Free Webinar
Predict, Plan, Profit: Mastering Pricing in the AI Era
How behaviour-led AI transforms pricing decisions from elasticity guesswork to profit-accurate optimisation.
Why Pricing Needs Behavioural AI
AI is rapidly entering pricing conversations, but most tools still stop at conceptual guidance rather than real decision support. In this session, Ingo Reinhardt shows why general-purpose AI struggles with profit optimisation and how traditional elasticity-based methods fail once prices, competitors, or pack structures change.
Drawing on real pricing examples and Buynomics’ virtual shopper technology, the session explains how modelling shopper behaviour, rather than static averages, enables accurate predictions across price, promotion, and mix. The focus is not automation for its own sake, but improving commercial decisions in complex, competitive markets.
What You'll Learn
- Why generic AI tools fall short in pricing decisions
Learn where language models help conceptually but fail when economic logic and optimisation are required. - How elasticities change as soon as prices move
Understand why static elasticity tables break down in volatile pricing environments. - What virtual shoppers model that regressions cannot
See how individual-level preferences and decision rules explain real market outcomes. - How to simulate price, pack, and promotion interactions together
Explore portfolio-level optimisation across brands, sizes, and competitors. - How behaviour-led models improve forecast accuracy
Learn why modelling the decision process delivers more reliable commercial predictions.
Meet the Speaker
 |
Ingo ReinhardtCo-founder and Managing Director at Buynomics |
Before Buynomics, Ingo was a Senior Director with Simon-Kucher & Partners, a global leader in pricing. He holds a Ph.D. in Management from the University of Cologne and Master's degrees in Management and Mathematics. Ingo was a PostDoc at the University of Oxford and published in the Strategic Management Journal.
Session Highlights
Generic AI can explain pricing concepts, but it cannot make pricing decisions
The session opens by showing how ChatGPT can describe elasticity theory correctly yet fails to answer a concrete profit-maximisation question. This gap illustrates why language models are useful for framing problems, but insufficient for real pricing decisions that require numerical optimisation and behavioural modelling. [05:53]
Price elasticity is a weak decision input once prices start moving
Using a simple demand curve example, the speaker shows how a 10% price increase can shift elasticity from −2.0 to −2.7 immediately. This makes static elasticity tables unreliable for decision-making, especially in inflationary or volatile environments where prices and competitive contexts change frequently. [15:19]
Most ‘AI pricing’ claims stop at better estimation, not better decisions
The session challenges the idea that machine learning adds value by producing more accurate elasticities. The real limitation is not estimation quality, but the fact that elasticities collapse complex shopper behaviour into a single, unstable number that ignores portfolio and competitive dynamics. [16:15]
Optimisation requires modelling the decision process, not just the outcome
A key distinction is made between observing sales outcomes and understanding the underlying choice process that produces them. By focusing on shopper preferences and trade-offs, rather than aggregate sales responses, AI can model how decisions shift when prices, pack sizes, or features change. [18:19]
Virtual shoppers make portfolio and competitive interactions explicit
Through a live example, the speaker demonstrates how virtual shoppers reallocate demand across sizes, brands, and competitors when one SKU changes price. This reveals cannibalisation and substitution effects that single-SKU or single-elasticity approaches systematically miss. [22:03]
Better predictions matter because precision compounds into profit
The session concludes by comparing traditional regression forecasts with virtual shopper outputs, showing materially closer fit to actual sales. The argument is not academic accuracy, but that more realistic behavioural modelling enables more confident decisions across pricing, promotions, and product changes. [27:26]
Q&A
What is the main limitation of using ChatGPT for pricing decisions?
ChatGPT can explain pricing concepts but does not optimise profit or model demand responses. It lacks behavioural and economic structure required for pricing decisions [06:06].
Why are price elasticities unreliable for decision-making?
Elasticities change with price levels, competitive moves, and assortment context. Static values quickly become inaccurate when conditions shift [15:19].
How do virtual shoppers differ from traditional regression models?
Virtual shoppers model individual preferences and choices, while regressions only fit average relationships. This allows more accurate prediction of market dynamics [18:30].
What data is used to train virtual shopper models?
Models are trained on historical sell-out, panel, and promotion data to infer shopper preferences and behaviours [24:26].
Can virtual shoppers simulate competitive reactions?
Yes, they account for substitution across brands and pack sizes when competitors change prices or features [22:03].
Why does behavioural modelling improve forecast accuracy?
Because it reflects how shoppers actually choose, rather than assuming stable averages that break under change [26:20].