What happens when traditional forecasting isn’t enough? In this episode of GAINS On, host Joe Davis sits down with GAINS product owner Aurellia Sudihardjo to unpack what makes demand prediction different, and why it’s become a must-have tool for today’s planners. From regional spikes in demand to changing market signals, this conversation covers how GAINS’ machine learning-based demand prediction delivers insights that traditional methods miss. You’ll hear how companies are using it to get more accurate forecasts, smarter inventory planning, and better business outcomes.
Three reasons to listen:
● The key differences between forecasting and demand prediction
● How external signals like item features or material costs inform predictions
● Why accuracy still matters—and how better predictions help unlock working capital
Be sure to subscribe for more expert insights from the ever-evolving world of supply chain
strategy and planning.
Joe (00:01):
Welcome back, business buffs, transformation trailblazers, and the tech curious. Joe Davis here, your Friendly Neighborhood podcast host fearlessly guiding you through another episode of GAINS On the podcast that unpacks the future of supply chain strategy, planning and design. Powered as always, by the brilliant minds at GAINS. Today we’re turning our attention to a deceptively simple question. How do you know what to plan for? We’re talking demand prediction. What is it? How is it different from traditional forecasting and why it’s become a must have for planners navigating unpredictable markets, shifting demand and increasingly complex portfolios. To make it all make sense, I’m joined by GAINS product owner, Aurellia Sudihardjo, who’s leading the charge on our new demand prediction solution. We’ll explore how machine learning unlocks smarter forecasting, how external signals like material costs or regional trends, shape predictions and why accuracy isn’t the whole story, but it sure helps. Let’s get into it. Aurellia, welcome to the show.
Aurellia (01:04):
So good to be here.
Joe (01:05):
Excellent. So tell me, what do you do at GAINS specifically?
Aurellia (01:10):
I’m a product owner here at GAINS, so I help the product team develop new services or new products that can be backend service or improvements in the UI, things like that.
Joe (01:25):
Cool, that’s great. I know we just rolled out a new UI not too long ago with GAINS X. Is that something that you were part of?
Aurellia (01:34):
I, I joined the company a couple years after that was developed, but I was involved in minor improvements here and there.
Joe (01:45):
Oh, I’m sure they were more than minor. Your contribution, they may have been minor changes, but I’m sure your contribution was major. So the reason I brought you on the show today is I want to talk about GAINS is rolling out a new demand prediction product and I guess you are the product owner for that, right? You’re our subject matter expert for the demand prediction product.
Aurellia (02:09):
That is right.
Joe (02:09):
Alright, well first lemme start off with the real basic question because as the audience knows, and you’ll likely no doubt, we’ll find out. I’m a bit of a novice when it comes to supply chain. So I guess to start things off, my question would be what is demand prediction?
Aurellia (02:24):
So demand prediction is the name of our product at the very basic level. It is a ML driven model forecasting tool
Joe (02:36):
And a forecasting tool, if I’m understanding it, that’s like forecasting sales or that’s forecasting demand,
Aurellia (02:44):
Forecasting demand. So I’ll give you a forecast of one or two years of what your SKU location demand can be.
Joe (02:55):
One of the things that I know is that we’re able to look at things at the SKU level. So let’s say I, I always come up with some ridiculous company, but I always end up on dog toys. So let’s say I work for a dog toy manufacturer and for whatever reason people in the northwest part of the country are just going wild for my fire hydrant dog toy. And so I can say that this fire hydrant dog toy sells really well in the northwest and then based on that data, able to route things to the northwest based on my demand prediction. Is that right?
Aurellia (03:30):
That’s kind of right. But not in traditional forecasting sense. In uh, just to go through what the major differences in this new ML demand prediction product and traditional forecasting is that traditional forecasting only uses your past sales. So in your example, they wouldn’t know it’s a northwest specific spike in this red fire hydrant toy. But with demand prediction, because we include other features like let’s say item description or locations, our ML model can pick that up, say there’s going to be an uptick in your forecast because this item has these features.
