This interview analysis is sponsored by 7Learnings and was written, edited, and published in alignment with our Emerj sponsored content guidelines. Learn more about our thought leadership and content creation services on our Emerj Media Services page.

Retail runs on thin margins. General merchandise retailers posted an average net margin of 5.6% in the most recent NYU Stern analysis of public company filings, and grocery retailers cleared just 1.3%. Retailers have also added products and sales channels faster than their systems track them, multiplying the number of pricing, marketing, and inventory decisions that must be made every day.​

Demand moves fast enough that the U.S. Census Bureau updates its retail sales estimates every month rather than on a slower cycle. Inside most organizations, the teams responsible for those decisions work separately, and the underlying forecasting models often operate in silos, solving one problem at a time rather than accounting for their interactions. Margin absorbs the difference.​

Felix Hoffmann, Founder and CEO of 7Learnings, joined Emerj’s Yolandi de Weerdt on the AI in Business Podcast to discuss how retailers can unify pricing, marketing spend, and inventory into one coordinated commercial system that delivers stronger margins, clearer demand signals, and higher‑quality decisions.

This article examines how unified commercial decisioning, predictive simulation, and sequenced automation can materially improve retail margin and decision quality.

• Unified pricing–marketing–inventory decisioning for margin improvement: Use one commercial view so teams act on the same demand and stock picture, reducing margin loss from misaligned discounts, campaigns, and replenishment decisions.
• Predictive simulation for outcome‑based commercial optimization: Use models that reveal the expected impact of each pricing or marketing move so leaders can select the option that best advances growth or profitability before execution.
• Sequenced automation for provable commercial ROI: Apply automation to one high‑leverage decision area at a time so returns can be measured cleanly and transformation risk stays contained instead of spread across all commercial functions.

Listen to the full episode below:

Episode: Unified Predictive Decision Making for Retail Growth – with Felix Hoffmann of 7Learnings

Guest: Felix Hoffmann, CEO at 7Learnings

Expertise: Retail AI, Pricing Optimization, Revenue Optimization, Data Science

Brief Recognition: Felix Hoffmann is founder and CEO of 7Learnings. Previously, he led global price optimization at Zalando, where he managed the company’s pricing algorithm, and spent six years as a strategy consultant at Kearney. He holds a master’s degree in Management from ESCP Business School.

Unified Pricing–Marketing–Inventory Decisioning for Margin Improvement

Retailers often treat commercial misses as surprises, but Felix’s experience shows they’re usually the predictable result of teams acting on isolated signals.

Felix Hoffmann’s view of siloed commercial decisions is structural: retailers have multiplied their SKUs, channels, and promotional levers, but the underlying decision processes have not kept pace. Pricing, marketing, and inventory teams still operate on spreadsheets and isolated rules, each optimizing for its own function without visibility into how those decisions affect the others. The result is not random commercial misses — it is predictable margin leakage created by teams acting on partial information.

Felix’s Zalando example makes the blind spot concrete. A UK marketing push sold out a limited run of sneakers almost immediately, and the local team celebrated the result as a clear win. But because the campaign wasn’t connected to global stock levels or pricing strategy, the company lost the chance to sell those same units at a higher price in other markets where demand was still unmet.

What looked like success in one region was, in commercial terms, a margin leak — the direct consequence of pricing, marketing, and inventory acting without visibility into each other’s decisions.

The same blind spot shows up directly between pricing and marketing. Pricing and marketing teams typically operate without a shared line of communication, even though a pricing decision changes what a marketing decision should be. A significant price increase on a product drives conversion down; once conversion drops, the marketing spend or targeting behind that product needs to change as well.

Without a connection between the two functions, marketing continues operating on assumptions the pricing decision has already invalidated.

