How I Used Purchase Data to Build Smarter Product Bundles
Combining transaction data, SQL, Power BI and product expertise to create relevant Frequently Bought Together recommendations.
Products bought together are not necessarily products that belong together.
Basket data can reveal an opportunity. In a technical category, only product expertise can decide whether the recommendation is genuinely useful.

Can historical baskets identify the right accessories?
A plumbing installation often requires several complementary components. The objective was to turn those relationships into a useful product-page module.

6-outlet manifold
3/4" connection
Mounting brackets
3/4" compatible
End cap
Male 3/4"Use behavior to find candidates, then validate the installation logic before publishing.
From raw orders to a measurable customer experience.
Data identified the opportunity. Product expertise validated the recommendation. Tracking measured the impact.
Which products are actually purchased together?
For every order, SQL generated product pairs and aggregated their frequency across historical transactions.
- Manifold
- Mounting brackets
- End cap
- PTFE tape
Illustrative case-study data
Orders containing both products.
Share of Product A orders also containing B.
Significance across the order base.
Relationship strength beyond popularity.
Explore, prioritize, then decide.
The dashboard translates thousands of relationships into a review queue. It is inspired by the analysis workflow, not a copy of Power BI.
Data suggests. Product expertise decides.
Professional customers may buy many unrelated items in one order. Statistical proximity alone is not technical compatibility.
ManifoldSelected productExample: manifold + mounting brackets.
Added manually when order volume is too low.
Do not recommend despite the co-purchase signal.
A simple operational bridge to the product page.
Fréquemment achetés ensemble
The analysis becomes useful only when customers can understand and act on the recommendation.
Collecteur48,85 €
Paire de supports10,98 €
Bouchon3,97 €Launch was the beginning of the analysis.
Bundle interactions must be isolated from standard Add to Cart actions to understand adoption, engagement and attributed revenue.
Illustrative case-study data
of eligible page views generated an interaction.
Add to Carts originated from the module.
Tracked revenue associated with bundle-driven Add to Carts.
Tracked revenue is not automatically incremental revenue.
- Conversion Rate
- Average Order Value
- Revenue per User
- Complement attach rate
From data to a better buying experience.
Data identified the relationship. Product expertise made it relevant. Tracking measured its value.