So, Chakri, let's delve a little bit closer into the details of Demand Planning. What do you think are the key, the must-haves steps in a world-class Demand Planning process? Yeah, when we talk about Demand Planning process, first, let's recognize that there are different cycles within the overall Demand Planning process. There is the operational Demand Planning cycle can run weekly, a monthly Demand Planning cycle for tactical horizon, strategic Demand Planning process for the strategic horizon kind of run once a quarter or once a year.

Irrespective of the cycles, there are some standard steps that happen in each cycle, the degree of automation, the degree of review, the degree of manual review, the degree of risk analysis would be different in the operational, tactical and strategic horizons. But let me describe, kind of drill deeper into the tactical Demand Planning horizon. The key steps in that the steps are similar in others with, with you know, the differences that I talked about. So first, why is it important to get the process steps right?

The critical thing to get each of these steps, the roles and responsibilities kind of nailed out so that there's a consistency in which the Demand Plan is generated on time for the supply chain to respond. So what are those steps? The first step is I would call data collection and automation. For the inputs into any Demand Planning process on a monthly cycle or a weekly cycle.

You need to collect a lot of inputs.

The inputs are historical sales: What happened last month in terms of history?

Inputs are: drivers,

external drivers, market drivers.

And inputs are: commercial actions, commercial drivers. So for each of these we get the historical shipments. But with respect to the drivers, there was an assumption of what the driver was going to be versus what actually transpired. So you need to get the actuals for the drivers and potentially, if it comes to commercial drivers going forward.

Have any of those drivers been updated by the commercial team for what they think is going to happen on a go-forward basis, for example, changes to pricing, etc. The challenge today in the collection of data for demand planners is a lot of this data is in spreadsheets and disconnected and coming from different systems. One of the key benefits in a world-class Demand Planning processing system is the automation of the collection of all of this data into a combined data model that then makes it completely easy for the Demand Planner to access and use. So once we have the data collected, the next step and a very, very critical step that's often missed is what we call post-game analysis.

A lot of people jump to generating a demand forecast every week or every month, but not enough time and rigour is spent on what really happened last week or last month. What was the forecast? What was the actuals that transpired and why were there deviations between forecast and plan? There's a lot of gold in that when we when we dig deeper.

So if we can analyze the forecast versus plan deviations and you correlate with all the drivers that you had expected that were driving that forecast, well then the drivers transpired as expected. The market drivers, whether it is the commercial drivers or even supply chain drivers, what actually transpired, what explains the difference between the plan and the forecast and the actuals? That is a very important step because if coming out of that step are two key actions, one is should some of the drivers be changed going forward? If the drivers were not coming out as expected, should they be changed going forward because that can change the forecast for the future Or second step could be the forecasting models themselves need to be updated because the accuracy is not good enough.

So that's what we call post-game analysis. It's a very, very critical step in a good demand planning process. So once the post-game analysis is done, now you have updated information about all the drivers, what the driver values are going forward. Then you have a third step, which is the system-generated forecast.

This is a statistical forecast, a machine learning-based forecast. Whatever model you're using takes the drivers and the historical sales as an input, and the system is generating a forecast for the future periods. So there's a lot of details and you know, what kind of models to use at what levels you forecast, etc. . But let's assume that there's a process step, there's a critical step with the system-generated forecast.

Coming out of the system-generated forecast is a prediction now for the future. This is when a planning organization or Demand Planning organization starts the step of reviewing the forecast. What's the critical step in reviewing the forecast ? Well, the forecast has changed from last cycle.

One of the key things there is to really narrow down on the exceptions, what forecasts have significantly changed beyond thresholds from last time, what has gone up? What has gone down? And because any big changes there will drive a lot of impact in the supply chain, a lot of bullwhip. And there has to be good reasoning applied to any big changes.

In fact, explanations have to be given for those big changes. Otherwise, supply chain is going to second guess and not accept the forecast changes. So the purpose of the demand review from the demand plan is to focus in on the big exceptions. What changed from last cycle and start reasoning as to why those changes have come in, what drivers have changed?

What could be the reasoning? So once they've established that basic reasoning, then the next step is kind of bringing in all the stakeholders into a collaborative Demand Planning Process, where what we call the common level off of demand forecast, where it's a common level of interest between finance, between account teams on the supply chain and that common level, you want to drive a collaboration and a consensus around the forecast numbers. So the common level, as we discussed in other topics, the common level could be different than the level of detail with which the supply chain needs a forecast or the sales team needs a forecast.

But it is that one number, a common level where everyone agrees to what the market demand is. So at that level, you want to really focus in on what the changes are, putting the reasoning behind those assumptions. Do we agree on this new forecast? Do we agree on the changes to the forecast where the forecasts have gone up or down?

So that's putting the reasoning behind this forecast and that's a very critical step, documenting the assumptions and the reasons behind why the forecast has changed. There's another critical step that happens in the collaboration and consensus process, which is what we call risk and opportunity assessment. So the forecast has been generated by the system, but there could be potential downside risks to the forecast or potential upside risks, opportunities to the forecast that different stakeholders have inputs to your sales teams, your account teams, product marketing. They're seeing various opportunities and risks in the forecast.

These are used to construct what I call demand scenarios. So the forecast is not just one number, it's the possibilities of if some things different happen in the market, the forecast could be not 100, but it could be 120. More opportunity is coming. If something adverse happens because of a risk, it could be 80.

So the question is what is the range of forecast possibilities? What are forecast scenarios? So in this collaboration and consensus process, you're establishing the reason for the baseline forecast, what we expect, what the reasoning for the changes, but also the risks and opportunities, assessing those and establishing the forecast scenarios that we want supply chain to consider. So at this point, once we have established those finalized forecast scenarios, I would call them as pessimistic forecast, an optimistic forecast and a most likely forecast.

Let's assume those are three demand scenarios. Then those forecast scenarios need to be published to the supply chain for a response. But in terms of publishing, there's a very important step. Remember, we talked about the collaboration and consensus being at a common level.

Supply chain requires a forecast potentially at a more detailed level than what the consensus forecast is. So there's an automated disaggregation step of taking the consensus forecast and driving the supply chain forecast. There are rules and algorithms that can be applied to take the consensus forecast and automatically disaggregate it to what the supply chain needs to drive supply chain decisions. Similarly, the consensus forecast can be translated into what account teams need to manage the account sales forecast.

So this process of disaggregation translation is typically automated, and then the forecast is then automatically published to the supply chain planning process where they're going to do the demand supply match. To close the loop when the demand supply mathc process comes back with responses to these demand scenarios. That's when we are looking at an integrated business planning process, assessing the costs of supporting the demand scenarios, the ability of the supply chain to support the demand in the first place, and then the integrated business planning processes saying, okay, we're going to go with based on an analysis of risk and the supply chain's ability to support, we're going to go with one of the demand scenarios of the possibilities or demand scenarios that exist.

So that demand scenario, which is now blessed by, hey the supply chain can support it, is the amount of risk we can take in terms of cost and inventory is what we would call the demand plan that against the forecast that was initially generated. So. basically a forecast or a range of forecast scenarios has converted into a demand plan. And that's what we now as an organization cross-functionally have aligned to on what we expect to do to sell.

So that's, in a nutshell, the overall Demand Planning Process that's done in a tactical demand planning process. Now the same steps in a sense apply for operational Demand Planning, you know, collecting the data, generating forecast reviews, except that because of the shorter cycles of an operational Demand Planning, shorter timeframes, a lot of the steps, a lot more automation would be applied there and not as much rigorous review of not as much scenario planning, but a lot more automation in that process.