
How is Generative AI playing into that now? Or is it? I think it's playing into everything though.
(05:41):
Yeah, it's playing into everything. And what I'd actually, what we're finding is that organizations are really just scratching the surface, particularly with Generative AI.
In the last year, that's become the kind of buzzword. And some organizations are using Generative AI in fairly simple ways.
You know, they're using it to draft policies, to draft plans.
(06:04):
But others are actually using it. If you particularly if you layer it over the top of something like an open source GitHub model, for example, you can actually get some pretty amazing optimization models by feeding your data in directly from the machines into an optimization model.
The looks at, you know, where you might be able to drive efficiencies through your manufacturing system.
(06:26):
And that's something that you're basically talking about connecting their OT, their operational technology up with their IT sort of BI world.
Yeah, exactly. Yeah, and this is where we've actually focused a lot of attention recently with organizations.
So a lot of our conversations start with people saying, wouldn't it be great if my systems just talk to each other?
(06:50):
They talk about systems, they talk about an ERP layer, you know, manufacturing execution layer. But now with advanced manufacturing, you're layering over the top of that direct data from machines.
And at the moment, if you're running a more traditional model, that data has to flow up through your IT stack.
And we're starting to build models now for clients and see this more often where this is actually going direct into a data warehouse or a data lake.
(07:18):
And how are you going with data quality for that? Because that's always been the argument against that very thing.
Yeah, well, so the data quality that comes on the machines is incredible. And actually, we're seeing improved data quality because it hasn't had to go through multiple...
Making the people out of the loop.
Yeah, that's right. You take the people out of the equation and put it for one, but also, you know, with a unified semantic layer, you know, actually everything is referring to the same thing in the same way.
(07:46):
So actually, the translation of that becomes really simple up through the stack.
So actually, the data quality and potentially the data quality, the data validation is actually becoming much easier as a result.
And how are you? How many metadata in that world?
Yeah, so metadata, it really depends on the organization and the maturity of the organization.
(08:09):
So the metadata is trying to build a unified semantic layer in the data warehouse.
And that's really kind of the starting point that we're seeing for most organizations.
You know, a lot of organizations we go into have got, you know, they've got a spreadsheet over here that Bob maintains.
You know, we've got a web app that Susan built three years ago. We've got the ERP, MES, and all of them have their own semantic standards.
(08:36):
And all of them refer even potentially to sensors by a different name.
Yes, that's why I asked about metadata.
Yeah, well, that's right. So if you think about the manufacturing environment, the complication that we're seeing in the last few years is just the number of data points.
You know, so you might have thousands of sensors running through a manufacturing system and each one of those has its own naming convention and flows through to your IT stack.
(09:05):
So right back to the start of designing the architecture, you've got to have a common naming convention across all of those sensors, which is a huge activity to retrofit.
So if an organization can start that early in the process of their automation journey, then so much better.
So this is a pro tip for anybody who's thinking about this in the future. Name things, sensible common names. Don't give them all the names.
(09:33):
Absolutely. And that doesn't just apply to manufacturing, right? I think the complication for manufacturing is just the number of data points that you get through those sensors.
And the sensors are so cheap and easy to implement now and you can layer them on top of relatively old equipment and still get really good data.
(09:54):
So does that bring sort of networking architectures into play as to how you manage that? Are you sort of looking at edge-style computing for that sort of thing?
Yeah, exactly. So it's particularly when you're retrofitting equipment, if you're putting a PLC or a sensor on more traditional equipment, then you're looking at edge computing.
(10:19):
And it's actually relatively straightforward to do now with the right architecture. But this is the pro tip that you just mentioned, you have to start as early as possible.
For most organizations, that's not the reality, particularly manufacturing.
Yeah, manufacturing is like the dark ages. I remember going into a manufacturing floor and there was an AS400 computer in the corner that had never been turned off in the 30s that had been there and it was still working.
(10:47):
And that was the sophistication of their technology.
Yeah, yeah, it's amazing. I mean, you think the capital investment required is huge for most organizations and for a lot of organizations, they just don't have that capital available.
And this is part of the challenge, I think, for manufacturing as a whole.
It's very easy, particularly if you're a medium size enterprise, you look at videos on YouTube or LinkedIn of Amazon's warehouses in China and it's completely lights out.
(11:18):
There's robots everywhere. There's one person, there's an enormous warehouse, fully automated. And everybody thinks that that's where they need to get to.
