Corporate Finance Explained | Corporate Forecasting: Why Predictions Go Wrong

[00:00:00 – 00:00:09]
So I’ve been staring at this stack of research you sent over and I have to be honest. Uh oh. It’s kind of terrifying. We usually think of the corporate world as this,

[00:00:10 – 00:01:53]
you know, highly sophisticated machine, right? You have teams of PhDs, you have AI, you have these massive ERP systems that cost more than my house. It creates a real sense of security, doesn’t it? Like all that data must mean something. It does. You assume that with all that horsepower, these companies must know exactly what’s going to happen next. Like they have a crystal ball. Right. But looking at these notes, it seems like the crystal ball is cracked. Or maybe it never existed in the first place. That is the big secret. If you walk into any boardroom, anywhere in the world, there’s an unspoken truth. The forecast is always wrong. Always. Almost always. And here’s the thing. Everyone knows it. The goal of a corporate forecast isn’t actually to be perfect. The goal is to be credible. But the problem is, while everyone expects the numbers to be slightly off, the way they’re off can destroy a company. Okay wait, let’s back up a second. I want to make sure we ground this for everyone listening. Because when I hear corporate forecasting, my eyes kind of glaze over a little bit. I get that. I picture a guy in a basement with six monitors and a lot of coffee just crunching spreadsheets. Why does this matter to someone who isn’t a CFO? Because the forecast is the nervous system of the entire organization. It is not just a math exercise. I mean think about it. If that spreadsheet says demand is going to double next year, what happens? I assume you start hiring. You hire aggressively. You sign leases for new office space. You tell your factories to buy raw materials. You tell Wall Street, hey, expect big things. And if the spreadsheet was wrong. Then you have hundreds of employees you can’t pay. You have warehouses full of inventory that’s just gathering dust.

[00:01:54 – 00:04:15]
And your stock price tanks because you promised something you couldn’t deliver. The stakes are incredibly high. It determines everything from your bonus to whether the company even survives. That’s the tension that really grabbed me in this research. We’re living in the golden age of data, right? Supposedly. We have more information than any generation in history. So why are we still so bad at this? Why do these massive smart companies still miss the mark so often? It’s the million-dollar question. And the answer the research points to is, well, it’s fascinating. It turns out the failure isn’t technical. It isn’t that the algorithms are broken or the computers aren’t fast enough. It’s us. It is absolutely a human problem. It comes down to behavior, incentives, and how we communicate. I really want to dig into that because you’d think data would be objective. A number is a number. But the sources describe this almost like a game of tug of war happening inside the company. Oh, it is war. And it usually starts with what we call sandbagging. I’ve heard this term thrown around. It’s basically lowering expectations, right? Right. Imagine you’re the VP of sales. You’re sitting down to tell the finance team what you’re going to sell next year. Now, your bonus, your livelihood, depends on you hitting that target. Are you going to give them a number that’s a stretch, a number that requires everything to go perfectly? No way. I want a number I can hit in my sleep. I want to look like a hero. Exactly. So you sandbag, you tell finance, oh, the market is really soft right now. Competitors are aggressive. Clients are delaying decisions. You intentionally suppress the forecast so you can step over the bar easily later. Okay. So that’s the sales team dragging the number down. But surely the CEO isn’t doing that, right? No. The leadership team usually has the opposite incentive. They fall into what’s called the optimism trap. If you’re a business unit leader, you want resources. Of course. You want to launch that new product line. You want to hire 10 more engineers. You can’t just walk into the CFO’s office and say, “Growth looks flat. Can I have $10 million?” You have to sell the dream. You have to show growth. Yeah. So you push these overly optimistic assumptions. If we just launch this one feature, sales will triple. So you have sales pulling down to protect their bonuses and leadership pulling up to get their budget? Who on earth? I mean, who figures out what the real number is? The finance team. Specifically, the FP&A group financial planning and analysis.

[00:04:17 – 00:04:23]
They are stuck right in the middle. Wow. They’re essentially the referee in a game where both teams are trying to cheat.

[00:04:25 – 00:04:59]
Their job is to filter out the sandbagging and the blind optimism to find something close to the objective truth. I do not envy them. But even if you take the politics out of it, let’s say everyone is actually trying to be honest. The research mentions that our brains are just wired wrong for this. We have these cognitive biases. We do. And they’re so dangerous because they’re invisible to us. The most common one in forecasting is probably anchoring. Right. That’s where you get stuck on an initial number, isn’t it? Yeah, exactly. Let’s say I ask you to forecast sales for next year. The first thing you’ll probably do is look at what we sold last year. Let’s say it was 1,000 units.

