Small steps to harnessing supply chain data to enable performance improvements (aka chopping 'data mushrooms')
At the risk of overdoing this metaphor, shamelessly 'borrowed' in a previous article , this discussion explored potential solutions to the 'data mushrooms' and consequent quality and integration issues that we identified as being significant obstacles to supply chain improvement initiatives, in particular, for inventory optimisation and replenishment.
Summary
Data can be a major constraint to supply chain improvement and transformation initiatives of various types. In this discussion, we focused on how this manifests within inventory optimisation and replenishment, especially related to raws and packs.
Some of the participants reported frustrations with ERP-based solutions which, although generally useful, don’t offer the desired functionality for inventory management and replenishment which leads to lots of work being done on idiosyncratic spreadsheets, out of the system, which often adds to the confusion and siloed decision-making. Others are focusing on the forecasting element because, without that inventory visibility, forecasts can resemble wishlists which are hard to challenge without accurate and reliable data.
Underlying these initiatives, many are implementing group-wide data platforms to integrate data and get a single view of stock as a necessary starting point. Some, at more advanced stages, are now able to extrapolate the data and generate projections using machine learning and AI for scenario planning.
A key requirement is the ability to track exceptions and have visibility of excess stock and shortages. The lack of a unified capability leads to different divisions, departments and individuals coming up with their own ways of defining and measuring exceptions.
Two common factors underlying these kinds of challenges are:
Although some ERP-based systems may lack the desired capability, at the same time, ‘best of breed’ solutions are overkill for many businesses so attention returns to doing a better job with what you have or point solutions that can deliver the functionality.
Some relatively simple steps that can be taken include:
Anonymised transcript
JP?
Thanks to those of you who were able to provide some input. Hopefully you can see the graphs up there. The first point on the agenda is really just to talk about business drivers around data. I just wanted to show from the responses we had, this is what B and I went through yesterday afternoon just to try and orientate ourselves, where the centre of gravity is among the discussion participants. The top one was inventory optimisation, followed by supply chain visibility and then improved internal & external collaboration. That's what we're planning to focus on and just get into what data issues, what constraints people are having.?
A, welcome. First of all, just because you're the first one up here alphabetically, and I think you were one of those who mentioned inventory optimisation as being one of the things that you'd like to be able to improve, and that data was to some extent at least a constraint in being able to do that. Would you be prepared to just elaborate more on what you shared on the survey input to get us going??
A
Yeah, of course. I think one of the main issues is data, as we're all kind of experiencing, I'm sure, but it's how people are using that data. Obviously, XXX has predominantly been a B2B business, whereas now we've moved over to more of an ecommerce based platform. Having the demand planning, having the forecasting, having the stock positions available to the right people has been a real challenge. XXX has gone through quite a dramatic change in terms of the systems that it uses and the premises, because we're in the middle of moving, so there's a lot of variables going on. I think from my perspective, having only been with the business for a year, is understanding a bit more about or looking for best practice of how I can share the right data with the right people and understanding a bit more about other systems and protocols that may be out there that I'm just not aware of or need some guidance on.?
For me personally, the data aspect and the inventory optimisation is more around demand forecasting, demand planning, and then making sure the right stocks are in the right place.?
JP?
Thanks for that, A. I'll just get a few more people involved here who also highlighted inventory optimisation as a key driver. D, if you don't mind joining in, you highlighted that as being one of the key areas of focus for you. Could you similarly just expand on that, please??
D
Yeah, with pleasure.?I'm the Logistics Director at XXX. We've got an interesting and diverse product range and about eleven and a half thousand SKUs at the moment. We're omnichannel and we have very limited warehouse capacity, 86 stores and a growing ecommerce business. Multifaceted challenges for me: physically storing the stock is a big problem in the warehouses, understanding the layout of the warehouse in terms of the type and size and profile of the pick bins that I'm using, what stock needs to go where? How do I replenish that stock efficiently and understanding then around promotional activity, the movement of that stock within the store portfolio, what stock needs to be where and in what depth. Our WMS and ERP system is XXX…it’s is very good at some things and not so good at others.?
