But then a bright project leader decided to use predictive models on how children crops would behave based on the properties of their parents and environmental conditions they grow in, instead of actually planting them, waiting for the year to pass, harvesting and calculating the results.
Building models and discovering which properties of parent crops do impact the Key Performance Indicators - like yield and resistance, allowed the team to focus on 10 features like this instead of 400 in a several weeks, as well as run “in-silico” experiments on a computer by varying only the features that matter. Ten knobs to turn instead of 400 created the necessary focus for the team. Predictive models which instantaneously predict the likely outcome of your experiments saves massive amounts of time and resources
Quickly discovering what worked and what did not, and using predictive models to guide future experiments the team has come up with a final product in just six years (instead of the planned 15). The company reduced the time to market by 60% in a course of just two and a half years!
This did not only boost team bonuses and promotions, and equated to huge savings (tens of millions) on data collection, but gave the company nine years of competitive advantage of being the first on the market which made investors very happy:
- 60% shorter time to market
- Tens of millions costs saved
- Hundreds of millions ROI gained
...all because they KNEW what to do their their data!
IF knowing what to do with your data is something you are struggling with - be aware that the solution to your problem exists. Introducing advanced analytics during the data collection process always results in the massing ROI improvements and a shorter time to market for new products.
This is not difficult.
- To check whether you are ready to reap the benefits of data-driven product design go through this simple check list.
- Think about your data collection process. Are you extracting all the benefits of all that effort?
- Think about what you truly need to know how to optimize your product, i.e. make a list of your KPIs. Are you collecting data for each of them?
- Assuming you would know what truly drives your KPIs, think about how it would impact your personal and business goals. Write this down.
- Is your biggest challenge not knowing what the true KPI drivers are? If yes - search for space-filling designs, which would allow you to maximize the information of the problem with a smaller amount of experiments.
- Is your biggest challenge not knowing how exactly the KPI drivers impact the KPIs? If yes - check non-linear predictive analytics and ensemble-based modeling, which would allow you to build more robust predictions to guide future experiments