Bridging the Operational Gap in Business Central Forecasting

Emma: Welcome back to the deep dive. We’re shifting gears a bit today. You know, usually when we talk about AI in business, especially now, we get caught up in these huge sweeping ideas.

Ryan: The grand, the visions.

Emma: Exactly. Robot overlords, algorithms predicting the stock market. It’s all very sci fi. But I’ve been digging through the research for today and I think we need to talk about something much more well grounded and honestly, more frustrating.

Ryan: Frustrating is the right word. I mean, we’re swimming in data, we have these incredible prediction engines, Right. And yet businesses are still running out of stock or they’re drowning in inventory they don’t need.

Emma: It’s the operational gap. That’s the perfect term for it. And that’s our theme for today. Yeah, we have all this intelligence telling us what might happen, but we’re missing the mechanism to, you know, automatically do anything about it.

Ryan: It’s the differ between a GPS warning you about traffic ahead and a self driving car that actually reroutes you.

Emma: Oh, that’s a great analogy.

Ryan: Most companies today just have the gps. They see the traffic jam, the stock outcoming, but they’re still manually steering the car and they usually react too late.

Emma: That is the perfect setup for what we’re looking at today. We are diving into a really specific case study to understand this gap. We’re looking at the world of Microsoft Dynamics 365 Business Central, which is huge. And a specific tool from InsightWorks called the Enhanced Forecasting Worksheet.

Ryan: And let’s be really clear up front, our mission here isn’t to review a product.

Emma: No, this isn’t a sales pitch.

Ryan: Not at all. We’re looking at this tool because it’s a case study. It highlights a fundamental problem in how a lot of modern ERP systems handle the supply chain. It’s about bridging that gap between knowing and doing.

Emma: Okay, so let’s set the scene. If I’m a business and I’m running on Dynamics 365 Business Central, I’m feeling pretty modern, right? I’ve got the cloud, the Microsoft name, I’ve got AI.

Ryan: Yeah, absolutely do. Business Central has native integration with Azure AI and it’s actually pretty powerful. It’ll look at your past sales data, it’ll find trends, seasonality, and it spits out a forecast quantity. It’ll tell you, hey, based on everything, you’re probably going to sell 500 units of this widget next month.

Emma: Okay, so the system knows I need 500 units. Problem solved, right?

Ryan: You would think so, but this is exactly where that operational gap opens up. The forecast is just a number. It’s information. But the rules that tell the system when to actually buy more inventory, the planning parameters, they’re totally separate.

Emma: Wait, hold on. When you say planning parameters, are we talking about the absolute basics, like safety stock reorder points?

Ryan: Exactly those. Safety stock, quantity, reorder point, reorder quantity, maximum inventory. These are the actual levers that control your supply chain.

Emma: So play this out for me. The AI screams, demand is spiking. We need 500 units. But my safety stock is still set to.

Ryan: To whatever someone typed in six months ago, maybe a year ago. Let’s say you set it to 50 units when demand was low. The AI can see demand is rising, but unless a human planner goes into that item card and manually changes that 50 to, say, 100, the system is.

Emma: Still operating on the old rule.

Ryan: The system is flying on old data.

Emma: That seems, well, incredibly inefficient. It’s like having a smart thermostat that knows a heat wave is coming, but it won’t turn on the AC until you get up and turn the dial yourself.

Ryan: That is the daily reality for thousands of inventory planners. And think about the consequences. The AI predicts the spike. The planner’s busy. They’re putting out other fires. The inventory level drops. It hits the old reorder point.

Emma: So the system triggers an order.

Ryan: It does, but it’s a small order based on the old reality. And because demand is actually double what it used to be, you burn through that stock way before the new shipment arrives. Stock out, stock out. You get angry customers, you lose revenue. Yeah, and then what happens? The planner panics. They go in, they manually jack up the reorder point to compensate, but by then, maybe the spike’s over, and now you’ve got a warehouse full of stuff you don’t need.