It tries to be all things to all people and it's not doing it very well. We look at store allocation, for example, it's quite simplistic in that it runs on a min max and days cover. It's not as intelligent as perhaps we'd like it to be. Ultimately what that means is we're overstocked on a very large portion of that inventory and we're carrying too much dead and terminal inventory as well. We're very good at looking at what's coming in and dealing with that. We're not so good at dealing with the 35% worth the stock I've got that sits there gathering dust. There's a lot of challenges for us and we've got loads of data. Everybody's got a spreadsheet, everybody knows how to use their spreadsheet, but nobody's looking at all of those spreadsheets and making collectively sound decisions. So we're quite siloed in that respect.?
Buying and the Merchandising team are doing one thing, finance are over there doing something else, and me and my team are the buggers who have to try and piece it all together in the middle. So lots of challenges. What I want to understand is, similar to A really, what solutions are out there that I haven't seen? What's the right approach? Is there a right approach? AI is a new term. Everybody gets excited by the letters AI, but I don't really understand it if I’m being honest and I'm quite techy, so that's what I'm hoping to get from today.?
JP
Thanks D, we'll get to those points in a minute. H, I can see you nodding. You obviously recognise quite a lot of that. You also highlighted inventory optimisation. Can you tell us a bit more about that, please?
H
For XXX we've got about 400 shops, but we are growing our services business. For us on the retail side, we are very much XXX. It's our ERP system, an old version of XXX. For WMS, we're just putting XXX having had Manhattan before. From that point of view, we're looking to do a number of things: one is change not the ERP system but the forecast engine and the replenishment brain of that. We've been out to market looking at some of the top companies in that space because we feel that's what we need, because we're in a situation at the moment where everybody who's using XXX, it's not flexible enough for what we want to do.
It's great at the actual transactions, but not really in forecasting and planning space. We've got a number of people, much like you suggested, D, where people are off in spreadsheets building their own humongous models and then wonder why Excel crashes. In the background, what we are doing is putting in a group data platform, which will be great in essence for our services business, where we've acquired a number of smaller companies there, all with their own platforms and things. We need to integrate that data so we can get one view of the stock so that you can then start doing things like optimisation and things like that. We're just on that motoring side struggling to even see what inventory we've got where. Having a group data platform is a great enabler of just getting that first stage visibility and we are embarking on XXX for that, alongside using tools like XXX to present that data to people.?
I think we've got the right plan moving forward to be able to deliver those aspects that we're looking for. It just takes time to get there and it's all about data accuracy and avoiding double counts and things like that and making sure the definitions of data at all the various stocking points are all saying the same thing. XXX is great at looking at data based on movements. In other places, they don't even capture movements, they just show an end stock position. It's trying to meld all those bits together. Until you've got the data, optimisation becomes just a wish list for you to tackle later on. It's all about getting hold of the data first and then having the right tools to do the right things with it. I don't think once you've got the data being complex about doing something to optimise, you don't need to be complex.?
I think a lot of it also depends on your rates of sale as to how complex you feel your business needs to be. That's the journey we're on at the moment.?
JP?
Brilliant. Thanks for that, H. I'm sure this discussion is going to kind of bleed into the other areas around visibility and collaboration, but I just want to focus on inventory optimisation for the moment. C, if you wouldn't mind, I wanted to ask you because in a recent discussion, you coined the phrase (which I've been presenting as my own!) of ‘data mushrooms’, so thank you for that. This was an area that you felt was well worth focusing on and that you'd actually dedicated quite a major project to getting your data in order. Would you mind sharing to what extent you've had the similar kind of issues that have been mentioned just now and how you're getting on with trying to crack them??
C
Sure. All of what I've heard is the same story, different words, different acronyms, but very much the same things that we've been hearing. I work for XXX, I've had global roles across segments. We have different divisions with slightly different routes to market, slightly different customers, but very much the same issues when it comes to the things that we're talking about. One division has started pioneering this, trying to look at our data foundations and a lot of that includes the inventory side of things. We've started very much in the planning areas along with inventory and the optimisation and what that should look like. So we've done two things. We started our journey very much with inventory optimisation tools and what we then found was I actually need better data to feed them.?
