E-Commerce Growth Through AI-Powered Pricing Strategies

At Balkan eCommerce Summit 2026, Simeon Lukov, Founder and CEO of Dynamic Pricing, explored how eCommerce businesses can increase revenue and profitability not by spending more on advertising, but by optimizing pricing strategies through AI-powered dynamic pricing models. The lecture focused on modern pricing experimentation, intelligent discounting, inventory-aware pricing, and AI-driven personalization designed to maximize both revenue and profit from existing traffic.

Pricing Optimization Is Becoming a New Growth Lever

The presentation began with a simple but powerful idea:
instead of continuously increasing advertising budgets,
eCommerce businesses should focus on:

extracting more value from their existing traffic.

According to the lecture,
pricing itself has become one of the strongest growth levers available to online retailers,
especially when combined with:

• AI
• automation
• and real-time experimentation.

AI-Powered Price Testing Can Increase Margins Without Losing Sales

One of the core topics discussed was:

price testing.

The speaker explained that many merchants underestimate how much pricing flexibility they actually have. Instead of applying one fixed price,
AI systems can test:

• two
• three
• or even up to ten different prices for the same product.

For example:
a product priced at:

• €35
• €39
• or €45

may generate:
the same number of orders,
while significantly increasing total profit margins.

The lecture emphasized that:
small price increases across hundreds or thousands of products can create:

substantial additional revenue over time.

Bigger Discounts Are Not Always Necessary

Another major topic was:

optimal discount testing.

According to the speaker,
many retailers automatically assume they need:

• 30%
• 40%
• or even 70% discounts
  to generate conversions.

However,
AI-driven testing often shows that:
the same sales volume can sometimes be achieved with:

much smaller discounts,

preserving significantly higher margins.

The key message was:

“Don’t undercut unnecessarily.”

Instead of maximizing discount size,
brands should identify:

the optimal discount level.

Traditional A/B Testing Is Becoming Outdated

The lecture also compared:
traditional A/B testing
with newer AI-driven frameworks such as:

multi-armed bandits.

Unlike classic 50/50 split testing,
multi-armed bandit systems dynamically allocate more traffic toward the better-performing price in real time.

This creates several advantages:

• faster learning
• better performance during active campaigns
• less revenue loss
• and higher profitability while the test is still running.

The speaker gave Black Friday as an example:
traditional A/B testing may identify the best price only after the campaign ends,
while multi-armed bandits optimize traffic distribution:

during the campaign itself.

Pricing Should Be Optimized for Profit – Not Just Conversions

A central insight throughout the lecture was that:

more orders do not always mean better business performance.

The speaker explained that:
many businesses focus too heavily on:

• conversion rate
• transaction volume
• or total orders,

while ignoring:

profit per visitor.

The lecture introduced three key optimization metrics:

• conversion rate
• revenue per visitor
• and profit per visitor.

In some cases,
slightly lower order volume can still generate:
higher revenue and stronger profitability,
especially when return costs and operational expenses are considered.

AI Can Dynamically Manage Markdown Pricing

Another important topic was:

markdown optimization for inventory management.

The speaker demonstrated how AI models can automatically:

• monitor inventory velocity
• analyze sales performance
• and adjust discount levels dynamically.

For example:
seasonal products such as winter jackets can gradually move through a:

pricing ladder,

where discounts increase only when inventory movement slows down.

If products sell well,
the system avoids unnecessary markdowns.
If inventory remains unsold,
AI recommends stronger discounting before the season ends.

This approach was presented as particularly useful for:

• fashion
• grocery
• and products with expiration windows.

Personalized Pricing Uses Customer and Context Data

The lecture also explored:

adaptive and personalized pricing models.

The AI system can consider:

• customer behavior
• page visits
• holidays
• weekdays vs weekends
• seasonal demand
• competitor pricing
• and customer lifetime value.

For example:
prices may increase during weekends or before major holidays,
when purchase intent is naturally higher.

The speaker also explained that:
AI models help merchants avoid unnecessary competitor price wars by analyzing whether competing stores actually impact conversion performance.

AI Personalization Must Respect Legal Boundaries

An important point raised during the lecture was:

regulatory caution around personalized pricing.

The speaker noted that:
personalized pricing practices differ between:

• the United States
• and European markets,

where stricter regulations around discrimination and transparency apply.

Because of this,
the company applies personalized pricing selectively depending on geography and legal frameworks.

Similar Product Matching Enables Large-Scale Optimization

The lecture also addressed a challenge common in industries like:

• fashion
• furniture
• and lifestyle products:
  the absence of truly identical products.

To solve this,
AI models identify:

highly similar products

and use similarity analysis to optimize pricing across broader product groups.

According to the speaker,
the system can process:

up to half a million products per day.

Reporting and Transparency Remain Critical

The final part of the lecture focused on:

reporting and business visibility.

Since pricing directly affects:

• finance
• margins
• and profitability,

the platform provides dashboards comparing:

• previous periods
• current performance
• revenue changes
• and profitability impact.

This allows:

• CFOs
• finance teams
• and eCommerce managers
  to clearly evaluate whether dynamic pricing strategies improve overall business results.

Key Takeaways

The lecture demonstrated that:
AI-powered pricing is becoming a major driver of eCommerce growth
retailers can often increase margins without reducing order volume
small pricing adjustments across large catalogs can create major revenue gains
optimal discounts are frequently lower than merchants assume
traditional A/B testing is increasingly being replaced by multi-armed bandit models
real-time optimization during campaigns improves profitability
pricing strategies should optimize for profit per visitor, not only conversions
inventory-aware markdown optimization reduces unnecessary discounting
AI can dynamically react to inventory velocity and seasonal demand
personalized pricing can use behavioral, seasonal, and competitive data
competitor price matching is not always necessary
similar-product AI models enable pricing optimization at large scale
dynamic pricing requires strong reporting and transparency for business teams
successful pricing optimization combines AI automation with strategic business decision-making.