And then the question is, well, how? And the second question is, do you really need to?
Most organizations probably don't need to get there yet. It might be a roadmap that they're on, but actually there's a lot that they can do now with relatively low cost when you look at things like edge sensors and PLCs.
(11:47):
Well, just to reduce their wastage. That's absolutely wastage. That's a really good thing.
Yeah, that's right. But when you put those videos up, you know, the Amazon warehouse, a lot of people just sort of do in the headlights, you know, it's like, how do we even get there when we might be struggling today with cash flow or today with productivity or today to find the right people.
(12:11):
And often the challenge for organizations is to actually take a step back out of the day to day and say, what's the strategy around this? What's actually going to get us from here to here to here to here on a roadmap, rather than feel like they have to go from where they are now to that sort of full Amazon setup?
(12:33):
Well, you know, for every organization, you don't need to be like Amazon, because, you know, it might actually mean be better for an organization to keep people more in the loop.
Absolutely. And this is a conversation that we have all the time around what's the competitive advantage that you're trying to drive for your organization.
Now, because for some organizations, it is cost. So you have to manufacture as cheaply as possible. And therefore, you know, things like production efficiency are the number one goal.
(13:04):
But for others, it might actually be security. So for example, we're seeing a few organizations in New Zealand, actually the UK as well, we're there to winning work from China and other places because of security that they can guarantee through their process.
So they might not be the cheapest solution. Now, they still need to manufacture as efficiently as possible. But actually, their strategic advantage is security. It's not necessarily price.
(13:34):
And so when you start to have that conversation, then that you can then play forward in terms of data or technology strategy, because that's got to be your number one goal as maintenance of data security.
And so outside all of the really big manufacturing organizations, you know, uni leaders and GES and stuff, a lot of those organizations won't have thought of those kinds of things because they're just doing business.
(13:59):
They've always done business. They've always struggled to improve their margins and stuff, but they've never really taken a very scientific approach to it. So does data help lead that conversation?
Yeah, and it really comes down to the culture of the organization. So you if you've got people who are mature in terms of their data capability or even their mindset towards a data driven approach, then absolutely will try it.
(14:31):
But a lot of organizations are fairly immature when it comes to data capability and even how you use data. And most organizations will be absolutely on top of things like reporting, historical reporting.
And actually, how do you use data to forecast to scenario model to predict? That's a very big gap.
In fact, there was a lot of people will be using Excel. Absolutely. Yeah, ridiculous Excel spreadsheets that are so complicated if you put something in there just fall apart.
(15:01):
Well, that's right. But you know what? I'm always amazed at how what those Excel budgets can do.
Yeah, they're amazing. But that's scary because they're so fragile.
Yeah, because of the dependency they create. And it doesn't matter what else you do. If that's what the foundation of the model is, then it's a fairly precarious model.
(15:22):
We've actually been moving a lot of our old Excel models. We had Excel models for predictive models that you press the button on them to recalculate and then you go and leave the machine for two days while it did it.
You know, and it was quite scary. We've moved all of those into our cloud data platform, you know, where we do all the heavy lifting in the cloud, which makes a lot of sense, but they can still have an Excel front end on it so they feel at home.
(15:50):
Yeah, well, that's right. I mean, you know, we do a lot of our visualization work in Power BI. You know, there's amazing things you can do with Power BI that look very simple.
And sometimes people, you know, it's always deceptive how simple the front end looks when you think about what's going on behind the scenes and the, you know, the translation that's required, the transformation of data, the analysis of the data in order to be presented simply.
(16:15):
It's pretty deceptive when you look at what is effectively a fairly simple dashboard on the Facebook.
Are those companies that are building that sort of thing also realizing that they will need to beef up their recruitment and skills in the data space to feed, they buying a puppy, they've got to feed the puppy?
That's right. That's right. And there's really two ways of looking at that. So one is what's the specific technical resource that you need to build in and that might be, you know, data architects, VA's, data scientists.
(16:48):
And for a lot of organizations that that's going to be relevant for some, that's so depending on their scale. But then you've got to look at what's the, what's the capability that's required across the organization, across all roles, and particularly those sort of decision maker roles, you know, team leader level and above, where people are actually relying on data in order to make good business decisions.
(17:11):
And now we're talking about data literacy, which is the biggest elephant in the room worldwide.
