AI: do sales forecasts create their own confirmation?
A sales team may use AI to rank customers by potential and then concentrate its efforts on those with the highest scores. The managerial question is precise: how can we distinguish an accurate forecast from an outcome produced by the resources allocated on the strength of that forecast?
What the literature establishes
Robert King Merton (1948, The Antioch Review) describes the self-fulfilling prophecy: an initially false definition of a situation prompts behaviour that eventually makes it true. His article proposes a sociological mechanism, not an experimental evaluation of AI. Applied to sales rankings, this framework invites us to examine whether the predicted potential is discovered by the tool or partly created by the attention given to the customer.
When forecasts organise resources
Fabrizio Ferraro, Jeffrey Pfeffer and Robert Sutton (2005, Academy of Management Review) analyse how theories can become self-fulfilling by shaping institutional arrangements, norms and language. Their theoretical contribution illuminates how a representation can organise practices until it produces outcomes that appear to confirm it. For a team using AI, the mechanism to examine is concrete: rankings determine visits, offers or follow-up time, which subsequently influence sales.
The premature conclusion
It would be excessive to conclude that a forecast which influences action is necessarily poor. Directing more resources towards the most promising opportunities may be precisely the intended goal. The distinction is between two questions: does the tool anticipate outcomes under comparable resource conditions, and does the allocation policy it informs improve overall results? (our executive and employee training programmes)
What observed outcomes cannot settle
Neither article measures the accuracy of contemporary AI or its commercial effectiveness in Vevey. Simply comparing sales to high- and low-ranked customers conflates their initial potential with the effects of different sales treatment. Without records of the resources committed and a relevant comparison, the correspondence between scores and sales remains ambiguous.
A practical check in Vevey
As part of SHR — Swiss Human Resources’ « Manager à l'ère de l'intelligence artificielle » programme, an exercise could focus on a customer portfolio at the headquarters of a global food group or an SME on the Riviera. During a campaign defined in advance, retain the initial scores and record the sales time, contacts and orders for each customer. Within a subset of comparable customers, and where commercial constraints allow, randomly allocate portfolios to either score-guided resource allocation or usual practice, with equal time budgets. The measure to examine is the difference in orders per sales hour between the two approaches: it provides evidence about the allocation policy tested, without by itself establishing the model’s general validity. To go further: explore the Managing in the Age of Artificial Intelligence training in Vevey in the canton of Vaud, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Vevey
- managing in the age of artificial intelligence
- Vevey
- research
- training vevey