The same disconnect appears on the inventory side. Reorder decisions are often based on last year’s sales without accounting for whether the price is changing this year. Selling 1,000 units at a loss is not a reason to order 1,000 more, and a planned price increase should reduce reorder quantity, not repeat it. The reverse holds too: a planned price cut expected to lift demand should trigger a larger order, not the same one a pricing‑blind process would generate.​

Felix’s guidance for leaders is to interrogate decisions that appear successful at the local level. Three questions reveal whether a pricing, marketing, or inventory move was made without cross‑functional visibility:

• Would this product have generated more margin if allocated to a different market?
• Does the reorder quantity reflect the price the business intends to charge next season?
• Is demand being evaluated across the full commercial footprint, or only within the local signal?

These questions aren’t theoretical. They are the exact counterfactuals retailers fail to model — and the exact places margin disappears when commercial decisions are made in isolation. Felix’s point is that unified decisioning isn’t a technology milestone; it’s the moment teams stop mistaking local wins for commercial success and begin acting on a shared view of demand, price, and stock across the business.

Predictive Simulation for Outcome‑Based Commercial Optimization

Felix highlights a structural limitation in how commercial decisions are made today: retailers can describe the outcome of a price or marketing move, but they cannot quantify what would have happened had they chosen a different path.

The counterfactual is missing. In his experience, this gap exists because the underlying data foundation is incomplete; many retailers lack accurate purchase price records, have never reviewed which historical price changes actually succeeded, and do not maintain structured visibility into past demand and marketing activity. Without that baseline, alternative scenarios cannot be modeled reliably.

He uses a mapping analogy to illustrate the shift. A mapping app does not simply show distance; it shows multiple routes, the trade‑offs between them, and the fastest path once the user sets a destination. Predictive commercial systems behave the same way. Once a retailer defines a target — a margin threshold, a revenue lift, a demand outcome — the model evaluates the available commercial paths and identifies the combination that reaches the target most profitably.

The model compares:

• Alternative pricing paths — different price points, discount depths, and timing
• Alternative marketing paths — spend levels, channel allocation, and promotional intensity
• Profitability trade‑offs — the margin impact of each pricing‑marketing combination
• Route efficiency — the fastest or most profitable way to reach the defined commercial target

Felix explains the shift toward target‑driven commercial decisions:

“It’s the same for our algorithms. You can say, I want to grow 10% more than what I’m currently predicted to grow next week. Give me the decisions that get me there in the best possible way in terms of profitability.”

• Felix Hoffmann, CEO at 7Learnings

Simulation becomes operational only when the historical record is trustworthy. With accurate purchase prices, validated past price changes, and clean demand and marketing histories, retailers can compare multiple commercial paths before committing to one. The work shifts from debating actions to defining outcomes — letting the model work backward to the combination of decisions that achieves them.

Sequenced Automation for Provable Commercial ROI

Hoffman is clear that retailers run into trouble when they attempt to automate pricing, marketing, and inventory simultaneously. In his experience, the sequence matters because each function depends on signals produced by the others. Automating them in parallel forces teams to make decisions without the information those systems are meant to generate.

In most retail environments he’s worked with, pricing is the natural starting point. It moves quickly, affects margin directly, and provides immediate feedback on whether a decision worked. Once pricing decisions become predictive, marketing typically follows, because spend allocation interacts directly with price — promotional intensity, channel mix, and budget levels all depend on knowing how price will shape demand. Inventory comes last.

Reorder logic requires visibility into future price and expected demand, not just last year’s sales, and those signals only become reliable once pricing and marketing are operating predictively.

The progression isn’t universal. A luxury retailer with stable pricing behaves differently from an off‑price retailer where price changes constantly. In practice, he sees two factors determining the right starting point:

• Team readiness — automation succeeds first where a team is willing to change how it works.
• Data reliability — automation fails fastest in functions where historical records are incomplete or inconsistent.

Before scaling, he emphasizes the need for proof. In his approach, early automation is tested through controlled comparisons so retailers can see the commercial impact of predictive decision‑making before expanding into adjacent functions. The goal is not to automate everything at once, but to demonstrate measurable ROI in one area and extend only when that success creates the conditions for the next.

In his view, sequenced automation is less about technology maturity and more about operational honesty: start where the data will support predictive decisions, prove the return, and extend automation only when the first step has earned its right to scale.