SilaBG Case Study: How 82,166 AI Conversations Generated EUR 25,000 in Additional Revenue

A customer came into SilaBG's online store looking for a product worth 26 BGN.

The AI Sales Agent started a conversation. Asked a few questions. Understood the actual need. Recommended related products. Addressed every concern.

The conversation lasted 37 minutes.

The customer left with a 200 BGN purchase.

No human support agent would have had that patience at 11pm on a Tuesday. And no chatbot would have made it past the first response.

That interaction is a single data point inside 82,166. Here's the full picture.

The Challenge

SilaBG is a large online store in Bulgaria operating in a competitive category with high traffic volume and significant customer service demand. Like most stores at their scale, they had:

• Conversion rates stuck in the 2-3% range despite strong traffic
• Customer service costs scaling with volume, not efficiency
• High cart abandonment with limited automated recovery
• After-hours traffic that wasn't being served well by existing tools
• Operational processes that required significant manual team involvement

They weren't a company with basic tools. They had already invested in their infrastructure. But they'd reached the ceiling of what conventional tools could deliver, and they knew it.

The question was whether AI could actually change those numbers - or just add another layer of complexity to manage.

The Solution

SilaBG deployed HighSky AI's SkyCommerce platform across three areas:

AI Sales Agent

Real-time customer consultation, running 24/7, trained on SilaBG's actual product catalog and customer history. Capable of handling complex queries, product comparisons, objection handling, and guiding customers through checkout directly in conversation.

AI Search Engine

Natural language product search replacing the standard keyword-based system. Customers could now type as they think - not as a search engine expects - and receive a small set of highly relevant results instead of hundreds of marginally relevant ones.

RPA Services

Automation of post-sale processes - order entry, invoicing, shipping labels, inventory updates - using bots that operate without requiring API integrations.

Implementation took several weeks of training on SilaBG's specific catalog, language, and customer patterns. Not a plug-and-play deployment - a real integration built to deliver real results.

The Results

The numbers below are from live data across the deployment period. None of these are projections.

The cost savings alone - EUR 20,000+ - cover the implementation and then some. The additional EUR 25,000 in revenue is purely additive.

But the numbers that matter most to SilaBG aren't the totals. They're the patterns:

The after-hours problem is solved. A third of all customer interactions now happen outside business hours. Previously, those customers either waited for an email response or left. Now they get immediate, competent assistance and many of them complete a purchase they otherwise wouldn't have.

Conversion on complex queries improved significantly. Customers who previously abandoned when they couldn't find what they needed - because the search was too rigid or the product catalog too large to navigate - are now completing purchases with AI assistance.

Operational costs are no longer scaling with volume. Order processing, invoicing, and shipping label generation run on autopilot. The team handles exceptions, not routine work.

The 37-Minute Conversation

It's worth spending a moment on this particular interaction because it illustrates something that numbers alone don't capture.

The customer came in with a clear, simple intent - buy one specific product for 26 BGN. Under any normal scenario, this would have been a fast, low-value interaction.

The AI Sales Agent did what a great human salesperson does: it asked a question. The answer revealed something about what the customer actually needed - not just what they initially searched for. The conversation that followed was a real consultation. The AI understood context, answered follow-up questions with specificity, offered comparisons that were genuinely relevant, and stayed engaged for 37 minutes while the customer worked through their decision.

At no point did the AI rush. At no point did it repeat "please contact our support team." It just... helped.

The outcome - a 200 BGN purchase instead of a 26 BGN one - wasn't a manipulation or a dark pattern. It was what happens when a customer gets the information they actually need to make the best choice for them. The AI didn't upsell. It consulted. The customer made a better decision as a result.

What This Means for Stores With Similar Profiles

SilaBG's profile - large catalog, high traffic, competitive market, ambition for growth - describes hundreds of stores across Bulgaria and Central and Eastern Europe.

The results at SilaBG aren't exceptional because of something unique about their business. They're the result of deploying AI that was actually trained for the job, in a platform designed around sales outcomes rather than support ticket reduction.

If your conversion rate is 2-3%, those numbers aren't locked in. They're the floor, not the ceiling.

If your customer service costs are scaling with order volume, that's solvable. It just requires the right automation architecture.

And if 30% of your traffic is showing up after your team goes home - that's money you're leaving on the table every single night.

About HighSky AI

HighSky AI builds SkyCommerce - the AI autopilot for large e-commerce stores. The platform combines AI Sales Agent, AI Search Engine, and RPA Services into one integrated system that increases revenue and reduces operational costs simultaneously.

We don't give guarantees for specific numbers. We guarantee visible, measurable change - backed by evidence like what you just read.

SilaBG is a Bulgarian e-commerce company. Results shown are from live deployment data and represent actual outcomes, not projections.