Demand Forecasting Retail Buyers Actually Trust
Overview
Our client had an accurate forecast that buyers overrode nine times out of ten, which made it worth precisely nothing.
Nikolaos Papadopoulos
Sr. Data and BI Analyst / fairsystems
A forecast nobody acts on is not a forecasting problem. It is a trust problem, and it is usually earned: models that cannot explain themselves, that are wrong in ways buyers can predict, and that ignore the commercial knowledge buyers hold. fairsystems was engaged by a retail group whose forecasting investment had produced good statistics and no behaviour change.
The Client
Our client is a multi-category retailer operating physical stores alongside a growing online channel. They had invested in a demand forecasting platform two years earlier. The statistical accuracy was respectable in aggregate, but category buyers overrode the recommendation in the overwhelming majority of cases, reverting to their own spreadsheets. The result was the worst of both worlds: the cost of the platform, and inventory decisions still made on individual judgement with no consistency across categories.
Roadmap
1
Assess
fairsystems started with the override data rather than the model, because every override is a buyer telling you something the forecast missed. Our consultants analysed two years of overrides against subsequent outcomes to establish where buyers were genuinely adding information and where they were simply anchoring on last year. The results were split almost evenly, which was the crucial finding: the model was wrong in specific, learnable situations, principally around promotions, weather-sensitive lines and new product introductions, and buyers had real knowledge in exactly those cases.
2
Deliver
Our team rebuilt the forecast to incorporate the signals buyers were using informally, adding promotional calendars, weather data and structured attributes for new product introductions where no sales history exists. fairsystems made every recommendation explainable, showing the drivers behind each number so a buyer can see why it moved. We redesigned the override workflow so that a buyer records the reason, turning disagreement into structured training data. We also set category-level accuracy targets and gave buyers a dashboard showing their own override performance over time.
3
Continue
fairsystems runs a quarterly forecast review with the client's planning function, examining override patterns to find the next set of situations the model handles badly. We retrain on the accumulated override reasons and extend coverage as new categories come on stream. Our consultants continue to coach buyers on when to trust the model and when their own knowledge genuinely adds value, which has proved more valuable than further statistical refinement, because the constraint was never the mathematics.
Solution Details
A forecast nobody acts on is not a model problem. It is a trust problem.
Buyers overriding the forecast in most decisions
Deliverable: fairsystems analysed two years of overrides against outcomes, establishing precisely where the model was wrong and where buyers were merely anchoring.
Model blind to promotions, weather and new products
Deliverable: Our consultants added promotional calendars, weather signals and structured attributes for products with no sales history, closing the real accuracy gaps.
Recommendations arriving as unexplained numbers
Deliverable: We made every forecast explainable, exposing the drivers behind each figure so buyers can evaluate rather than simply accept or reject it.
Overrides discarded instead of learned from
Deliverable: fairsystems redesigned the workflow to capture a structured reason with every override, converting disagreement into training data.
No visibility of whether overrides improved outcomes
Deliverable: Our team gave each buyer a dashboard showing their own override performance over time, making the question answerable individually.
Parallel spreadsheets running alongside the platform
Deliverable: We closed the gaps that made the spreadsheets necessary, removing the reason buyers maintained a shadow process.
Accuracy measured in aggregate, hiding category failure
Deliverable: fairsystems set category-level accuracy targets, exposing the categories where aggregate figures had concealed poor performance.
No mechanism to find the next model weakness
Deliverable: Our consultants established a quarterly override review that systematically surfaces the situations the model still handles badly.
Result:
Buyers now accept the great majority of recommendations, and the overrides that remain are concentrated in situations where buyer knowledge genuinely does add value. Forecast accuracy improved, but the decisive change was that the forecast is used: stock availability rose while inventory holding fell, which cannot happen when every category is planned on a different person’s spreadsheet. The platform the client had already bought finally returns something, and the override review keeps finding the next weakness before it erodes trust again.
0
Reduction in forecast overrides
0
Forecast accuracy improvement
0
Inventory holding reduction
0
Stock availability improvement
0
Categories meeting accuracy target
Testimonials
How our services bring about success
Annika Nkosi
Lead Process Analyst
fairsystems’ team brought a unique combination of skills to our business assignment and was often sought out to tackle the most complex and integrated challenges.
Marie-Anne Badeaux
Programme Manager
fairsystems will challenge your thinking, provide you with alternative perspectives and, where appropriate, actively coach/mentor/advise you to generate and co-create solutions.
Justin J. Spratt
General Manager
Fairsystems consults impeccably to understand organizational needs and then brings innovative business relevant ideas and solutions to the table.
Mohamed Bin Naser
Chief Financial Officer
I am impressed with fairsystems’ team ability to grasp complex, accounting system problems and quickly find a simple, straightforward and affordable approach.
Ella-May Robinson
HR Manager
fairsystems has an incredible gift for making the complex appear simple and without being simplistic. It is a pleasure to work with them as they always deliver to an exceptionally high standard.
Jamal Aagha
Chief Executive Officer
There are some great companies in the media industry and the social space, but we have yet to see anyone take it to the level that we have with @fairsystems.
Allen Gibson
Operations Manager
The most valuable lesson we learned is that there are no shortcuts. Every minute we spent dissecting and learning was rewarded.
