All Data Comes from One Place: The Past

Rory Sutherland said it the best: all data comes from one place – and that is the past. Every trend, every customer behaviour, and every sales pattern has its roots in historical data. As much as we speak about forward-looking insights, artificial intelligence, and predictive analytics, it is important to understand that all of these mechanisms are reliant on one single, irreplaceable source - past data.

The Foundation of Forecasting

Forecasting is not a guessing game. It is a structured exercise in pattern recognition. Whether you’re trying to anticipate consumer demand, optimise inventory, or plan promotions, the accuracy of your future view depends entirely on the depth and cleanliness of your historical records.

At BD-Nav, we store and analyse up to 36 months of detailed transactional history, enabling us to identify, as close as possible, the real demand signals. What is the real Rate Of Sale (ROS)? What happened between time of out of stock to time in stock?  What was the root cause creating the situation and what can be done to eliminate this going forward?

Where did the sales go? How did the promotion influence the sales of the competitor during the period over the promotion? Was it only the price point or were there other influencers?

These are questions requiring clear answers.

Forecasting is a combination of past fact analyses (science) and the predictions of what could or should’ve happened (art).

The Illusion of Real-Time Without Context

Many businesses invest in “real-time data” tools, hoping that immediacy will drive better decision-making. But real-time without context could lead to misinterpretation. For example, a sudden sales spike could be viewed as a demand surge - unless history tells you it's a seasonal promotion or once-off event. More importantly, was there enough time for your operational teams to attend to the identified problem?

 More data and analytics don’t necessarily empower you to make better decisions.

Why minimum 24 to 36 Months of transactional “influencer data”?

The choice of timeframe is not arbitrary. In retail, a 24-to-36-month history provides enough data to capture seasonal patterns, promotional cycles, and year-on-year category shifts. It allows for the exclusion of outlier events and gives weight to emerging trends. Never

Too short a timeframe, and your data becomes reactive. Too long, and your data may become irrelevant. The 24 to 36-month window strikes the balance between relevance and statistical power.

Clean Data, Real Decisions

Access to historical data is not enough - it must be structured, cleansed, and organised in a way that reflects the business as it operates today. Duplicate SKUs, outdated hierarchies, and poor master data all distort reality.

At BD-Nav, we place equal weight on the quality of the data and the quantity. Because clean history leads to clean insights, and clean insights drive confident decisions.