Forecasting to be proactive, rather than reactive
According to Greg Fazzaro, Director of Finance for Novartis, the challenge in todays environment, where there always seems to be some kind of bad news coming out, is that we need to have up-to-date insight on the pharma marketplace and what that actually means for our sales forecast.
Fazzaro told attendees at a recent eyeforpharma forecasting event that we also need to have a clear view of the risks and opportunities above and beyond that forecast.
Its one thing to know what youre predicting sales will come in at, but what are your risks and opportunities so that you can make an informed decision? he says. We also need to understand things like inventory levels, stock outs and write offs. These are things that of course, in the end, impact sales.
From a supply chain standpoint, Fazzaro says, we also must manage our production plans, lead times and capacity constraints.
At Novartis, were a very large pharma company as well as generics maker and our products have very different lead times, he says. Some are very short and others are very long and you need to take that into account when doing forecasts.
Here a forecast, there a forecast
Historically, Novartis has had two different forecasts, Fazzaro says. The demand forecast is generated using the Futurecast system to produce a 72-month forecast by skew. And while he says it sounds like a lot, Fazzaro stresses its actually needed from a production standpoint.
It takes three to five years to build a plant, so we have to have some insight into the future and its much easier to forecast at a consistent granularity, he says.
Fazzaro says the forecasts are also built with different assumptions, events and trends and although you may have a demand manager and finance manager sitting no more than 10 meters apart, often they dont know that the other exists. So you would have two different forecasts and then which do you use to make a decision?
Its not that we want to go to one number, but clearly you want some commonality in how those forecasts are built, he says. There was very little, to no cross-functional effort between the different customer groups of marketing, finance, supply chain. Everybody was coming up with their own numbers. There were different timelines, different granularities. The demand plan was at SKU, the finance plan was at brand.
With two or more forecasts, Fazzaro says, the trouble is knowing where the truth lies. Although some argue the benefit of having a couple of different views, he says, there are some key issues to consider.
As he explains, the demand plan is used for production plans to determine capacity increases, to understand where investments should be made as well as to facilitate long-term supply agreements. And from the financial forecast perspective, its used for resource allocation things like where scarce marketing dollars should be invested and how much should be spent on research and development. And pharma companies also need to provide guidance to the investment community, he says. If any of these are misaligned, wrong decisions are made in terms of investments or funding, he says.
You can have excess inventory, which equals increased write-offs, which is just a hit right off the bottom line, Fazzaro says. And thats actually how we were able to get the marketing folk to see why they need to pay attention to this. If you have to keep profitability at a certain level and youre basically throwing money away with write offs, the easiest way to get profitability back is to start to cut marketing spend and that hits home for the marketing people.
And he says if you have stock-outs, not only is it lost sales, but it starts to have negative impacts on patients, especially if its a lifesaving drug.
The typical sales forecast, Fazzaro says, was two distinct processes with no alignment the demand and financial forecasts are made in isolation using different assumptions, events and trends.
Everybody tends to know when a generic is going to happen, or when a launch is going to happen, he says. But based upon the different needs of the plans, there are sometimes different dates.
Within Novartis, Fazzaro says, the financial forecast was always perceived as conservative, due in part to the incentive process, while the demand forecast was always seen as optimistic, unconstrained.
So you have these large, unclear differences, he says. At the country level, the local forecast was done in the two areas finance and demand. Demand was submitted all the way up to global with no interference. But the financial forecast would go through a couple layers of review, so by the time you went from the country all the way up to global it started to skew either higher or lower, depending upon needs.
An integrated sales forecasting process
But in 2008, the group rolled out an integrated sales forecasting process a single process with better alignment, Fazzaro says.
What we want to do with this process is have mid-point forecasts based upon the best knowledge that we have today, using common assumptions, events and trends through a cross-functional effort involving all the parties that have the relevant information colleagues from supply chain, finance, marketing, trade, QA, DRA, etc., he explains. We also want to try for monthly alignment of the financial and demand forecast at least at the country level, with smaller and clearer differences.
One key point, Fazzaro says, is that they wanted to make sure there were no safety stock buffers in the forecast.
As we talked to many of the countries, we found out that as they were building the forecast, they would say well we really think its going to be 100 but just in case were going to put in 110 so that we have extra, he says. Well that safety stock is actually separately managed through the MRP2 process, so youre actually double counting safety stock. We really had to educate people to put in what they really thought because theres a separate process when its going to the production chain to handle safety stock and inventory levels.
In the end, Fazzaro says, Novartis wants to get to one process producing two forecasts.
A basic tenet of the companys approach for integrating sales forecasting, he says, includes making a monthly demand forecast the basis for the short to long-term financial sales forecast. Although every month they generated a 72-month sales forecast by SKU, Fazzaro says it wasnt leveraged as much as it needed to be.