Emma: It’s the bullwhip effect.

Ryan: Exactly. Magnified by stale, static data.

Emma: And the most frustrating part is that the intelligence, the forecast to prevent all of this was just sitting there the whole time. It just wasn’t connected to the ordering rules.

Ryan: Precisely. And traditionally, the only way you could fix this in Business Central was, well, what I call Excel. Hell, you export the forecast. You export all your items. You build a monster spreadsheet with VO ups everywhere to try and calculate what the new safety stock should be, and then you try to import it all back in.

Emma: Or you pay a Developer a fortune for a custom code.

Ryan: $50,000 easily.

Emma: Which brings us back to this enhanced forecast and worksheet. When I was reading about it, my first thought was, okay, it just connects the two. It moves number A into field B. But it’s. It’s more than just a copy paste job, isn’t it?

Ryan: Oh, absolutely. And that’s why we’re talking about it. If it was just copy paste, it would be boring. The enhanced part, the really clever bit, is the introduction of what they call expressions.

Emma: Expressions. Okay, now, it’s been a while since my last math class, so walk me through this. Why is an expression better than just plugging in a number?

Ryan: Because a number is static. An expression is. It’s logic. It’s a rule. It allows you to teach the system how to think. So instead of telling the system set safety stock to 100, you can tell it set safety stock to equal the forecast for the next month, multiplied by our supplier’s lead time, plus a 20% buffer for uncertainty.

Emma: Okay, that sounds incredibly powerful, but let me play devil’s advocate for a second. Isn’t that dangerous if I start automating all my ordering rules based on a formula, what happens if my formula is wrong? Am I just automating a disaster at scale?

Ryan: That’s a completely valid fear. Automation bias is a real thing. But you have to think about the alternative. The alternative is relying on a number you basically guessed at last year with expressions. You’re actually tying your decisions to the variables that matter right now.

Emma: Okay, give me a real world example. Let’s say I’m selling, I don’t know, big, expensive industrial machines versus cheap little plastic washers. I wouldn’t want the same logic for both, would I?

Ryan: Exactly. And you don’t have to. You can filter for those cheap plastic washers. You might write an expression that says something like, I don’t care about carrying cost. I never ever want to run out, so set my safety stock to three months of forecasted demand.

Emma: It’s aggressive because running out of a 10 cent washer could shut down the entire assembly line. And they cost nothing to store.

Ryan: Right, but for that heavy, expensive machine that ties up a ton of cash, you write a totally different expression. Maybe it’s set safety stock to only two weeks of forecast, but only if the forecast’s confidence level is high. You can bake your own risk tolerance right into the math.

Emma: Oh, I see. So you’re not just automating the number, you’re automating the strategy.

Ryan: That’s Exactly. And it’s dynamic. If your supplier for the washers suddenly emails you and says their lead time is now 30 days instead of 10.

Emma: And your formula includes lead time as.

Ryan: A variable, the system automatically recalculates the reorder point the next time you run the worksheet. Yeah, you don’t have to remember to go in and manually update it. The logic handles the shift in reality for you.

Emma: That sounds so much more robust than a spreadsheet. It really does change the system from just a place to store data into an actual decision engine. But I want to pivot to something else that really jumped out at me. The old saying, garbage in, garbage out.

Ryan: The golden rule of all data science.

Emma: Right. Because any AI is only as good as the history you feed it. If I had a bizarre sales month last year, maybe a competitor went out of business and I got a huge one time order, the AI is going to see that and think, oh, that’s a normal pattern. Let’s order a massive amount again this year.

Ryan: The unicorn order problem. It’s a classic.

Emma: Yeah.

Ryan: A single one off event that completely skews your entire prediction model. And if you feed that raw skew data into Azure AI, you get a skewed forecast. And if you feed that skewed forecast into your new fancy expression, you automate.

Emma: The very disaster I was worried about.