Joe (04:19):
I see. Okay. So you’re able to say, for instance, I know that there’s a hot seller, my dog toy fire hydrant is a hot seller, and the item description would be red, would be rubber, would be novelty size object. Right. So then is GAINS able to take the description of that project and product and say, because it shares characteristics with this top seller, your likely to sell more of it?
Aurellia (04:52):
Yeah, something like that.
Joe (04:53):
I nailed it. Demand prediction, it sounds pretty important, but I imagine it’s more important of a topic now than ever right just with all the kind of wild fluctuations we’ve seen in the market and economy and all that.
Aurellia (05:12):
Oh yeah, yeah, totally. Because with the recent volitility, our past sales doesn’t reflect that. Right. So we would need other signals to help us to adjust our forecast and demand prediction has those.
Joe (05:30):
Okay. Well what kind of stuff would you use to, I mean other than the history, what I sold last year I guess would be a very simple sort of idea. I’m going to sell things that are summer related more in the summer than I do in the winter. So I can kind of look at history for that, but what kind of other data would I look at to help inform that prediction?
Aurellia (05:53):
At the heart of it, it will take again the history or your other SKU location features. That can be your material description, the locations, maybe cost is a factor as well. But aside from that, we are also able to take external data. So with your example of a red toy, maybe the material of it, we are able to take data from outside like FRED or Y Finance and focus in on this maybe red plastic as an external data and feed that data as if it’s one of the features that might contribute to your predictions. So when there’s a dip or an increase in this red plastic, depending on how important that is in the model’s point of view, that might influence your predictions at the end of the day.
Joe (06:50):
So who in the company, in my company, my dog toy empire, who in my company would be using demand prediction? Would it just be purchasers or supply planners?
Aurellia (07:04):
Yeah, it would be your supply planners. At the end of the day, we give you this prediction, you would need to review whether or not you agree with it, but it would be your supply planners, anyone that’s interested as well in the data of how and why there’s an uptick in your forecast.
Joe (07:23):
Gotcha. Okay. So I think I have some understanding of what demand prediction is, but what might be some of the misconceptions that people have, particularly supply chain leaders might have about demand prediction?
Aurellia (07:37):
There is talks that sometimes in the industry that forecast accuracy isn’t everything, but I think what we realized in GAINS as a company is that yes, accuracy, forecast accuracy isn’t everything, but it is a metric that has a downflow effect or a downstream impact on your inventory. So at the end of the day, it’ll affect returns on your inventory, so it’ll make you buy more accurate and if you buy more accurate, you’ll be able to either save that money or funnel that money into some other initiative. But demand prediction is aimed to increase your forecast accuracy in which will hopefully better your inventory at the end of the day.
Joe (08:30):
So it sounds like it’s not, well, it’s something that’s important. You talk about forecast accuracy and how it is important, but is accuracy the most important thing in a forecast?
Aurellia (08:43):
Yeah, I would say so. I would say if you have a high forecast accuracy, you’ll buy better. You have the right items in the right time so you’re not holding onto your items or an indefinite time. Or if you have a big client in Northwest and suddenly you don’t have the items you lose sale, that client will probably go to your competitor.
Joe (09:06):
Right. So in talking about the factors that demand prediction addresses, when you look at something that has, or say you’re in an industry that is really heavily seasonal or is promotional driven or if suddenly demand unexpectedly spikes, how does demand prediction adjust for that or allow for that? I guess I should say
Aurellia (09:33):
In demand prediction, you are able to feed in those historical promotional data that you have. So we would have after in these promotions into your forecast or at the end to your two years predictions. But at the end of the day, the demand prediction allows you to see behind the scenes as well, like features. Maybe this red plastic is a high impact feature for your two years prediction or this particular item. So maybe you can use that as a negotiation tool or find a better supplier if, let’s say this red plastic supplier is unreliable, but unlike traditional forecasting method, they don’t tell you what’s influencing your prediction, right? Your format. Because at the end of the day, the only thing that is influencing your forecast is only your history. But the math prediction, we’re giving that access to the planners. So instead of let’s say them reacting to changes, sudden changes and manually scrambling, adjusting forecasts with demand prediction, they would I guess just be able to understand, okay, here’s influencing my predictions and not necessarily adjusting forecast because we haven’t put into account this spike in red plastic costs, let’s say.