We've gone back to the basics of where is that data sitting, where is it duplicating, where are the sources of it, putting together data layers underneath things like our data lakes where we are not manipulating data. We've gone back really to the basics to build that, build on top of that. Things like our KPIs, how we measure our inventory, looking at AI over the top of that to look at how we can extrapolate some of those projections, bringing in warehouse planning, all the different levers that we can. Have demands and elements into one place so that we can try and then look at how we can use those levers in a different way to the way we do today. Everybody is the same. I think, in as much as your stock value is going up, we still have the same pallets in, but the value goes up.?
My finance guys say I need to keep that value at the same as it was last year. All of the inflation and everything else is hitting it. What are the other levers we want to use? So, inventory is where we started, just try and showcase those data foundations underneath. It's a long process of getting data in the right place to bring it through and it doesn't show things very quickly, but we are getting there. So, long story short, they have now brought together that data in one place that we're looking at. What it's telling us also that is we need some simpler tools to look at safety, stock management, how we actually manage that inventory and the basics of our parameters. We're also then putting on things like the scenario planning elements to say how can I build those levers over the top? It is important to us and it's one of the real hot spots for all of our segments.?
We're all trying to not reinvent the wheel but learn from each other and actually bring the whole company forward in those different areas. It's a big project for us and a big focus area for the next, probably going to be in the next couple of years to get our full data level sorted. The inventory piece is definitely coming first because that's where people are saying we need to focus most.?
JP?
Thank you so much. C, before I bring B in, is there anybody else that would like to explain any other challenges that they're facing with regard to inventory optimisation so that we can make sure we're addressing as much as we can on this point too??
T
Yeah, apologies for not filling the pre-survey, but I think one thing to add with inventory management is the ability to track exceptions to actually have the visibility of where you have your excess. Is it in your stores, is it in your DC, is it in a replenishment, is it in an RC? I feel like part of the legacy forecast and replenishment tools, they don't really give you that visibility of where your excesses are, but also where your shortages are. I feel like a lot of our analysts at XXX. and here at XXX, they have their own ways of measuring these exceptions and often they only come up when it's way too late i.e. we got ten years worth of stock of this in the DC or we're out of stock so go and sort that out. That's something that we're trying to address now.?
H, I think we're kind of on a scarily similar trajectory with XXX here as well, also with the supply chain visibility tool and also with the lake. Definitely similar initiatives and yeah, I think that exception bit I think would be really where the unlock would be. So I think that's my tuppence.?
JP?
Lovely. Thanks for that T. F??
F
Similar challenges with XXX trying to build an XXX data lake, trying to set the right levels of inventory. The additional challenge we have is there's quite a lot of what I would call emotional inventory settings around the world managed through orders, dummy orders and things rather than the safety stock settings. You look at the fact of how much inventory you've got, which is very easy to get out of XXX, when you start adding the order book and saying, well, that all looks like it's going to ship, next week, but actually, a month later, it's still not shipped because it's still sat there withheld orders. How do you build the additional data on top of the facts that helps you analyse those exceptions you're just talking about? Is that ten on hand actually ten years worth or is it going to sell next week??
How do you add additional qualitative information on top of the facts to help you make better decisions about those inventory levels that you've got.?
JP?
Thank you F, anybody else before I turn to B. No. Okay. B, thanks for joining us on this one. Maybe before you kind of give your response to the points that have been raised, just give us a brief introduction, your background, what you're doing now.?
B
Yes, hi everybody. The bit that we're focusing on is the data platform, data engineering part of this. There are lots of clever people that can do data science, do the analysis, and do the visualisation of data. As you've all said, getting good quality data in the right place, available to the right people is a critical aspect of it. My background, a lot of background in ERP supply chain manufacturing operations and distribution wholesale type businesses and one of my bugbears I guess, is that the ERP world has kept a lot of this data locked up and not very accessible to people. The other one is the whole business of what we set up: if you take just a basic MRP engine, it's really just a big calculator, but it's a calculator that works off a set of predefined parameters.?