[00:05:00 – 00:06:14]
You’ll instinctively start at 1,000 and then nudge it up or down a little bit. That seems logical though, doesn’t it? History repeats itself. Sometimes. But often the market has fundamentally changed. Maybe you should be starting at zero and building up from scratch. But because you’re anchored to that 1,000 number, you can’t see the cliff edge that’s approaching. You’re just tweaking the old reality instead of seeing the new one. Exactly. And then there’s overconfidence, which I assume is, well, exactly what it sounds like. Right. It’s the belief that we have more control than we actually do. And it’s often compounded by pressure. If the CEO says, “We need to grow 20% to keep the investors happy,” the finance team might start tweaking the spreadsheet sales just slightly until the bottom line says 20%. You torture the data until it confesses. Precisely. You aren’t forecasting anymore. You’re just reverse-engineering a fantasy. Okay. So we’ve established that humans are messy, political, and biased. I want to see what happens when this hits the real world. The source material had this really stark comparison between two retail giants. Let’s talk about Target first. This is a textbook case of what happens when you misread the signals. And Target is a sophisticated company. I mean, they have endless data.

[00:06:15 – 00:16:08]
But after the pandemic, they hit a massive wall. What exactly happened there? Well, think back to 2020 or 2021. We were all stuck at home. We weren’t spending money on vacations or restaurants. So where did all that money go? Stuff. I think I bought a new patio set, a new desk, probably way too many kitchen gadgets. Right. Demand for discretionary goods, home decor, electronics, furniture. It went through the roof. It was a historic spike. Target saw those sales numbers, and they made a critical error. They extrapolated that trend into the future. They assumed the spike was the new baseline. They fell into the trap of looking at lagging indicators. They looked at what happened yesterday and just assumed it would happen again tomorrow. They didn’t account for demand normalization. Which is just a fancy way of saying we eventually ran out of room for new patio sets. Exactly. The world opened up. People wanted to travel and eat out. They stopped buying lamps. But Target had already placed the orders. Ships were literally crossing the ocean full of inventory that nobody wanted anymore. And this isn’t just a whoops moment, right? This cost real money. It’s devastating. When you have too much inventory, you run out of warehouse space. You have to pay to store it. And eventually, to get rid of it, you have to slash prices. Yeah. You see those 70% off stickers? That’s the sound of profit margins dying. So the lesson there is, don’t assume the future will look like the past. But then you have the counterexample. The research points to Walmart as the gold standard here. What did they do differently? It’s really a tale of two philosophies. Target was relying on a static view of the world. Walmart, on the other hand, relies on agility. They don’t just look at last month’s sales report. They use near-real-time data. Meaning they know what I bought five minutes ago. It’s pretty close. They integrate point of sale data, inventory levels, supplier lead times, even regional demand signals. If they see a dip in electronic sales in Ohio on a Tuesday, the system flags it immediately. But knowing it is one thing. Being able to do something about it is another. That is the key. They use something called rolling forecasts. Okay. Break that down for us. So, traditionally, companies do an annual budget. In October or November, you plan out the entire next year. You lock it in, and that’s your Bible. Right. But Walmart says, “That’s ridiculous. The world changes way too fast.” So they update their forecast constantly every month, sometimes every week. Doesn’t that drive people crazy? I mean, if the plan is changing every week, how do you focus? It requires a culture shift. You have to stop thinking the forecast as a target to hit and start thinking of it as a navigation tool. Like a GPS. Exactly. If you’re driving and there’s a traffic jam ahead, you don’t keep driving into it just because you planned that route yesterday. You reroute. Walmart’s system allows them to reroute their supply chain in real time. So the big takeaway there is forecast agility beats forecast precision. Yes. Yes. You will never predict the future perfectly. So stop trying. Instead, build a system that can react fast enough when you’re wrong. That makes total sense. But I want to get practical. The research included this toolkit for fixing the forecasting process. And honestly, some of it sounded a bit like jargon. I want to unpack the variance analysis piece. Okay. The text says you need to look at price, volume, and mix. I get price and volume. Did we sell more or did we charge more? But mix seems to be the one that trips people up. What does that actually mean? Mix is the silent killer of forecasts. Let’s do a simple math example. Imagine you run a software company. You have two products. Product A is your software license. It’s almost pure profit. Product B is consulting services. You have to pay humans to do the work, so it’s low profit. Okay. High-margin software, low-margin people. Got it. Now, let’s say you forecast a million dollars in revenue. And at the end of the quarter, you hit exactly 1 million. The forecast was perfect. On the top line? Sure. But what if you forecasted selling mostly software, but you actually sold mostly consulting? Your revenue is the same, but your profit is vastly lower because your costs are higher. Ah, so you hit the revenue number, but you missed the profit number. That’s a mix variance. If you don’t analyze that, you might high five everyone for hitting the revenue goal, not realizing you’re actually bleeding money. You need to understand