We showed the finance people and the market people that we already had a forecast going out into the future and encouraged them to use it, he says. The financial forecast can then be derived from that forecast using management adjustments. Theres always going to be the need to tweak it a little bit, but in the end youre reducing parallel work because its a single, cross-functional effort. Youre aligning everything up front as much as possible, rather than waiting to the end and seeing this large divergence and trying to figure out how to get it back to somewhere in the middle.
Novartis also has established a defined governance policy which defines clear roles and responsibilities for everyone in the sales forecasting process, Fazzaro says. In Novartis Pharma, finance plays the lead role, he explains.
The CRO in the country actually owns the sales forecasting process, he says. That may be somewhat unique, but our thinking behind is that finance, and specifically the CFO, sit in a unique position to bridge between marketing and demand. And the CFO also has a certain level of power in the sense that if there are differences between the forecasts, he or she can negotiate as to where it should go. And when you have a sales forecasting meeting, if the CFO is leading that process, you can be sure that the brand manager will attend.
Novartis also has an escalation route, Fazzaro says, for forecast misalignments. There are going to be times when you cant agree, he says, and in the end the CFO has to be aware of that and sign off on it because it will impact profitability, the sales forecast and resource decisions.
We also want to facilitate operational efficiency, he says. By doing this process and putting everybody together, we can optimize inventory levels, reduce stock outs and minimize write-offs and waste.
Its a monthly process, Fazzaro stresses, driven and owned by the CFO, with the key inputs being actual sales and prior forecast submissions and trends, as well as updated assumptions, events, market information and pricing.
Forecasting in a changing world
You need to say, OK here is what happened. What does it mean going forward as well as whats changed in the market since our last forecast and in todays world? he explains. Every day the world is changing, so we need to try to take that into account and at least take a snapshot and based upon our best information and knowledge, say heres whats going to happen.
Fazzaro stresses that companies need to understand their inventory, stock-out and write-off positions.
You dont want to keep producing if you have a lot of inventory, or you need to ramp up if youre running out of stock because your product is taking off faster than expected, he says.
Globally, Fazzaro says Novartis brand managers own the launch forecast, because the countries dont always know all the details, but locally the marketing side needs to own the SKU forecast what is actually sold in the country and they need to own the local forecast assumptions.
And the supply chain organization needs to globally manage capacity to ensure supply that we have enough plants, that the production schedules are set, that we can produce what we need in a timely and cost efficient manner, he says. They own inventories and manage customer service levels so that if you place an order, its delivered and you get what you expected. The supply chain organization needs to own the local parameters and local customer service making sure that products are delivered to wholesalers, for instance, and they need to manage supply and demand, while also analyzing and challenging forecasts.
Finance, Fazzaro explains, sits in the middle. At the global level, they provide the process and training support and overall forecast validation, collection and analysis. They communicate the aligned assumptions between the different parties, he says, and also own the consensus forecast.
The financial forecast is built off of the demand plan and of course at the financial level you need to assess and optimize the inventory levels, he says.
Measuring the results
Fazzaro says that its one thing to roll something out, but if you dont measure the success in implementation it easily dies off.
So Novartis put in place two sets of KPIs process and business. From a process perspective, theyre measuring process compliance. Is it truly a cross functional effort? Is the CFO driving it? Are financial forecasts based off of the demand forecast?
We needed to be sure to put the pressure on people that this is a process thats been approved by senior management and its what we need to implement, Fazzaro says. As we rolled this out in the countries, we saw many times we had a nice forecast for 18-24 months and then it would simply drop to zero. We would explain that they didnt have to have the same level of accuracy as in the first 18 months, but at least we need to know what the future looks like through their best estimate. If you have a forecast thats going to deviate significantly from the trend, you want to explain that by an event is there a generic entrant, is there a competitor, etc.? You need to understand that and it needs to be transparent and documented.
Fazzaro said as they rolled out last year, they saw a variety of feelings about the process.
Some said it was a must that they couldnt live with large differences between the forecasts, it drives the necessary behaviors for integrating across the different functions, and brings in place a discipline to how we plan, he says. On the negative side, we heard there are good reasons why you should have two different forecasts, that we need a clear policy on how to handle things like safety margins and some felt it wasnt possible given the timelines. So we really needed to show people the benefits of the process and bring naysayers along by showing what you can get out of it.
Fazzaro says the benefits Novartis has seen include lower stock outs and write offs across the organization, much higher transparency in forecasts, and an uptick in forecast accuracy.
This all yields increased visibility into the direction of the company, he says. We can make better decisions about where we should invest money and so we have more funds for reinvestment because were not wasting it on write offs or stock outs. Weve increased our overall customer service levels internally and externally, and upfront alignment and communication results in reduced workloads because theres less time spent bridging between forecasts and sending out ad hoc requests.
Fazzaro says the approach allows everyone to better understand the business and the challenges and to start to formulate things they can do to be proactive rather than reactive.