Yeah, well, that's right. And, and that's so, you know, lifting up. There's so much in there. It's very a whole other podcaster on data literacy.
You know, there's, there's how do you interpret data? How do you read data? But there's also how do you set up your organization as an ecosystem when it comes to decision making using data? You know, so if you're, for example, we have conversations with HR departments, for example, and HR departments are actually getting really good when it comes to data analytics for
(17:51):
workforce planning, for example. At the moment, that's sometimes disconnected to things like the production plan, and particularly things like investment in automation. So if you look at your investment in a machine, a lot of organizations will factor in purely what the depreciation of that machine is in order to
(18:14):
determine its ROI. When you start to factor the resourcing requirement in, you know, people, total cost of ownership, over cost, relationship, exactly. And in the last sounds like it should be a fairly simple concept. It requires an ecosystem driven
decision making. And it means that it means that it can't be financed driving it because they like that method because it's easy. And it's the way they've always done it. So it's a really, it's a really interesting conversation inside an organization.
(18:44):
I remember, I was at GE early this century, but I was able to say to my CEO, based on data, that an idea that he had wasn't good, and he looked at the data and said, Yeah, you're right. And we dropped that idea. And that, to me, is a good data driven organization when you have that
here is evidence, don't do this. And it's pretty rare.
(19:08):
Yeah, well, that's right. And there's this multiple so maturity, there's the people need to come up on this. So, you know, often, we get really good buy in when the data supports what people already thought.
The challenge comes, you know, what happens when the data, you know, as in, in that case, actually, you know, goes against what was previously thought. And the maturity there is, you know, is letting go of, you know, previous assumption or previous ego even going with the data.
(19:37):
You know, sometimes what we find is, as long as the data is reinforcing what people thought, then they're happy to go with it. As soon as they go against it, suddenly become, you know, the machine whisperer, the only person that knows how to tune that piece of equipment.
You know, the old ways come back in.
Oh, I was just wondering, so what advice would you give to people who are in that situation where they've not presented data that reinforces management's beliefs. So how do you manage that?
(20:06):
Yeah, well, I think that that's, in some ways, a classic change management, you know, people have to understand the context of the problem.
I think where you can put a dollar value on, on information or use a metric that's going to resonate with that particular audience.
Then that's really where you start to talk their language. So in the manufacturing world, they're starting to talk about productivity, over equipment effectiveness, you know, cost of quality failure, those sorts of things that have an immediate impact on bottom line.
(20:40):
And then you'll start to get people's attention.
If you can't do that, then it's, you know, it's about providing the context for that decision and trusting, encouraging people to trust where that data has come from and, you know, the data sources, how that data is collected.
But it's really going back to first principles of change management, because for a lot of people, there's an identity tied up with the ability to make decisions based on, you know, their own experience.
(21:13):
Yeah.
And a lot of the time, you know, if you're contradicting what their, their decision would be.
Or their identity, really, their identity.
That's right. It comes down to identity, right? So their identity in the past might have been that, you know, I know this process better than anybody else. Therefore, I am able to make really good decisions about this.
And, you know, part of what we try and do is actually get in behind those individuals and say, you know, absolutely, you've got that experience. But now here's another tool that you can use to actually improve, you know, the decision making.
(21:46):
You can stay the expert. You don't have to know.
Yeah, exactly. Exactly. Yeah. And, you know, it's about putting those people, you know, front and center and helping them do their job, you know, rather than try and challenge it in any way.
Yeah, well, that's the thing, make people look like heroes with the data, not against them.
(22:09):
Yeah, that's right. And it's something we talk about all the time, you know, particularly it's not just in manufacturing, but manufacturing probably feels there's more than other industries where they've been very labor intensive.
Once you start to automate lines, you do start to eliminate the need for some human, human roles.
But we've seen really good success stories where organizations have retrained people actually to start to use the data to do things like predictive maintenance that improve the production output overall, even if they've moved away from the more hands on.
(22:44):
So it's kind of interesting that we're not seeing necessarily a lot of job losses through digitization or through automation. It's just an augmentation of what those capabilities are.
And the organizations that invest in those capabilities that help people through that and critically, you know, to the point that you made help people shift their identity through that process.
(23:05):
Those are the organizations that are successful.
And that is an excellent note to end on. Thank you so much, Adam. It's been really great chatting to you. And thank you, Adam, from DeCipher.
