Why recommendation engines aren’t just for Netflix

Published: 23/12/2019


Netflix now has over 139 million subscribers worldwide, so the chances are you're familiar with its recommendation engine—a combination of relevant content and the odd curveball thrown in to pique your curiosity. But have you thought about the difference a recommendation engine could make to your business? The team at Ballance Agri-Nutrients did and together we brought the idea to their frontline staff. 


Most recommendation engines follow two traditional methods.  To deliver a solution for Ballance Agri-Nutrients—a farmer-owned cooperative with annual revenue of more than $900Million—we needed to take the traditional methods and add a few curve balls of our own to solve some unique challenges like:

Most Recommendation engines follow two traditional methods

Because you watched…

Ballance have seized the opportunity in customers’ historical purchasing data to make product recommendations at the point of sale, freeing up their frontline staff to have more meaningful and efficient conversations with their customers. The primary goal is to target improvements in the team’s productivity, with current systems and processes taking up to 9% of the “day in the life of” a nutrient specialist.  As the recommendation engine produces results based on the tacit knowledge of the whole Ballance organisation, the hidden benefits have been to help speed up the process of inducting new starters and provide capability for new business models—like outbound sales calls. 

And of course, farmers are benefiting from ideas across the region.

"It saves me doing basic calculations and allows me to focus on adding value and accurate recommendations.” - Tom Aitken, Nutrient Specialist

An original production

The pathway to every production is unique and, even though the idea was simple, Ballance knew there would be challenges along the way that would mean changes to core applications and processes. To ensure there were no dramas, we followed five key principles:

1. Start with a Proof of Value. Even though the idea seemed sound from the outset, the initial short sharp proof the value won the hearts and minds of sponsors—particularly the executive and the frontline staff.

2. Focus on a Minimum Viable Product. This helped us make the concept real from day 1. By starting with the end result in mind, we were able to bring algorithms to life.

3. Use a multi-skilled team with representatives from the frontline. In this case, it was Ballance’s nutrient specialists and science extension team whose specific knowledge and feedback was invaluable.

4. Have mechanisms to systematically capture the feedback and then learn from it. This is how we continued to improve the end-product, using feedback to polish various iterations.

5. Build incrementally. The original idea of the recommendation engine was expanded to calculate the farm’s needs. Now that the end to end process is proven, the recommendation engine is now being fully integrated into the MyBallance digital platform.

Produce your own feature

Ballance took a data-led idea to deliver a solution that can make a real impact at the front line. And you could do that for your business too. More importantly improved recommendations allows Ballance to put data at the heart of helping the farming community to be more productive, profitable and sustainable.

The credits

If you’d like to get into more of the detail on the Ballance and Ignite partnership, you can find a full case study here. Alternatively reach out to us.