Joe (11:10):
So it helps you adjust for that unexpected. Is it sort of building in a margin of error, so to say? You don’t, I guess that you kind of stay in your lane, right? You don’t go too far under stock, you don’t go too far overstock, but you’re able to kind of maintain because you have a good idea of what to expect, you’re able to maintain sort of a steady course.
Aurellia (11:35):
Yes yeah. And you’re able to save costs and not lose sales at the same time. Right? Because if you don’t have the item because you’re not accounting for this promotion, then you’re just going to lose sale.
Joe (11:51):
Right. And I that this, imagine that this affects not only just the purchasing function, but other business functions too, like the financing and marketing, not just the procurement.
Aurellia (12:02):
Yeah, like we’ve said before, accuracy is a downstream impact on the business. When you forecast accurately, you’ll have these, you buy at the right time in the right amount, and if the right amount is five instead of 10, you’ll have five times the cost extra money that finance can use for something else.
Joe (12:27):
So Aurellia, how can better demand prediction help align strategic planning with execution?
Aurellia (12:33):
Demand prediction would help your the planners to plan better because they’re using less of their time, let’s say firefighting or forecast, because on account promotions or account economic shifts with demand prediction, they can see the features affecting it, which in turn makes the planner trust the system more and have to less validate the predictions. And so they can use the time to do some other strategic things rather than firefighting, adjusting your forecast manually.
Joe (13:16):
So if I’m not constantly answering the phone from angry customers asking me where their orders are, and instead I’ve got an algorithm telling me what to expect so we don’t run out of stock, then that does free up a lot of my time. I imagine
(13:34)
As a planner to do all sorts of things, I know that every person that I’ve talked to, and certainly some of the interviews in this season of the podcast, we have heard that when planners have more time or when supply chain leaders have planners with more time, the first thing they do is how can we start becoming more efficient? How can we start improving the process? And I mean, I have found it, I don’t know if this is the case with you, is that once you start making improvements, they’re addictive. Once things start to get better, you want more and more things to get better.
Aurellia (14:11):
Yeah.
Joe (14:12):
Well, Aurellia, thank you so much for coming by. We’ve really learned a lot. I know if people want to learn more about demand planning and demand prediction, they can go of course to our website gainsystems.com and look up there. But what if they want to learn more about you?
Aurellia (14:26):
I’m on LinkedIn, so feel free to connect with me on LinkedIn.
Joe (14:29):
Alright, well thanks again, Aurellia for coming in. We really appreciate your time and look forward to hearing more from you next time on GAINS On.
Aurellia (14:36):
Appreciate it.
Joe (14:38):
And that’s a wrap on today’s episode of GAINS On a big thanks to Aurellia Sudihardjo for joining us and shedding light on the next evolution in demand prediction. With machine learning driven insights, external signal inputs, and deeper transparency into what’s driving your predictions, GAINS is taking out the guesswork and giving planners the power to make better decisions faster. So whether you’re tired of firefighting forecast misses, or just want to put your working capital to better use, demand prediction could be the key to moving from reactive to ready. If you enjoyed this episode, share it, subscribe, and stay tuned because we’ve got a lot more insights, tools, and transformation stories coming your way. Keep learning, keep planning and remember in this predictably unpredictable world, we’re all in this together. This is Joe Davis signing off from GAINS On. Until next time. Want to stay connected with all things GAINS and continue to explore the exhilarating world of supply chain planning and design?
(15:39)
Then don’t forget to follow GAINSon LinkedIn where you can be part of our growing and vibrant professional community. And for more content, engaging posts and updates, don’t forget to like and subscribe to GAINS On on YouTube. Trust us, you won’t want to miss what we’re sharing. If today’s podcast episode left you hungry for even more insights, we’ve got you covered. Every episode of GAINS On is accompanied by a detailed blog post for those who wish to dive deeper into the topic. Whether you’re looking to expand your knowledge or find that special morsel of information, our blogs are designed with you in mind. Visit gainsystems.com for more. All the links you need to be found in the description below. Thanks once again for tuning into GAINS On. And remember, we are here to help you to code the world of supply chains, one episode at a time.