Thank you so much, Kate. It's been great to chat.
And that is it for another episode of the Data Revolution podcast. I'm Kate Crothers. Thank you so much for listening.
(23:28):
Please don't forget to give the show a nice review and a like on your podcast app of choice. See you next time.
Welcome to another episode of the Data Revolution podcast.
(00:19):
Today, my guest is Adam Sharman, who's from a company called Decipher, who are data specialists for manufacturing.
And he's based in New Zealand. Welcome, Adam.
Thanks so much for having me. Okay, that's a pleasure to be here.
Yeah, I was really interested when we were chatting about some of the data challenges that you face in manufacturing, because, you know, it's an area that not many people think about data being connected to.
(00:43):
So tell us a bit about your organization and what they do with data.
Yeah, fantastic. Well, thanks so much for having me. Yeah, it's a really interesting space.
So our organization focuses on all things data for manufacturers.
So that's everything from setting up data architecture, visualization of business intelligence, and then analytics and insight off the back of it.
(01:08):
And we're really supporting advanced manufacturing, so digital manufacturing.
There's been a real shift in the last five years to use data, particularly as manufacturers start to gather more data from PLCs, from their machines, and try and connect that to business systems.
We're finding a lot of manufacturers are starting to understand the power of data, but they're probably not using it effectively yet for a connected decision ecosystem.
(01:38):
And that's really what we want to be trying to support is not just manufacturing data, but actually how does that connect to the overall business and how do you use that then to drive decision making at the business level.
So many years ago I used to work for GE, you know, and I had some involvement with manufacturing.
And a lot of people don't understand how it used to be sort of big gray smokestacks, but now it's advanced manufacturing. Can you talk a bit about that?
(02:08):
Yeah, absolutely. So there are still those big gray smokestacks about the place.
But actually what we're seeing is even in those more traditional heavy industries, the digitization of the manufacturing environment really focuses on automation of process, automation of system.
And that could be to reduce things like labor consumption.
(02:32):
So you're moving away from factory lines that have hundreds of people on it doing repetitive tasks.
And that can all be automated through robotics.
But then using the data that comes from those machines to actually drive accuracy, productivity and reduce things like health and safety risk, quality risk, and ultimately increase productivity.
(02:55):
And what we're seeing is that a lot of organizations are starting to realize that they cannot just replicate what they currently do or used to do more efficiently, but they can actually use it as a strategic advantage.
So it might be that they can start to rethink how they position themselves in the market based on things like increased what we call diphots are delivered in full on time.
(03:21):
So making what you say you will when you say you will improve quality and improved product.
So all of these digital tools are actually taking data from the manufacturing environments are from the tools themselves and then putting them into optimization models.
So optimization of your production planning, for example, inventory planning and then feeding them back into the machine so that machines can be using things like video monitoring for quality, video monitoring for quality.
(03:50):
So it sounds to me like, you know, they're taking all of those old fashioned Japanese production ideals like Kanban and stuff and sort of ramping them up on steroids.
So what are some of the things that the organizations are finding interesting in their data?
Yeah, so it really is seeing a vast array of maturity when it comes to how people are using the data. So sometimes it's a case of, you know, we've got the spreadsheet that we've built that looks at things like on product time or overall equipment effectiveness.
(04:27):
And then they're replicating that and that's sort of you know, standard reporting.
But then you can actually start to use some of that historical data to get into prediction models. So particularly things around predictive maintenance, for example.
So, you know, if you're running a machine for a certain number of hours or certain type of job, you can tell when that machine will need to be maintained or when it's likely to fail.
(04:52):
And the prediction models will take historical data for that machine and then information from the manufacturers of that machine, how you're running it, what the maintenance history is of it.
And then can actually predict when that machine will need to be maintained.
And that reduces things like unplanned downtime, which can be very costly for particularly for the organizations.
(05:16):
So every sort of half an hour, every minute even of unplanned downtime can be thousands if not millions of dollars.
Yeah, I remember those.
So really what I'm interested in though is that sounds like pure play old fashioned machine learning.
How is Generative AI playing into that now? Or is it? I think it's playing into everything though.
(05:41):
Yeah, it's playing into everything. And what I'd actually, what we're finding is that organizations are really just scratching the surface, particularly with Generative AI.
In the last year, that's become the kind of buzzword. And some organizations are using Generative AI in fairly simple ways.
