Sign up

The wisdom of the staff

May 12, 2017

Merging AI & human intuition: the secret sauce. An alliance between artificial intelligence and human experience can perfect your sales forecast, and therefore align all data-driven strategies: inventory, pricing, promotions.

Big data + human intuition = profits

Algorithms are fed by big data. But even real-time data represents what has already happened, a past event. Human experience, however, allows managers to quickly formulate hypotheses about the future.

Profitably combining AI and human intuition results in 40 percent lower forecast error, in turn making pricing and sales decisions more accurate, and therefore growing sales by 10 percent or more.

How to create structure around human intuition

Evo Pricing’s model is based on predictive analysis of big data and input from management. At one retailer for example, over 300 shopkeepers, front-line staff, and managers, feed their structured input through a simple “trading-like” system based on non-monetary incentives.

Variables such as past sales, geographical area, climate, social media, and product characteristics—in essence, big data— are used for the initial prediction. The staff can then change this forecast, even to the point of proposing the direct exchange of goods across stores (when profitable).

In essence, the system creates a first proposal, field staff and management can change it, and finally the system clears the maximum possible number of human inputs based on budget and stock constraints, starting from the requests with the highest potential. This leads to better inventory allocation, and in turn, to more targeted promotions and better pricing.

The joint requests of the staff, in fact, have been proven to predict sales up to 8 weeks in advance – therefore pointing out popular items for re-order, opportunities to increase prices, and local areas in need of promotional stimuli.

Proven results

A team of business experts, data scientists, and researchers from universities like Harvard and MIT have proven the effectiveness of this model. In effect, human intuition improves the results of artificial intelligence significantly more than using either method individually.

As an example, an Italian fashion retailer has implemented Evo Pricing and noticed a +16% increase in sales during the initial pilot. This was mainly thanks to “middle of the pack” items: while slow movers and fast sellers tended to perform similarly across stores, the bulk of the inventory had wide store-level variations that could be profitably exploited by improving the store-level forecast.

Turin (in northern Italy, with typically colder and earlier winters) and Palermo (in southern Italy, with typically shorter and warmer winters) have a markedly different climate — what sells at any given time in Turin can be different from what’s best to sell in Palermo.

The stock of padded jackets, which don’t usually sell in warmer climates, can be reassigned to satisfy demand in cooler areas, when the opportunity arises.  Easy to describe, hard to take all these idiosyncrasies into account and leverage them with profit. But local staff is very well aware of their own rapidly changing needs!

The Novara store – somewhere far from the sea – attempted, for the first time in history, to request beach products such as costumes. Since this category was never sold before, no big data system would have ever predicted an opportunity. However, lo and behold, the small initial sample sold quickly, and thereafter the machine learned a previously unknown trend and developed it further.

Solid scientific roots for the ‘wisdom of the staff’

Giuseppe Craparotta, senior data scientist at Evo Pricing says, “In a collaboration between the University of Turin and Harvard, we have proven how the requests from the staff ‘trading system’ forecast supply and demand 8 weeks in advance and grow sales and profits significantly, by a double digit percent figure”.

This approach is revolutionary, and not just for retail forecasting. But retail staff is often surprised to discover how their insider customer and product knowledge are key to the success of Evo Pricing’s algorithm and science.

About the author

Kathy Edens is a digital content creator who enjoys writing about cutting edge technology and how it can disrupt and innovate the way a business operates.

Hey! Was this page helpful?

We’re always looking to make our docs better, please let us know if you have any
suggestions or advice about what’s working and what’s not!

Send Feedback