We typically set those up and then we ignore them. You talk about min max levels, lead times, order quantities, minimum order quantities, all those different parameters. Whether you're in XXX or whether you're in XXX, XXX input, whatever the number of those variables that you've got, the number of those parameters you've got to play with increases. There are way more in the XXX instances than there are in your basic XXX thing. The fundamental thing is we don't look after those parameters. I don't think we look after them well enough. One of the things that we're at the proof of concept stage with at the moment is just a utility that looks at the overage, is the underages, the excesses, where you've got stock that sits within an ideal range. How do you plan to replenish that? How do you plan to order it? What periodic review should you use against individual parts??
Rather than using the very easy ABC Pareto type stuff or nine box model, then the model is flexible enough to be able to seven, nine, whatever. It's the thing you would run periodically to check that you've got the different SKUs in the right categories. I think the other bit is what's the time horizon that you want to work with your data on? At the moment we’ve only really talked about internal data, what inventory you've got, where is it, what the parameters are. It really depends whether you're trying to drive operational intelligence, what should you be doing today, tomorrow? Are you trying to pull it through so that you can look at long term, medium term trends? Are you trying to look at the longer term horizons to build and plan networks, capacity where I should put my warehouses? Those broader strategic decisions?
We've partnered with and got to know quite a few different companies. I won't go into them now, but if anyone wants to ask questions about potential solutions, particularly against the backdrop of XXX. My background is XXX. I started at XXX in 1997. Well, I largely got disenfranchised by the whole ERP world and one of the projects that we did pre COVID was working with XXX, looking at their demand-supply planning engines and the things you said before about XXX's existing tools being pretty clunky and the new ones are probably overkill for what most businesses need. We went through a whole process of trying to look and identify tools that might be useful for a fairly large, fast moving food production unit. The bit for me though, is that parameter management. I think the other thing is that we tend to look at things as just one holistic blob, if you like, and we're not very good at breaking it down to that whole thing about eating the elephant.?
Understand what is controllable, what is not controllable? What could you have anticipated? What is a complete black swan or whatever term you want to use? Breaking things down so that you start to identify where your problems come from. Upstream variability that actually you can do more to control: suppliers not delivering on the lead times you expect them to deliver. You could just adjust the lead time and say, okay, well typically it comes in ten days rather than five days, let's just update the parameter. It actually might be better off to work with the supplier to drive that down so that they are consistently delivering. Every time you get any of that variability in your process, it has all that downstream effect. Looking at tools that liberate data from within ERP systems particularly.
JP?
That's great, thank you B. I'm going to open it up now and take a step back. Invite you to either put your hands up or just unmute and jump in to ask questions of B or anyone else.
F
Yeah, a quick question on that, B. You talked about min max, what I would call the dynamic settings and this is a question for everyone really. Do you have challenges with just master data settings that I would say are the ones that control your distribution network? How you choose whether you buy something from another location or whether you buy it or make it. I think that there are the dynamic settings but there are also the static parameters that control the way your supply chain is replenished or managed and whether they're active or inactive. I calculated we've got about a quarter of a billion data points if you look at absolutely everything. You narrow it down with, well, only these items are active and only these items actually have sales orders at the moment. How on earth do you work out which of those quarter of a billion you should be looking at??
Do other people have those same challenges with their master data, both the dynamic values and the static values? I see a few nods.?
A
Yeah, we have the same issue. A lot of is identifying the dead and live data points. You're absolutely right, F. One of our biggest challenges is we've got a lot of people looking at data points that there’s no need for and there's no reason why we haven't dealt with them. You could call it lazy data management. I think from my perspective, it's just understanding how we can structure the data so that we can identify those dead links, clean them out. We certainly don't have the same value that you do, but it's still the same problem. So it's a very valid point.?
