In this episode of Pricing Friends, I had the pleasure of discussing the future of price optimization in retail with Dr. Sebastian Voigt. I shared how, at Quicklizard, we empower companies to design flexible pricing strategies that work seamlessly across various channels—whether that’s in different countries, online and offline, or between discount and premium banners. Our software enables real-time price adjustments, driving revenue increases of 8–10% and profit growth of 3–5%.
For me, pricing is a lifelong journey—a process of constant prototyping. Strategies must evolve to keep pace with market dynamics. That’s why I advocate for a blended approach: instead of choosing between “make or buy,” why not embrace “make and buy”? Combining internal expertise with external solutions often delivers the best results.
We also discuss my nuanced view on price elasticity: “It’s not the holy grail of pricing.” While it’s undoubtedly important, psychological and intangible factors also shape consumer behavior. At Quicklizard, we leverage AI to account for these additional trends and influences, all while enhancing—not overhauling—our clients’ existing pricing processes.
The transcript below summarizes the key Q&A highlights between Dr. Sebastian Voigt and Dr. Fabian Uhrich and is not a direct reading of the conversation.
Fabian, what was your journey into the pricing world? And how did you come to join Quicklizard?
I transitioned into pricing by combining psychology and analytics during my business administration studies. After roles in consulting (BCG), operational pricing (Zooplus), and academic research on behavioral pricing, I joined Quicklizard in 2024 to focus on product development for pricing tools.
What is Quicklizard, and how has it evolved?
Quicklizard began as a price crawler over a decade ago but shifted focus in 2018 to pricing optimization and AI. The pricing software integrates customer or third-party crawling data and caters to retailers, airlines, and hotels by optimizing prices dynamically using attribute-based algorithms.
We work extensively with retailers and brands, particularly in multi-channel and -country pricing and direct-to-consumer contexts. Retail, both online and offline, is a primary focus. Interestingly, we also have airlines and hotels on the platform. That wasn’t originally the plan, but we found our software fits well. Instead of products on a shelf, it optimizes routes or room bookings for specific calendar days.
We work with pure online players, pure offline players, and mostly multi-channel businesses. Our clients range from €50–100 million in annual revenue to several billions. In terms of enterprise readiness, we’re well-prepared for large-scale operations. Quicklizard is well-established across continental Europe — with 40-50% of revenues sourced from the region, but also serves clients in the US or even Australia.
Our software is attribute-based—a term that may come up frequently today. Essentially, every attribute describing a product, route, or service is utilized in the software and incorporated into the algorithms. This makes it highly adaptable to yield management scenarios, such as optimizing based on load factor or time to flight.
What are Quicklizard’s unique selling points?
- AI and Rule-Based Combination: We blend rule-based pricing with AI—the most effective results come from combining the two. Rules are essential to account for factors like manufacturer requirements, strategic business goals, how different categories are priced, and constraints such as corridors. Within these constraints, AI serves to challenge and refine the decisions of category managers or pricing managers, deriving better prices by letting the data speak.This combination is a clear competitive advantage for us. We offer both approaches and empower customers to work transparently within these frameworks. It’s a key strength we emphasize in our features and implementation.
Amongst German retailers, the majority still base pricing decisions on rule-based pricing. When it comes to pricing maturity, most companies are still at the Excel stage, which doesn’t involve much AI. Most of the clients Quicklizard works with are starting from a lower baseline. Among those who use Quicklizard software, there’s always a blend of the two concepts—some are 80% rule-based and 20% AI, while others are at 50/50.
- Omnichannel Flexibility: We support distinct pricing for multiple channels; any channel requiring a different price is treated as its own channel. Our software allows for independent pricing of each channel or dependency-based pricing, where Channel A is priced first, and Channel B follows as a derivative. Alternatively, both channels can be priced independently, with a harmonization corridor to ensure prices don’t deviate too much.This approach is ideal from a price fairness perspective—it’s the best compromise between differentiation and maintaining fairness. Once you move into full personalization, you have to be extremely cautious about which factors you use to avoid being perceived as of discriminating. That’s why channel-based differentiation can be the perfect balance.
