Adriana, chef and owner of a neighborhood restaurant with an online shop

I run a small restaurant. During the pandemic, we started selling meal kits — pre-portioned ingredients for people to cook at home. The kits saved the business. But selling online is a different skill from running a kitchen, and I struggled with the "recommended for you" section on our site. It kept suggesting random things. A customer ordering a vegan curry would see a recommendation for our steak kit. It was embarrassing.

I don't know how I found this book. Maybe an ad. But the chapter on catalogue architecture changed everything for me. The author explains, in plain language, that a recommendation engine is only as smart as the data you give it. If the only attribute you track is "meal kit," the system will suggest meal kits at random. If you track protein, spice level, dietary tags, cooking time, and season, the system can suggest things that actually make sense together.

I sat down with our web person and we added tags to every kit. Vegan, gluten-free, mild/medium/hot, quick (under 30 min), date-night, kid-friendly, seasonal. It took two afternoons. Now when someone buys a mild vegan curry, the system suggests a medium-spice vegan soup or a quick vegan stir-fry. The recommendations aren't perfect, but they're not embarrassing anymore. And our average order value went up a little — not double, but enough to notice.

The book isn't written for people like me. It's written for tech company people and consultants. But the basic ideas are clear, and the author takes the time to explain things without making you feel stupid. I didn't finish every chapter. I skipped the one about B2B sales altogether. But the parts I read were worth the price.