B
I think in terms of prioritisation, we tend to focus on annual usage values, adding that financial dimension. I think the other bit is you've also got that challenge of do you look at your potential forward demand by trying to forecast the demand, or do you look at historical sales? Different businesses work in different ways on that. That bit about not addressing, if you like, the detritus that tends to accumulate within systems, I think that comes down to that very cultural thing where you need to treat the behaviours and not the condition.
T
What do you mean by conditions?
B
Let's say that you're constantly out of stock or something. You could come up with a solution that tries to address all your out of stocks. Actually, the reason why you're out of stock is because people aren't recording inventory accurately enough.?
T
Yeah, I see.
领英推荐
B
The behaviour bit is to address the recording of inventory accurately rather than trying to treat the condition of a symptom of that poor behaviour.
T
I have a project that we're now working on to attempt to reduce lead time and to reduce MoQs in order to reduce and also theoretical stockholding within the planning tool. However, about 70% of all of our orders are placed manually. We can go and change that lead time. We will go and change that MoQ, absolutely no problem. If we don't address the fact that everyone's placing manual orders and whatever they want to order then nothing is going to change. What we did is I took a relatively simple planning tool that's called XXX and I've taken data out of XXX, out of our data layer and put it into XXX. All you need is some demand and historical demand stock levels and some rudimentary master data. You can then kind of start to model on a high level how much stock you need and when you compare that with the current stock you just find that there's so much inventory to be saved by just ordering and optimising your order workflow. You can change your settings as much as you want, but if you don't change your behaviour it's not going to be much budget.?
C
I found very different approaches to process and ways of working and the strictness towards those pieces of the dead wood of going out between finished goods and raws. In our business, finished goods had a much better process of dealing with end of life. When we looked at raws and packs, you are still building if you like the process to make sure that we see all of the information together, the coordinated processes. We found a huge amount of raws that were just sat there with no demand going forward. I've talked about the big stuff and the big overwhelming foundations and data lakes. This was a very small piece put together in a spreadsheet pulling information from XXX and very much as B was talking about. Bringing together the information you needed to look at, is it still used? Do we still have forward demand? We looked forward demand versus the back demand on these particular raws and packs, finding them in a bit of a segmentation so, looking again at cost, at size of pallet spaces it was taking up, those types of things, looking a bit at lead times as well…we have ocean freight as well, so they're very different. What are those lead times? What should they be? These are just tabs in a spreadsheet. This wasn't anything too fancy. Gave us huge insight to see where to focus going forward. It's definitely worth delving into that information you have it available. It doesn't always need to sit just in XXX, bring it out and do some playing around and actually with basic Excel skills can give you some really good insights of where to focus next.?
JP?
Thank you, C. H and then M.?
H
Just a bit of a build on T's theme, I guess, and B’s. I think the challenge we have around XXX with people going back onto spreadsheets is that just that bottom up visibility. The downside of not having that bottom up visibility is when people are forward planning for the budgets and things like that, it becomes a wish list. You've got nothing to challenge back and say, well, the bottom up is suggesting this the gap, how are we going to deliver the gap? What are the events, the promotions, whatever it may be in a retail environment to close the gap? If you don't have those behaviours and drive that conversation, you end up under and over ordering for the obvious reason you can't see the gap. I think getting data into a data platform is one thing, but being able to put some rudimentary maths over it as, T, you described to support you doing that, which is the journey. We're trying to get this forecast or a replenishment platform that enables everybody to be one solution with the flexibility to offer the things that are forcing them into spreadsheets, enabling us to get that bottom up view to do some comparison. I agree that I think it's the behaviours and other people in the business on the trading side who are coming up with what our future budgets are and things need to be tampered by seeing what the natural progression from a bottom up plan is going to deliver.?
JP?
Thanks. M, did you still want to come in??