Ryan: Exactly. And this is why the data hygiene parts of this tool are so important. The worksheet actually lets you do what’s called historical data adjustment. It lets the planner be a detective before the AI jury even sees the evidence.

Emma: So I can literally go into the system and say, hey, see that massive spike in sales back in July 2025? Ignore it. That was a unicorn.

Ryan: Yes. Or you can smooth it out. You can say, that wasn’t a real 10,000 unit sale for forecasting purposes. Let’s create it as a thousand. Yeah, you’re cleaning the history. So the prediction is based on reality, not on weird anomalies.

Emma: It also mentions supporting multiple forecasting algorithms. Why is that important? I kind of thought AI was just AI.

Ryan: Not really. No. Different products have different demand patterns. Some are steady, like toilet paper, some are seasonal like snow shovels, and some are just totally erratic. Microsoft’s AI is great, but sometimes a much simpler algorithm, like a basic moving average is actually more accurate for certain items. This lets you pick the right tool for the right job, item by item.

Emma: Okay, so we’ve cleaned the data, we’ve picked the right algorithm. We’ve used these expressions to turn a smart forecast into a smart reorder.

Ryan: Point.

Emma: Now what? Let’s talk about the downstream effect. The source material kept mentioning MRP and.

Ryan: MPS Material Requirements, Planning and Master Production Schedule.

Emma: I feel like those are acronyms that people in business nod along to, but maybe don’t fully grasp. Why should someone listening care about MRP if they’re not a factory manager?

Ryan: Because MRP is the engine of the company. It’s what tells you what to buy and when to buy it. But here’s the catch. MRP is dumb.

Emma: Dumb.

Ryan: It’s just a calculator. It doesn’t predict anything. It just looks at your planning parameters, those reorder points and safety stocks we’ve been talking about, and it does the math. It just says you have 50. Your reorder point is 100, therefore you need to buy 50. That’s all it does.

Emma: So if your reorder point is wrong.

Ryan: The MRPA device is wrong. It’s that simple. And that’s why this operational gap costs companies millions. They have these huge expensive ERP systems, but if the parameters are stale, the MRP engine is just generating bad advice really, really fast. By using a tool like this to dynamically update those parameters, you’re finally feeding the MRP engine high quality fuel.

Emma: So the purchasing manager can come in on a Monday morning, open up their purchase advice screen and actually trust what it’s telling them.

Ryan: That is the holy grail, trust in the system. When you can trust the system’s advice, you stop second guessing every single line item. You stop keeping your own secret stash of safety stock in a different spreadsheet. You can finally let the system actually run the supply chain.

Emma: There’s a note here about system integrity that I think is important for any of the IT folks listening. We’ve all seen those third party apps that come in and try to rewrite the core code of an erp, and it just becomes a total nightmare when it’s time to upgrade.

Ryan: Right. Spaghetti code. It’s the worst.

Emma: Does this tool do that?

Ryan: No. And the source is very clear on this. It uses standard business central hierarchies. It is not replacing the MRP engine. It’s not trying to rewrite inventory logic. It’s just a very smart feeder system. It does its calculations and then places the results into the standard fields that Microsoft already built.

Emma: So it’s non invasive.

Ryan: Exactly. It respects the architecture, which makes it much safer and easier to deploy.

Emma: Speaking of deployment, this is usually where the conversation gets a bit depressing. We talk about this amazing technology, but then we find out it’s 50 grand a year and takes six months to install.

Ryan: The enterprise software tax, I call it.

Emma: Right, but the notes here say something really specific about the licensing for the enhanced forecasting worksheet. It says it’s free.

Ryan: It is, but we need to be very precise here. It is free for one named user per business central environment.

Emma: Okay, let’s just pause on that for a second. Is this a free trial? You know, use it for 30 days and then we lock all your data hostage.

Ryan: No, the source is explicit. The the single user license includes full functionality with no time limits. It is the complete full production tool.