About a quarter of large retailers differentiates prices for offline vs. online, even by search engine. It varies significantly by industry. In pharma, it’s nearly impossible to compete without channel differentiation. It often correlates with the intensity of competition in a given category. In electronics, for example, channel-based differentiation was once widespread but seems to be declining due to the generally high dynamics in the sector.
- Open Platform: Our pricing software is not a black box. Every price recommendation comes with a decision tree, showing exactly why the price was set as it was. This transparency is essential for price acceptance among those responsible for the P&L, such as category managers. They need to be able to understand and justify prices, both internally within their organization and externally to manufacturers.That doesn’t mean prices must be calculated in a simplistic way; it just means the logic behind them must be made transparent. This is particularly important when incorporating AI, as that can often feel less intuitive. It’s a solid compromise for making even AI-generated prices understandable within a given corridor, presented clearly in the decision tree.
Defining the pricing rules upfront is critical—it’s a key part of onboarding. Quicklizard helps clients define these rules, leveraging extensive experience. This step isn’t just about determining prices; it’s also about change management. Staff need to understand what’s happening: workshops to discuss what factors matter, which parameters to include, and how to prioritize them. This serves two purposes:
A. Ensure the mathematical logic fits within realistic constraints.
B. Give employees a voice, making them feel like co-developers of the pricing algorithms.Implementations where the “perfect” algorithm is deployed can be frequently overridden by staff due to mistrust. Over time, the automation will be bypassed, and the impact diminished. Starting gradually, building intelligence step-by-step, and involving employees ensures the opposite happens—acceptance and trust increase over time.
Additionally, we give customers the option to customize the software. Instead of presenting a hard-coded system, our pricing software allows clients to create custom functions using a Python interface.
What’s unique is that these custom scripts blend seamlessly into the software. For category managers, it looks and feels the same as the standard functionality. They won’t notice whether the algorithm running in the background is a Quicklizard default or a custom function created by their team.
This flexibility means we can handle 100% of use cases without locking clients into our system. I often compare Quicklizard to Excel. Like Excel, we provide built-in formulas for most needs. But when you hit a limitation, you can create your own VBA scripts—or, in our case, Python code.
If it feels too complex, clients don’t have to use this feature. Our solution architects and technical operations team are always there to assist. It’s worth mentioning that all data and models remain client-specific. Elasticities and algorithms are trained exclusively on each client’s data to maintain strict data separation.
- Speed: Initial deployment occurs within 12–16 weeks, assuming basic data like prices and costs are available. The solution is expanded step-by-step—adding categories, channels, and attributes—adapting to market changes and client-specific needs through continuous refinement. For smaller-scale clients, who might only operate in one country and one channel, implementation can be quicker relative to larger enterprises with multiple countries and channels, where organizational agility might be the bottleneck. The limiting factor is usually the client’s pace.
What measurable outcomes can clients expect?
Typical results range between 8–10% increase in revenue ora 3–5% improvement in profit. Another major benefit is the reduction—or even elimination—of manual work. Plus, the entire product catalog can be priced regularly. Without a tool, category managers can’t focus on the long tail.
The primary benefit depends on the client. For example, in direct-to-consumer businesses, the room for dynamic pricing is limited by nature. In those cases, the key benefits are consistency in pricing and a significant reduction in manual effort.
On the other hand, in sectors like consumer electronics retail, where margins are slim, even small pricing improvements can have a big impact. Efficiency is less critical here—extracting that extra 0.5% profit is what matters.
“Price elasticity is not the holy grail of pricing.” Why?
Although it’s used extensively, elasticity alone isn’t the solution. In theory, it could solve all pricing problems, but only if we could measure every influencing factor, which is impossible. For instance, we can measure competitive prices, stock levels, and trends on Google. However, factors influencing consumer behavior often remain intangible.
Moreover, the data collected must be statistically significant. As the granularity of data increases, the number of observations per segment often becomes too small to draw reliable conclusions. In practice, elasticity only covers a portion of the catalog, typically the faster-moving SKUs. For slower-moving items or the long tail, we rely on intelligent clustering to estimate elasticity.