what you sold, not just how much. That is a huge distinction. The other part of the toolkit that made me laugh, mostly because I’ve lived it, was the section on model governance. Ah, yes. The Excel hell. It warned against the dreaded final version V3, the real final dot Excel. It sounds funny, but it is a massive risk. I have seen billion-dollar decisions made off a spreadsheet where someone hard-coded a number in cell D1 and forgot to change it back. That is terrifying. It happens all the time. Governance is just a fancy word for discipline. It means locking down the model so people can’t break it. It means version control and really important. It means clearly separating your scenarios. This was the base case versus stretch case idea. Right. You need a base case. This is what we actually think will happen. This is what we tell Wall Street. Then you have a stretch case. If everything goes perfectly, this is what we could do. And the problem is when people mix them up. Yes. Leadership falls in love with the stretch case and decides to make that the budget. Now you have a plan that requires a miracle to succeed. You have to keep them separate. The sources also mentioned Salesforce as a company doing something really smart called driver-based forecasting. How is that different from just asking a sales guy, hey, what are you going to sell? So, asking a salesperson that question is asking for an opinion. As we discussed, opinions are biased. Driver-based forecasting ignores the opinion and just looks at the math. Give me an example. Okay. So instead of asking for the number, you look at the activity. How many phone calls did the team make? How many new leads entered the pipeline? What is our historical win rate on leads like this? So if I know I usually close one out of 10 deals and I have 50 deals in the pipe, then the math says you’ll close five. It doesn’t matter if the sales rep feels lucky and thinks they’ll close 20. Yeah. The drivers, the activity data say five Salesforce applies probability weighting to every single deal. If a deal is in the early handshake stage, maybe they only count 10% of its value. If it’s in the contract stage, maybe 90%. It takes the emotion completely out of it completely. It turns forecasting into a probability equation rather than a wishlist. We’ve covered the psychology, the case studies, and the technical toolkit. But there’s one last piece that the research really emphasized. And it might be the most important one. I think I know where you’re going. You can have the best model in the world, but if you can’t explain it to the CEO, it’s useless. Communication. This is where most finance careers live or die. The phrase that stood out to me was false precision. Executives apparently hate it. Oh, they despise it. If you walk into a board meeting and say, revenue next year will be exactly $104,230,000, you lose credibility instantly. Because it sounds like you’re just making it up. Because everyone in that room knows the world is volatile. Pretending you have 100% precision implies you don’t understand the risks. Executives don’t want a single perfect number. They want a range. And they want the story. Okay, talk to me about the story because storytelling is such a corporate buzzword. What does a good forecasting story actually sound like? A bad story sounds like this. We missed the forecast by 5%. That’s just reporting the score. A good story explains the mechanism. We missed the forecast by 5% because supply chain delays in Vietnam pushed $2 million of revenue from Q3 into Q4. However, our order volume is actually up. So this is a timing issue, not a demand issue. That is a totally different message. Completely. One says we failed. The other says, here’s the context. And here’s why the business is still healthy. That’s what leaders need to know. Right. They need to know if the problem is structural like with Target, or if it’s just a blip. It’s the difference between being a score keeper and being a navigator. That is the perfect analogy. The finance team needs to be the navigator. There’s a storm ahead. Here are three route options and here are the risks of each. That empowers the CEO to make a decision. Just saying it’s raining doesn’t help anyone. So pulling this all together, we started with the idea that forecasting is broken. But it seems like the fix isn’t just buy better software. No, not at all. It’s about recognizing our human biases, the sandbagging, the anchoring. It’s about building a system that values agility over precision, like Walmart. And it’s about having the discipline to analyze the variants, the mix, the drivers, so you can tell a credible story to the business. I want to leave the listener with one final thought that I saw in the notes. It was a bit of a provocation. It suggested that maybe the annual budget itself is obsolete. It is a growing sentiment. Yeah. In a world where supply chains break overnight, pandemics happen, and AI disrupts entire industries in months. Does it really make sense to spend three months planning exactly what you’ll spend on paperclips next November? It feels like an exercise in futility. It might be. The future might not be about setting a static target for the year. It might be about building a continuous rolling system that can survive, not knowing what happens next week. Building a business that doesn’t need a crystal ball because it has good reflexes. Exactly. Well, this has certainly changed how I look at those quarterly reports. It’s not just numbers. It’s a story of human psychology and survival. Always is. Thank you so much for breaking this down with me. My pleasure. And to our listener, next time you see a projection, don’t just look at the bottom line. Ask about the assumptions. Ask about the drivers. And remember, the number is almost certainly wrong, but the story behind it might just be true.

[00:16:09 – 00:16:12]
Thanks for listening to The Deep Dive. We’ll see you next time.