M
Yeah, just a couple of reflections from my side. I work for XXX, so durable consumer goods. More than 200 factories worldwide. The impression I get is that not everybody is really doing inventory planning based on a forward forecast. I get the impression that some or at least parts of your planning is done looking backwards. Is that correct? Because some of the problems you have been talking about don't entirely resonate. In XXX, we have a very strong focus on S&OP, and that includes tactical planning. There is a strong leadership involvement on tactical planning and measuring results, at least on bias and depends a bit on the granularity. Not everything is accurate, but at least bias is a strong focus and then parameters are key. So rubbish in, rubbish out.?
I think, B, you said it, we actually use XXX. Is it the best? No, it's not. Are we so mature to go to for best of breed everywhere? No, we are at least not that mature. We really try to get to a high maturity on what we have. Mainly XXX in our case is good enough. We actually use the XXX inventory optimiser and do review and assess the input parameters quite a lot. There's a lot of pressure on the quality of the parameters that in turn does enable us to steer the inventories actually quite well. That doesn't mean that we don't have inventory problems. We do have inventory problems on finished goods. We do have them, especially in the recession. We were over optimistic on the economic environment, of course, and we had to bring all those down. Instead of focusing on service levels, we actually, during a period of four or five months, focused on days of sales now, for example.?
More an asset priority than a service level priority, which we normally have. I don't know if that was helpful, but this is a bit how we drive inventories. I think on finished good inventories, we do have quite a good grip, to be quite honest. In those areas where we have, let's say, XXX in the full range, we have a lot of legacy. That's a different question, I think.?
JP
Thanks, M. G, you have your hand up.?
G
Yes, it was actually more about what M was saying. That was resonating more with me. We're in a slightly different situation than some of you. Over the last three years, we've reduced our inventory to all time low levels and we're now in this position, within supply chain, where we believe we are at the minimum we can go with our current flexibility. However, we have both pressure coming from the leadership of the business, which doesn’t want to see inventory grow again because we've seen the benefits of the operating cash flow opening up. We are now having pressures from the market, where we have been very fortunate over the last few years, where we've been in a very short market. Therefore, customer service was not as much of the priority and we focused more on inventory reduction and trying to be as flexible.?
With the talk of the recession and everything coming, we will be in a situation where demand will not outstrip supply, we believe. What we want to focus on, I suppose, is using all the data we have and maybe touching a bit on the AI, but we don't have the facility yet, to determine what is our optimum inventory right now. We just have a gut feel and knowledge within people's heads of, well, I can't go lower than this, or we stock out and this is the stock and we're being able to trade off and say, well, we can supply this for the customer, but they will be out of stock of this for a number of weeks or months and the decision is made. We need to come to a place where we can articulate to the business: this is actually the inventory that we need to be at, either from a bottom up point of view or a segmented point of view to better challenge how we can then swing that focus back to customer service because it will not be as simple as, oh, will just build out inventory.?
I feel the pressure will come to manage the assets. In some instances I feel we need to be able to use this data to go back and say this is actually what we need to be holding.?
JP?
Thanks, G. O?
O
I'm not sure how my comments are going to help, but it's just a reflection really. My background is Food Service Logistics, where I previously worked for XXX for 18 years, where our customers were people like XXX and XXX in this country. I worked for a company called XXXs who are European and again have a similar portfolio of customers XXX, XXX, XXX and the way we work is there's a list of products that their restaurants can buy from us. We manage the suppliers, we manage the inventory, we pick it, pack it, distribute it. Everything's normal, that's fine because you can quite easily forecast how many chips you're going to need for theXXX restaurant, all that kind of stuff. It's where things that are different come in and affect that. This is the bit where I'm not sure if it helps, but what we were able to do with our customers because we have very intense dialogue, the relationships are very close to talking with them every day, we will say your Normal run is X on these particular products, but it's now Y and the implications of that is that we're going to fill up or we're not going to have enough.?
What you want us to do about it? That's why it's not particularly helpful because we kind of moving the problem to the customer in that case. It does make for a relationship that's quite dynamic. There's an understanding from our customer, from someone like XXX or XXX of what the implications are of the poor intelligence that they provide to us in situations like promotions or new product launches. I just wonder if that point that someone earlier made around discipline could be the solution there…I don't know? In the case of your XXX depots, if there's a bit more of a tighter relationship between the replenishment team and the people who are ordering willy nilly and filling the deposit stock that they don't need in the same way that we have very close relationships with our customers.?