You know, they're using it to draft policies, to draft plans.
(06:04):
But others are actually using it. If you particularly if you layer it over the top of something like an open source GitHub model, for example, you can actually get some pretty amazing optimization models by feeding your data in directly from the machines into an optimization model.
The looks at, you know, where you might be able to drive efficiencies through your manufacturing system.
(06:26):
And that's something that you're basically talking about connecting their OT, their operational technology up with their IT sort of BI world.
Yeah, exactly. Yeah, and this is where we've actually focused a lot of attention recently with organizations.
So a lot of our conversations start with people saying, wouldn't it be great if my systems just talk to each other?
(06:50):
They talk about systems, they talk about an ERP layer, you know, manufacturing execution layer. But now with advanced manufacturing, you're layering over the top of that direct data from machines.
And at the moment, if you're running a more traditional model, that data has to flow up through your IT stack.
And we're starting to build models now for clients and see this more often where this is actually going direct into a data warehouse or a data lake.
(07:18):
And how are you going with data quality for that? Because that's always been the argument against that very thing.
Yeah, well, so the data quality that comes on the machines is incredible. And actually, we're seeing improved data quality because it hasn't had to go through multiple...
Making the people out of the loop.
Yeah, that's right. You take the people out of the equation and put it for one, but also, you know, with a unified semantic layer, you know, actually everything is referring to the same thing in the same way.
(07:46):
So actually, the translation of that becomes really simple up through the stack.
So actually, the data quality and potentially the data quality, the data validation is actually becoming much easier as a result.
And how are you? How many metadata in that world?
Yeah, so metadata, it really depends on the organization and the maturity of the organization.
(08:09):
So the metadata is trying to build a unified semantic layer in the data warehouse.
And that's really kind of the starting point that we're seeing for most organizations.
You know, a lot of organizations we go into have got, you know, they've got a spreadsheet over here that Bob maintains.
You know, we've got a web app that Susan built three years ago. We've got the ERP, MES, and all of them have their own semantic standards.
(08:36):
And all of them refer even potentially to sensors by a different name.
Yes, that's why I asked about metadata.
Yeah, well, that's right. So if you think about the manufacturing environment, the complication that we're seeing in the last few years is just the number of data points.
You know, so you might have thousands of sensors running through a manufacturing system and each one of those has its own naming convention and flows through to your IT stack.
(09:05):
So right back to the start of designing the architecture, you've got to have a common naming convention across all of those sensors, which is a huge activity to retrofit.
So if an organization can start that early in the process of their automation journey, then so much better.
So this is a pro tip for anybody who's thinking about this in the future. Name things, sensible common names. Don't give them all the names.
(09:33):
Absolutely. And that doesn't just apply to manufacturing, right? I think the complication for manufacturing is just the number of data points that you get through those sensors.
And the sensors are so cheap and easy to implement now and you can layer them on top of relatively old equipment and still get really good data.
(09:54):
So does that bring sort of networking architectures into play as to how you manage that? Are you sort of looking at edge-style computing for that sort of thing?
Yeah, exactly. So it's particularly when you're retrofitting equipment, if you're putting a PLC or a sensor on more traditional equipment, then you're looking at edge computing.
(10:19):
And it's actually relatively straightforward to do now with the right architecture. But this is the pro tip that you just mentioned, you have to start as early as possible.
For most organizations, that's not the reality, particularly manufacturing.
Yeah, manufacturing is like the dark ages. I remember going into a manufacturing floor and there was an AS400 computer in the corner that had never been turned off in the 30s that had been there and it was still working.
(10:47):
And that was the sophistication of their technology.
Yeah, yeah, it's amazing. I mean, you think the capital investment required is huge for most organizations and for a lot of organizations, they just don't have that capital available.
And this is part of the challenge, I think, for manufacturing as a whole.
It's very easy, particularly if you're a medium size enterprise, you look at videos on YouTube or LinkedIn of Amazon's warehouses in China and it's completely lights out.
(11:18):
There's robots everywhere. There's one person, there's an enormous warehouse, fully automated. And everybody thinks that that's where they need to get to.
And then the question is, well, how? And the second question is, do you really need to?
Most organizations probably don't need to get there yet. It might be a roadmap that they're on, but actually there's a lot that they can do now with relatively low cost when you look at things like edge sensors and PLCs.
