Demand planning is just educated guessing — here's how to make better guesses

One of our demand planning consultants said something recently that stuck with me.

"Even with a plan, we're still making a guess. But it's an educated one."

If you've been avoiding demand planning because it sounds like something a $500M company does in a conference room with enterprise software and a team of analysts — you've been avoiding the wrong thing. Demand planning is just the work of making better guesses about how much product you'll sell, and when.

You're already guessing. The question is whether your guesses are educated.

What happens when you guess without data

A few years ago, a demand planning team we know was running S&OP for a major personal care brand — $150 million in annual revenue, market leader in its category. And yet every single month, they missed their forecast by 10 to 20 percent. That's roughly $3 million off, month after month.

Nobody could explain it. The brand was strong. The promotions were running. Everyone in the room kept saying the same things: back-to-school season would drive orders, a big retail event would move product, a major retailer was running a feature. The forecast was built on those assumptions, not on what the data was actually showing.

The data was telling a different story. Consumers were complaining online about a recent reformulation. Retailers had stocked up heavily the year before, running buy-one-get-one promotions that led consumers to buy six months of supply at once. In year two, those same consumers didn't need to repurchase. Retailers weren't reordering because their shelves were still full.

The team wasn't making bad guesses because they were bad at their jobs. They were making bad guesses because they weren't looking at the right information. They were guessing based on what they hoped would happen, not what the data suggested was happening.

What an educated guess actually looks like

We worked on a new product launch for a personal care brand — one of the first in its category at that price point in the mass market. The forecast came from marketing. Our demand planner looked at it, considered the category dynamics, thought about what similar launches had done, and decided the number was too low.

She didn't override marketing or argue in a meeting. She quietly added a 33 percent buffer above their number before committing to production.

The product came in 40 to 50 percent above forecast. Without that buffer, the brand would have been out of stock at their major retail partners within two weeks of launch.

The buffer wasn't a lucky guess. It was an educated one — based on category knowledge, launch experience, and a clear-eyed read of the situation. That's demand planning. It's not a formula that produces a perfect answer. It's a disciplined process that makes your guesses better.

What the process actually involves

When we build a demand plan for a new client, we start with two years of historical sales data by SKU. We take that data and run it through several forecasting techniques — things like weighted moving averages, linear regression, and seasonality index — to see which method best matches the actual sales history for each product. Then we use MAPE (Mean Absolute Percentage Error, which just means: how wrong was each method, on average?) to pick the best one for each SKU.

That sounds technical. The underlying idea isn't. We're asking: given everything we know about how this product has sold in the past, what's our best estimate of how it will sell going forward?

Then we layer in the other information that affects whether you have enough stock: how much inventory you have on hand right now, what orders are already on their way to your warehouse, how long it takes to get product from your supplier to your shelf, and what the minimum you can order at once is.

Put all of that together and you get a 12-month view of when you'll need to place orders and how many units to order. Not certainty. An educated guess.

Where founders usually start

You don't need six forecasting models to start. You need three things.

1. Plan in units, not dollars. How many cases, how many each — not how many dollars. This matters more than it sounds. A price increase inflates your dollar forecast without changing how much product you actually need to make. A promotional discount does the opposite. Units tell you what's really happening.
2. Know your real lead time. Not just how long your co-manufacturer takes to produce. The full chain: production, transit to your warehouse, transit to the retailer's distribution center, time from the DC to the shelf. Most founders think their lead time is three weeks. It's usually seven or eight. That gap is where stockouts are born.
3. Look at what actually happened last month before you forecast the next one. The most common forecasting mistake isn't a bad model — it's not updating the model when reality doesn't match the plan.

The point

Nobody has a perfect demand forecast. You're always guessing about the future. The brands that get into trouble aren't the ones with imperfect forecasts — they're the ones who mistake their assumptions for data, who keep running the same numbers and expecting different results, who don't notice what's actually happening until it's expensive.

Better guesses start with looking at the right things. That's what demand planning is.

If you want to see what that looks like for your specific products and channels, we're happy to take a look. The first conversation is free.