JP?
Thank you for that, O. M, you did have your hand up? Yeah, that's just one question which I found very interesting to B. Just to offer one more insight here how we do it. Of course, we bring everything into a global data platform as you guys do. We then visualise it with XXX and actually maybe something interesting, if it becomes a certain excess actually in our company, we convert it into a loss. That means it is added as a cost. For the sales companies and the sales planners, there's a huge pressure on having the right stock because otherwise it's affecting their monthly P&L. You can imagine that's an interesting driver really of trying to anticipate excess stocks down the line also because they need some time to activate, of course, any promotions or anything out of that.
B, to come back to your points, I found it very good how you brought this together and see if I understood it right? You say ERP and other systems are like a big calculator which I liked. Secondly, it depends on what parameters, how you set the parameters, which I like as well. See if I understood it right then you identify problem areas based on that? That means based on performance metrics, I would guess. What you then do is you simulate better outcomes by simulating different parameters. Is that how you do it? Hence identify maybe the biggest parameters which could drive the biggest improvement and if there's anything which could be changed than in the real world, is that right??
B
I think, ultimately, I'm not sure that too many people have actually got to that last bit yet. It's about the levers. Which levers can you pull and which ones have the most impact? I think that's still a way off for most organisations. I think the main thing is to identify what are those different levers and to start doing some basic sensitivity analysis on them. If, for example, you change order quantities, if you change lead times, if you change service levels or whatever, the sensitivity analysis, which is where I think this idea of bringing things into a data platform which can be then used by those people that can do something with a bit more data science. If you've got it in the data platform and it's clean and the rest of it, then you can start to apply some of the machine learning and other techniques.?
All of those techniques are useless if you don't address the point that you've all made, which is that the data quality needs to be there. You need to have the data and then you need to have the quantity of data and you need to have looked after the data and the master data, the metadata, the parameters that you've got, as well as the quality and accuracy of the transaction data. There are tools out there that can go to that ultimate level, but I think for most people come, back to that…start simple. If you can remove the variability, it's a risk mitigation strategy isn't it? You choose, whether you accept, O, as I think as you said, you go back to the customer and say well this is the problem. Effectively it's just transferring the risk onto them isn't it??
It's no different to normal risk management approaches really. Do you accept it, do you transfer it, do you avoid it or do you reduce it? Those are the four main strategies that you have for managing risks and disruption. I think I would say that sensitivity analysis bit I think is a step on but start off with really just understanding what can you control, what can't you control, what variability is there? Particularly the upstream variability. If you get variability in demand, I think the last session, JP, one of your previous sessions looked at probabilistic forecasting techniques rather than your typical linear regression type models. I think that has a part to play for people. Again, it's not treating everything as if it's one thing and we have to apply the same technique to all the universe of problems that we have. Segmenting it, classifying it, breaking it down, applying the right tools and the right techniques to the particular segments, I think breaking the problem down.?
M
Thank you. That made it very clear to me.
JP
We’re into the last ten minutes or so and just wanted to circle back to where we started and make sure that we're getting some answers or making some progress. A, I started with you at the beginning of the session and hopefully we're touching on that data linking. Are some light bulbs going off or are there other questions stemming from this conversation??
A
No, definitely light bulbs going off. I think there's two heartening points that have come out of this. It's actually good to see that there are other brands having the same issues and it's not just poor management of what we're doing here. Whilst it's not nice that we're going through it, it's nice to see that this is a widespread problem and not just for us. I think what this really highlighted to me is something that I knew and I've been trying to impress on the team here, is that it's the data management that I think is the issue, not how we get that data or how we present that data…crap in and crap out. I think the reason we're getting poor reporting or poor results in certain areas, certainly within the supply chain is because of the quality of the data.?