Elasticity is a crucial piece of the pricing puzzle, but it’s just one piece. AI can enhance our ability to capture trends and include more influencing factors, but in pricing—where structured data and numerical inputs dominate—the room for disruption is inherently limited compared to fields like image or speech recognition.
Factors like long-tail items and new products, where historical data is scarce, will always present challenges. That’s why a multimodal approach works best: combining AI, rules, elasticity, stock levels, and cost-based minimum prices. Pricing will always require a mix of methods.
How does Quicklizard account for challenges like trends?
Stable assortments are great for algorithms because they have time to learn. Temporary assortments, like seasonal items or fast trends, require a different approach.
Take an iPhone, for instance. When a new model launches, we might manually classify it as the successor to its predecessor and assume its characteristics will be similar. This “inheritance” gives algorithms a head start.
One of the first steps we take in most projects is segmenting products using 20 metrics. These include factors like online traffic costs, shopping cart position (e.g., first, second, or third item in the cart), and lifetime value. We apply these metrics to the new iPhone, assuming its performance will align with its predecessor. Then, as real data comes in, the scoring model is updated every four to five weeks to reflect the market reality.
Now, for fashion: here, consumer behavior is influenced by learned patterns, like end-of-season sales. Algorithms can handle these markdowns effectively. For example, based on the historical performance of similar items, we might set a discount schedule.
The algorithm continuously learns from demand data. If a predicted 20% boost in sales from a discount doesn’t materialize, the algorithm adjusts for the next week, possibly by increasing the discount or accelerating the timeline. Short-term algorithms work well for these assortments, compensating for the lack of long-term data by borrowing patterns from broader categories and making adjustments in real time.
What is Quicklizard’s cost structure?
Quicklizard’s pricing structure is based on several factors, including (1) how often prices are adjusted, (2) the number of SKUs being priced, and (3) the number of channels involved. These factors determine the computational resources required on our end, which drives our licensing fees.
How does a retailer decide between building their solution or buying one?
It doesn’t have to be “make or buy”—it can be “make and buy.”
This ties back to Quicklizard’s open platform philosophy. Many elements of a pricing tool don’t need to be reinvented: the user interface, governance models, approval workflows, APIs, upload interfaces, etc. These standard features don’t drive profitability or revenue—they’re just foundational.
It makes sense to use an off-the-shelf solution like ours for these elements. The real differentiation comes in defining pricing strategies, segmenting products, and developing algorithms. That’s where clients can add their “magic sauce.”
Quicklizard’s pricing software allows them to do this directly by customizing their approach using Python.
Has there been any price that stood out to you in the last few weeks, e.g. “Price of the Week”?
It’s been a little while now, but it’s still a pretty blatant and probably a running joke in the pricing community — Black Friday, or rather, Black Week by now. And also a good reason why Omnibus Compliance was invented. But the classic example is the price that stays the same, and stays the same. Then Black Friday comes along, and suddenly, the price is still the same, but it’s marketed as if it’s especially cheap by inventing an imaginary higher price. And I still find this a very suitable example of why Omnibus Compliance in Europe is such a sensible regulation – that we can of course automatically incorporate. At the end of the day, consumers aren’t stupid, and fair pricing is, of course, very important.
Why Quicklizard? Where does the name come from?
It comes from the old days of price crawling, where the lizard, which can change its color, would inconspicuously but quickly crawl through the relevant websites and collect competitive prices. We thought the metaphor still fits well because we’re still fast in our implementation. That’s why we kept the term.
About Dr. Sebastian Voigt, host of 'Pricing Friends'
Before joining hy, Dr. Sebastian Voigt held leadership roles at Axel Springer SE and ProSiebenSat.1 Media SE, where he focused on pricing, monetization, and digital transformation. Earlier in his career, Sebastian spent eight years at Simon-Kucher & Partners, specializing in strategy and pricing. He holds a PhD from TU Darmstadt, specializing in digital market monetization, and a degree in Business Administration and Computer Science from TU Braunschweig.
