Sunday, 16 September 2012

Predictive Modeling and Planning & Budgeting

Embarking on something big or out of the turn? Its time to spice your Planning and Budgeting exercise with with Predictive Modeling Tools. Predictive Modeling in Planning and Budgeting is useful for generating realistic scenarios using data like past performance, business drivers etc. Admitedly, this is more than simulation of scenarios by varying underlying numbers that have gone in as Plan and Budget i.e., sales volume and price in certain territory or plain statistical extrapolation of your past data.

Lets look at it in the following context

• You have medium term strategic direction in place.
• Your annual planning and budgeting is in line with your strategic direction and objectives
• Your strategic direction and annual plan have undergone rigorous analysis like Porter's - Five Competetive Forces that shape Strategy

You have to take some calls on your annual and mid-term plans. Here's where Predictive Modelling Tools come in.

• You have sufficient past data to base your analysis
• And you have sufficient business drivers in place to base your calculation

Here I would like to move away from analytics at an operational level i.e., analytics crunching immediate to short term influencers to predict stock out or demand planning etc. Walmart uses services like Weather Trends International to correlate demand with weather condition. They are demand humunguous correlated data and are more complex to execute.

For example, you have grown 25% QoQ for the last 4 quarters and economy looks good. To reach your strategic goals, you have to add new capacity to meet demand, increase the marketing spend, invest in new products, beef up supply chain and expand overseas. Some of the questions in front of you are - would supply chain be able to meet the demand, are any disruptive technologies emerging, is competition catching up with better features, is demand elastic to pricing, is sufficient cash flow available? And you have to take calls on reducing your risks and spreading the resources most profitably.

Your past data can help answer some of the questions. Where there is no past data, build it. Since you have more than few variables to consider, that too with assumptions, its time to use Modelling. Apply on the data your important assumptions and the ones most likely to happen. Take a call taken on business drivers you can control and the ones lying outside the span of control of management. The important aspect lies in identifying the key business drivers & assumptions and making them as realistic as possible. This will throw up scenarios which will help in decision making. And what you decide goes into your Plan and Budget.

Lets take an example where business drivers were within the span of management control. Automotive industry suffers excess capacity, demand robust cash flow for operations, investment in new models, battle perception in crowded market place and finally, have a strong line up. With finite resources available, call must be taken to manage all of these at the same time. 6 years ago, Ford NA took few strategic decisions which proved to be the best. And these were not simple standalone decisions.

- They beefed up cash position by selling off non core brands like Jaguar, LandRover.
- Invested in new models and strengthened by focusing on Ford and Lincoln (they let go Mercury)
- Lowered cost by moving to towards common platform and investing in global products.
- Finally, they focused on QA to climb back on JD Power rating.

Now each one of these i.e., cash flow, manufacturing cost, quality and line of vehicles are key business drivers towards profitability. May be this is a over simplified example, but Predictive Analytics helps test your plan against key business drivers, how finite resources can be allocated and how to still reach your strategic goal.

 'Subbu' Balakrishnan | EPM and BI CoE

'Subbu' is a qualified Cost & Works Accountant, Company Secretary and Law Graduate, with more than 20 years experience in operations and IT.

No comments:

Post a Comment