Selkirk Sport Cuts 15 Hours of Manual Work with AI-Powered Recommendations

Challenges

Hand-picked recommendations with no testing capabilities limited scale and optimization.

Solutions

Implemented Maestra with its AI-powered product recommendations to deliver:• Always in stock and relevant product suggestions• Flexible A/B testing• Synced recommendations across website and emails

Integrated with:

Shopify

Results

15 hours saved weekly on manual curation

January 31, 2026

Result

15
hours saved weekly on manual product curation

Testimonial

Cory Wynhof

Site Merchandising Manager at Selkirk Sport

Profile

Before Maestra, I had to manually select products for each recommendation placement – individually, for every single product. There was no room for nuance, and a lot of our catalog never got proper recommendations. I knew better merchandising could add 10–20% more revenue, but the effort required made it impossible to scale.

Now it runs automatically, and the cascading effect of removing this from my plate has been huge. That extra 15 hours has allowed me to focus on other needs of the business – and that’s definitely already paying dividends. And while I’m focused on strategy, every page still shows relevant, in-stock products.

Selkirk's Recommendations Transformation

Manual recommendation	Maestra's AI-driven recommendations
❌ Consumed 15–20 hours a week that could be spent on strategy	✅ Set up once, runs automatically
❌ Required constant updates as inventory changed	✅ Dynamic updates as inventory and trends change
❌ No personalization based on customer behavior	✅ Real-time personalization using browsing data
❌ Made A/B testing practically impossible	✅ Built-in A/B testing to optimize performance
❌ Limited to website	✅ Same logic powers emails and on-site widgets

Product Recommendations Examples

Selkirk automated product recommendations across key website pages. On the PDP, customers see "Frequently Bought Together" and "Similar Products" suggestions, while on the checkout page they are shown recently viewed items.

Maestra powers the full recommendation logic, and Selkirk's website displays the products delivered by Maestra:

Maestra-powered Frequently Bought Together recommendations on Selkirk product page

Maestra-powered "Frequently Bought Together" recommendations on product pages

Product Recommendations A/B Testing

Before Maestra, testing different recommendation strategies wasn't an option — every product suggestion was selected manually, and there was no infrastructure to compare what works better. Now, the team can run A/B tests across any recommendation placement, swapping algorithms to see which drives more engagement.

For example, Selkirk is currently testing two approaches in the "You May Also Like" section on product pages: one powered by the Similar Products algorithm, the other by Best Sellers.