Understanding your competitors is essential in a marketplace so focused on selling a largely commoditised product differentiated primarily on price. Not only does this information allow you to understand price competitiveness at a tactical level, but it allows you to pick realistic business strategies based on the cost of acquiring customers. Essential for feeding into optimisation and demand models, it also plays a key role in helping you to understand risk cost outside of your sales footprint.
- Improve conversion model accuracy by calculating price distance to the market and feeding in as a new feature.
- Use as a proxy for risk cost where data in unavailable or unsuitable, such as for a startup, product, or footprint expansion.
- Understand strategically which customers you can compete for and which segments are unachievable without wider changes.
- Our models achieve ~15% of customers within +/1% of the true price, and ~50% within +/- 5% of the true price.
- Mean absolute error of £72, and an r2 score of 0.946.
- Consistent accuracy results across all premium ranges, indicating that both low and high risk customers are modelled accurately.
- Our market leading models are built in python and deployed in ML Ops environments, but we can adjust to suit your platform.
- We take great care to prepare the data correctly and efficiently, creating new features where they will add benefit.
- We fully test our models on test data batches, both within the time period and outside it for longevity assurance.
- Often startup companies have no risk data meaning that any attempt to build a risk price can be perilous.
- Startups can instead use our market model (plus overlays) to set their prices, gaining customers without incurring poor results.
- This can be thought of as a data acquisition strategy, with the market model eventually being replaced by risk models.