How Bikes Online Uses REP AI to Improve Sales, CX, and Operational Efficiency

When Bikes Online first introduced AI into its business, the goal was straightforward: make it easier for customers to find answers without contacting support.

Like many growing ecommerce teams, they didn’t start with a vendor. They built it themselves.

It worked on the surface. But as the business scaled, so did the cracks.

What started as a promising automation project quickly became something harder to manage.

That’s when they started looking for a better way. 

Rafael, a Solutions Architect at the company, led the effort. His role spans integrations across platforms like Shopify, Google Analytics, and Klaviyo, as well as bringing REP AI into the business to automate key workflows.

Bikes Online isn’t a small operation anymore

What began 14 years ago in Australia—shipping bikes out of a garage—has grown into a global business with operations in both Australia and the United States. Over the past few years, the company scaled rapidly, growing from a team of around 20 employees to more than 80.

With that growth came increased complexity.

The company expanded beyond a small set of bike brands into a broader catalog that now includes accessories, helmets, and riding gear. At the same time, customer expectations grew alongside order volume, especially given the nature of the product.

Bikes often sell for $1,000 to $2,000 or more. Customers want to feel confident they’re choosing the right model, the right size, and the right setup before they commit.

That creates pressure on both the customer experience and the support team behind it.

To meet that demand, Bikes Online started building its own chatbot internally. The chatbot they originally built could access internal data, answer product questions, and handle a portion of customer queries.

But over time, the limitations with responses, formatting, and UX became clear. More importantly, accuracy became an issue.

As policies changed or product details evolved, the chatbot would sometimes surface outdated information. In other cases, it would hallucinate answers entirely.

“The objective of the chatbot is to answer customer questions accurately & immediately so they don’t feel the need to reach out to customer service… but because of its limitations, it was creating a high volume of general enquiry tickets,” Rafael explained.

That’s when the team decided to look for a third-party solution.

Why REP AI stood out

As Bikes Online evaluated alternatives, a few things immediately set REP AI apart.

The first was the experience itself.

Compared to what they had built internally, REP felt polished. The interface was intuitive, visually clean, and fit naturally into their website. 

The second was capability.

REP wasn’t just answering questions. It was guiding conversations, helping customers explore products, understand trade-offs, and make decisions based on their needs.

And the third was the backend.

Rafael explained it best:  “I’m a data guy, so I really loved their backend… it gives very comprehensive analytics from our website, customers, and orders.”

What started as a search for a better chatbot quickly became something broader: a tool that could influence not just support but also merchandising, marketing, and overall site performance.

How REP supports the buying process

One of the biggest differences between REP and previous solutions is how it supports the buying process.

Customers don’t always come to Bikes Online knowing exactly what they want. Many are beginners or casual riders trying to figure out what kind of bike fits their needs.

REP helps bridge that gap.

Instead of static answers, it asks questions about budget, riding style, and experience level, and uses that context to recommend products.

It also pulls from a wide range of internal sources, including product data, blogs, and policies, to provide detailed, relevant responses.

The result is a more guided, consultative experience.

Customers can:

• Ask about sizing based on height and measurements
• Compare options based on usage (trail riding vs. casual riding)
• Get reassurance around policies like returns and warranties

That matters, especially for high-value purchases.

“It gives them the confidence that we are a professional, trust-worthy business… and that they’re making the right choice,” shared Rafael.

While exact attribution across channels remains complex, the team consistently sees cases where REP answers questions, recommends products, and directly contributes to completed purchases.

“It doesn’t depend on your team spending hours every week to make it work… even with default settings, it brings results.”
- Rafael Campanhola, Solutions Architect at Bikes Online

Reducing support load one insight at a time

While sales impact is important, one of the most immediate wins came from something much simpler.

Pedals.

Some higher-end bikes sold by Bikes Online don’t come with pedals: a small detail that was technically listed on product pages, but easy to miss.

This led to a steady stream of frustrated customers asking the same question, like “Where are my pedals?”

In some cases, the team would send pedals to resolve the issue, adding unnecessary cost on top of support volume.

REP surfaced this pattern quickly. It identified that a disproportionate number of conversations—and support tickets—were tied to this exact confusion.

That insight led to two changes:

1. Making the information more prominent on product pages
2. Having the chatbot proactively communicate it during conversations

The impact was immediate. Rafael shared how the team “went from hundreds of these tickets a week to less than 10.”

It’s a simple example, but it highlights something bigger: the value of understanding what customers are actually struggling with, not what teams assume they understand.

The results

Across more than 20,000 interactions, the following figures are reported by REP AI:

• 8.4% of total revenue is assisted by REP AI
• 48% lift in AOV for orders assisted by REP AI
• 4% conversion rate from shoppers who were guided to a product page by the REP AI’s Sales Agent
• 92.53% of questions asked were able to be successfully answered by the AI chat

These successes prove that shoppers were able to easily find answers to their questions and build confidence in their purchase decision.

Looking ahead

As Bikes Online continues to grow, so will the role of AI in the business.

But the reason the team loves working with REP extends so far beyond just the technology. It’s also been about the team behind it.

From onboarding to ongoing support, REP has taken a hands-on approach to proactively check in, gather feedback, and make improvements based on real-world usage.

And increasingly, a system the team can rely on without needing to manage it day-to-day.