(11:47):
Well, just to reduce their wastage. That's absolutely wastage. That's a really good thing.
Yeah, that's right. But when you put those videos up, you know, the Amazon warehouse, a lot of people just sort of do in the headlights, you know, it's like, how do we even get there when we might be struggling today with cash flow or today with productivity or today to find the right people.
(12:11):
And often the challenge for organizations is to actually take a step back out of the day to day and say, what's the strategy around this? What's actually going to get us from here to here to here to here on a roadmap, rather than feel like they have to go from where they are now to that sort of full Amazon setup?
(12:33):
Well, you know, for every organization, you don't need to be like Amazon, because, you know, it might actually mean be better for an organization to keep people more in the loop.
Absolutely. And this is a conversation that we have all the time around what's the competitive advantage that you're trying to drive for your organization.
Now, because for some organizations, it is cost. So you have to manufacture as cheaply as possible. And therefore, you know, things like production efficiency are the number one goal.
(13:04):
But for others, it might actually be security. So for example, we're seeing a few organizations in New Zealand, actually the UK as well, we're there to winning work from China and other places because of security that they can guarantee through their process.
So they might not be the cheapest solution. Now, they still need to manufacture as efficiently as possible. But actually, their strategic advantage is security. It's not necessarily price.
(13:34):
And so when you start to have that conversation, then that you can then play forward in terms of data or technology strategy, because that's got to be your number one goal as maintenance of data security.
And so outside all of the really big manufacturing organizations, you know, uni leaders and GES and stuff, a lot of those organizations won't have thought of those kinds of things because they're just doing business.
(13:59):
They've always done business. They've always struggled to improve their margins and stuff, but they've never really taken a very scientific approach to it. So does data help lead that conversation?
Yeah, and it really comes down to the culture of the organization. So you if you've got people who are mature in terms of their data capability or even their mindset towards a data driven approach, then absolutely will try it.
(14:31):
But a lot of organizations are fairly immature when it comes to data capability and even how you use data. And most organizations will be absolutely on top of things like reporting, historical reporting.
And actually, how do you use data to forecast to scenario model to predict? That's a very big gap.
In fact, there was a lot of people will be using Excel. Absolutely. Yeah, ridiculous Excel spreadsheets that are so complicated if you put something in there just fall apart.
(15:01):
Well, that's right. But you know what? I'm always amazed at how what those Excel budgets can do.
Yeah, they're amazing. But that's scary because they're so fragile.
Yeah, because of the dependency they create. And it doesn't matter what else you do. If that's what the foundation of the model is, then it's a fairly precarious model.
(15:22):
We've actually been moving a lot of our old Excel models. We had Excel models for predictive models that you press the button on them to recalculate and then you go and leave the machine for two days while it did it.
You know, and it was quite scary. We've moved all of those into our cloud data platform, you know, where we do all the heavy lifting in the cloud, which makes a lot of sense, but they can still have an Excel front end on it so they feel at home.
(15:50):
Yeah, well, that's right. I mean, you know, we do a lot of our visualization work in Power BI. You know, there's amazing things you can do with Power BI that look very simple.
And sometimes people, you know, it's always deceptive how simple the front end looks when you think about what's going on behind the scenes and the, you know, the translation that's required, the transformation of data, the analysis of the data in order to be presented simply.
(16:15):
It's pretty deceptive when you look at what is effectively a fairly simple dashboard on the Facebook.
Are those companies that are building that sort of thing also realizing that they will need to beef up their recruitment and skills in the data space to feed, they buying a puppy, they've got to feed the puppy?
That's right. That's right. And there's really two ways of looking at that. So one is what's the specific technical resource that you need to build in and that might be, you know, data architects, VA's, data scientists.
(16:48):
And for a lot of organizations that that's going to be relevant for some, that's so depending on their scale. But then you've got to look at what's the, what's the capability that's required across the organization, across all roles, and particularly those sort of decision maker roles, you know, team leader level and above, where people are actually relying on data in order to make good business decisions.
(17:11):
And now we're talking about data literacy, which is the biggest elephant in the room worldwide.
Yeah, well, that's right. And, and that's so, you know, lifting up. There's so much in there. It's very a whole other podcaster on data literacy.
