and modeling techniques Econometric modeling
What is it and why do you need it
Econometric models are a way to predict the sales volume of a product with an accuracy of up to 95% and increase the profitability of a business.

We create sales models that take into account the weight and influence of each factor (advertising, price, distribution, competitors, and so on) on product sales.

The model makes it possible to check the scenarios of the intended marketing activities, advertising and price changes before they are implemented, which helps to make the right decision and brings the cost of the test error to zero.

At the same time, the resulting formula is able to show the levels of investment in a particular type of advertising in order to get a response that maximizes ROI; to determine the price of the product that maximizes off-take in the current market situation.

In complex and complex models, it is possible to determine the cross-dependence of one product on the promotion of another within the brand portfolio.

The model breaks down the final level of sales into influencing factors, and studies the dynamics of the response to certain changes.
What mathematical methods do we use?
• time series
• different types of multiple regression analysis
• MARSplines curves
• regression trees
• cluster analysis
• factor analysis
• logline analysis
• neural networks
• other methods necessary for a specific task (specialized software and mathematical algorithms are used in the process of work)
Modeling technique
1
Factors
• we define a set of factors that affect the final result;
• we leave the significant ones;
• we remove or neglect the insignificant ones.
2
Model
• we build a model of the dependence of the result on the parameters:
• we create a formula or a system of formulas that most correctly describes the existing behavior of the brand;
• we use various optimization algorithms to determine the coefficients of the model.
3
Verification
• the model is verified on the historical sales data provided by you, with a depth of at least 1 year, and preferably 3-5 years
We have moved away from the linear regression model, which works in a very limited range and gives a wrong interpretation of the results, in particular, the influence of price and distribution factors
In our model, we study:

• how a consumer buys a brand
• what factors influence the purchase
• how factors affect each other As a result of applying this technique, our models have an average accuracy of 95% and reliably work in a wide range of source data
What do you get from the sales forecast model?
ROI increases to +30%
• You can see the response charts for advertising and marketing events
• You determine the levels of investment needed to achieve your goals
The risks of marketing activities are leveled
• You don't try blind promotion approaches
• You can test the promotion scenarios yourself
• The result of the campaign is known even before the launch
The model in your preferred form
• A formula or a group of formulas
• An MS Excel workbook with an interface, graphs, and a data set that you can immediately start working with and forecasting.
• A *.dll or *.ocx library that connects to existing software
• A software product with a user-friendly interface that can make forecasts, build graphs, and make recommendations 