I think actually what this conversation has given to me is a bit of an appetite to go back and cleanse what we've got first before we start looking at how we move forward, whether it be with expensive systems or basic systems. I think my issue is poor data before I start looking at anything else. I think listening to you guys has really helped solidify that and it's a stupid, really basic thing. Actually I think I've lost sight of the importance of that data being right before we can start making decisions on it. This has been really useful for me personally.?
JP?
Thanks A. Thanks for that. D, could I ask the same question to you? What thoughts has it provoked or any other questions that you'd like us to address??
D
A lot of things. I think certainly the similarities in there, which again is encouraging. I think there is a business behaviour issue here, isn't there, that's coming out departmentally. How do you take a business on a journey to do something like this? I am looking at it through the lens of warehouse optimisation. Where do I put the stock? The buying team are looking at it from a margin and sales figures perspective. We know that we've got challenges and we know data is one way to resolve them. What is, I think, coming out amongst everybody's comments is how do you start the journey? Where do you start the journey? How do you get people on the journey? Who owns the journey and how do you change, ultimately, the core business behaviour on what data have you got, how accurate is it and how you use it??
My question perhaps has changed from 55 minutes ago about not what solutions are out there, but how do we understand what solution we're looking for in the first place? That's the big conversation that we need to have. I mean, ultimately we still need to talk to vendors, we still need to look at solutions so that we understand what's available and how we might use those things. I think we do need to have a little look in-house about what have we got, why have we got it and what we do with it.?
JP?
Thanks, D.
B
Sorry. Can I chip in there?
JP?
Yes, please.?
B
Sorry, D. I think the bit that always strikes me is people not being very clear about what problem are you trying to solve, rather than what's the solution? Again, it's about identifying the root causes and the problems. I mean, it's basic stuff. It's basic stuff that we forget and I think the other bit, and I don't know how it applies in your businesses, the other thing that I've noticed as well is that we probably invest a lot on the demand and the forecasting and the sales side of things, right? Your average material planner and scheduler, the person that's calling in the orders and the rest of it, I don't think we've invested enough in those people for them to understand the art and the science and the maths behind the engineering, behind managing inventory and managing material stocks and that ordering process.?
I don't think we do enough to educate people at that upstream end of things. I think there's a great tendency to employ the biggest and brightest to try and come up with more and more complex and sophisticated forecasting, demand planning models and employee data science. We forget about the fundamentals at the other end.?
JP?
Thanks, B. Just in the last few minutes we have, actually D mentioned the issue of ownership and that's something that we talked about yesterday, with B, when we were thinking ahead about what are the common issues that are likely to come up. I think we framed it as the supply chain often owns the problem, but they don't always own the solution. I wonder whether you have any nuggets of wisdom in how to address that ownership problem and how to get people on board?
B
I think it was M who implied that there's a comprehensive collaborative planning process, shared objectives, is that right, M, in your business? Unless you get that conversation up and understand how the cost of buying, the customer service, the financial implications, whatever it might be, how those different priorities for individuals map against each other to decide what's most important for the organisation, it’s difficult. But it's a journey of maturity as well. You've got to raise it up a level, I think, in order to have a view on what is the principal set of drivers or constraints that the business needs to work to and therefore who needs to be subservient to those and then gave the alignment of the measures further down so that people aren't individually motivated within different departments.?
JP?
That's probably one of the most common topics that we have around S&OP and IBP and trying to make it not just a supply chain thing so other departments feel that there's an ownership around that. In a couple of minutes, any other questions, comments, reflections before we wrap it up? It looks like we've answered everyone's questions, I think, B! Any final thoughts yourself??
B
No, just, we've been doing quite a lot of work on the whole topic of liberating data from ERP Systems, particularly XXX, so if anyone wants to have a conversation about some of the tools and utilities that we've identified I'm very happy to have that conversation.?
JP?
Thank you very much. As usual, what I will do is once I've written up the notes, I'll send an email around sharing contact details so that you can pick that up with B or indeed anyone else that you might have identified common cause with on this session. We're going to give you a couple of minutes back at the end of the hour for a cup of tea or something. So thank you very much for joining.