Advanced pricing software accounts for historic sales data as well as current inventory levels and operating costs, alongside detailed classifications of customer types by demographics and location. Further, companies also need to account for competitors’ prices of complementary goods, as well as the broader state of the market.
With big data, many observations means that every effect size differs from zero (i.e. is statistically significant), and the issue is deciding which effects actually matter. Hence our new problem is ‘identification’— ensuring there is a clear causal path from one variable to the other, that doesn’t mix up other influences. AI won’t do this automatically, and without guidance it can fall prey to correlations that are only statistical artifacts.
Our software uses causal graphs, a popular tool in social science and epidemiology that embeds statistical best practices into easily-understandable visual form. They are often called DAGs, or directed acyclic graphs. ‘Directed’ means each arrow points in one direction. ‘Acyclic’ rules out feedback loops where a variable causally affects itself. Finally, ‘graphs’ mean networks, where variables are the nodes and influences are the links.
Not everyone in pricing needs to be an expert in statistical methodology, as long as they have a platform that can promote best practices, and steer them in the right direction. Far from a black-box AI model, our software helps you learn as the market grows, giving insights that can improve strategy for your entire firm.