You know, there's, there's how do you interpret data? How do you read data? But there's also how do you set up your organization as an ecosystem when it comes to decision making using data? You know, so if you're, for example, we have conversations with HR departments, for example, and HR departments are actually getting really good when it comes to data analytics for
(17:51):
workforce planning, for example. At the moment, that's sometimes disconnected to things like the production plan, and particularly things like investment in automation. So if you look at your investment in a machine, a lot of organizations will factor in purely what the depreciation of that machine is in order to
(18:14):
determine its ROI. When you start to factor the resourcing requirement in, you know, people, total cost of ownership, over cost, relationship, exactly. And in the last sounds like it should be a fairly simple concept. It requires an ecosystem driven
decision making. And it means that it means that it can't be financed driving it because they like that method because it's easy. And it's the way they've always done it. So it's a really, it's a really interesting conversation inside an organization.
(18:44):
I remember, I was at GE early this century, but I was able to say to my CEO, based on data, that an idea that he had wasn't good, and he looked at the data and said, Yeah, you're right. And we dropped that idea. And that, to me, is a good data driven organization when you have that
here is evidence, don't do this. And it's pretty rare.
(19:08):
Yeah, well, that's right. And there's this multiple so maturity, there's the people need to come up on this. So, you know, often, we get really good buy in when the data supports what people already thought.
The challenge comes, you know, what happens when the data, you know, as in, in that case, actually, you know, goes against what was previously thought. And the maturity there is, you know, is letting go of, you know, previous assumption or previous ego even going with the data.
(19:37):
You know, sometimes what we find is, as long as the data is reinforcing what people thought, then they're happy to go with it. As soon as they go against it, suddenly become, you know, the machine whisperer, the only person that knows how to tune that piece of equipment.
You know, the old ways come back in.
Oh, I was just wondering, so what advice would you give to people who are in that situation where they've not presented data that reinforces management's beliefs. So how do you manage that?
(20:06):
Yeah, well, I think that that's, in some ways, a classic change management, you know, people have to understand the context of the problem.
I think where you can put a dollar value on, on information or use a metric that's going to resonate with that particular audience.
Then that's really where you start to talk their language. So in the manufacturing world, they're starting to talk about productivity, over equipment effectiveness, you know, cost of quality failure, those sorts of things that have an immediate impact on bottom line.
(20:40):
And then you'll start to get people's attention.
If you can't do that, then it's, you know, it's about providing the context for that decision and trusting, encouraging people to trust where that data has come from and, you know, the data sources, how that data is collected.
But it's really going back to first principles of change management, because for a lot of people, there's an identity tied up with the ability to make decisions based on, you know, their own experience.
(21:13):
Yeah.
And a lot of the time, you know, if you're contradicting what their, their decision would be.
Or their identity, really, their identity.
That's right. It comes down to identity, right? So their identity in the past might have been that, you know, I know this process better than anybody else. Therefore, I am able to make really good decisions about this.
And, you know, part of what we try and do is actually get in behind those individuals and say, you know, absolutely, you've got that experience. But now here's another tool that you can use to actually improve, you know, the decision making.
(21:46):
You can stay the expert. You don't have to know.
Yeah, exactly. Exactly. Yeah. And, you know, it's about putting those people, you know, front and center and helping them do their job, you know, rather than try and challenge it in any way.
Yeah, well, that's the thing, make people look like heroes with the data, not against them.
(22:09):
Yeah, that's right. And it's something we talk about all the time, you know, particularly it's not just in manufacturing, but manufacturing probably feels there's more than other industries where they've been very labor intensive.
Once you start to automate lines, you do start to eliminate the need for some human, human roles.
But we've seen really good success stories where organizations have retrained people actually to start to use the data to do things like predictive maintenance that improve the production output overall, even if they've moved away from the more hands on.
(22:44):
So it's kind of interesting that we're not seeing necessarily a lot of job losses through digitization or through automation. It's just an augmentation of what those capabilities are.
And the organizations that invest in those capabilities that help people through that and critically, you know, to the point that you made help people shift their identity through that process.
(23:05):
Those are the organizations that are successful.
And that is an excellent note to end on. Thank you so much, Adam. It's been really great chatting to you. And thank you, Adam, from DeCipher.
Thank you so much, Kate. It's been great to chat.
And that is it for another episode of the Data Revolution podcast. I'm Kate Crothers. Thank you so much for listening.
(23:28):
Please don't forget to give the show a nice review and a like on your podcast app of choice. See you next time.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.