Emma: That is really unusual. I mean, insightworks is a business, not a charity. Why on earth would they give away the full tool? What’s the catch?

Ryan: It’s actually a fascinating adoption strategy. Think about who usually buys this kind of software. The cto, maybe the cfo, but who actually feels the pain of bad forecasts? It’s the inventory planner, the operations manager.

Emma: The person who’s staying late every night trying to fix the spreadsheets.

Ryan: Exactly. But that person doesn’t usually have the authority to go out and sign a big software contract. By making that first user free, InsightWorks is basically executing a Trojan horse strategy. But in the best possible way, they are empowering the end user.

Emma: So if I’m a planner, I can just download this app from AppSource. I don’t need to ask for a budget. I install it. I can run it on my own data, fix my own forecasting problems, and start saving the company money.

Ryan: And then you can go to your boss with proof. You can say, look at this. We reduced stockouts by 20% last month just using this tool on my machine. Now I want to roll it out to the rest of the purchasing team.

Emma: And that’s when they buy the additional licenses.

Ryan: Exactly. It completely de risks the buying decision. You’re not buying a promise from a salesperson. You’re buying a proven result that you’ve already achieved. And honestly, for a lot of smaller businesses, that one free user might be all they ever need.

Emma: It basically democratizes the technology. You don’t have to be some massive corporation to get AI driven. Planning. You just need one curious person with a free license.

Ryan: It’s a very smart move. It changes the whole conversation from sales to problem solving.

Emma: Before we wrap this up, let’s just quickly touch on who’s behind this Insight works. Because if I’m going to let an app touch my inventory data, which is essentially my company’s cash sitting on a shelf, or I want to know who.

Ryan: Built IT they’re a Canadian independent software vendor, isv, with a big presence in Europe as well. But the really relevant part is their focus. They don’t do HR software, they don’t build CRM tools. They focus almost exclusively on manufacturing, distribution and warehousing.

Emma: So they live and breathe the supply chain.

Ryan: They do. They’ve got over 30 apps and they’re all in this very specific niche. A tool like this, it feels like it was born out of frustration. They probably watched hundreds of their clients struggle with this exact operational gap and finally just said, okay, let’s build the bridge to fix it.

Emma: It definitely adds a layer of credibility when the developer speaks the language of the shop floor, not just the language of Silicon Valley.

Ryan: I agree.

Emma: So let’s zoom out and synthesize this all. We started with a the operational gap. We have these incredible AI brains predicting the future, but our actual operations, our ordering parameters are stuck in the past being updated manually.

Ryan: And we looked at solution that uses logic, these expressions, to connect the forecast directly to the execution. It unties those hands.

Emma: We talked about the importance of data hygiene, of cleaning your history so you don’t just automate a past mistake. And the impact how getting this right trickles all the way down to the MRP engine and changes what actually shows up at your loading dock.

Ryan: And finally, we touched on accessibility. The fact that the barrier to entry is literally $0 for that first user means there’s really no excuse to stay stuck in the dark ages of spreadshee.

Emma: It really challenges the way we think about implementing AI. It doesn’t have to be this giant million dollar project. Sometimes it’s just about intelligently getting the data from column A to field B.

Ryan: The value isn’t just in the prediction, it’s in the integration. That’s the key takeaway. Knowing it’s going to rain is interesting. Yeah, having a system that automatically closes the windows is useful.

Emma: So here’s the thought I want to leave you with today. If you’re listening to this and you’re running a business on a modern ERP like business Central, go look at your item cards. Look at those safety stock numbers. When was the last time they were actually updated with any real logic?

Ryan: If the answer is I don’t know, or sometime last year, then you’ve got a problem.

Emma: If your system knows what’s coming, why are you still the one manually telling it what to do? The tools are there. Maybe it’s time to let the system start doing some of the driving. Thanks for deep diving with us.

Ryan: Goodbye, everyone.