41 experiments, one goal:customers who buy without a sales rep.
Insuma is Alicorp’s B2B e-commerce for bakeries, restaurants and small businesses in Peru. In 2025 I ran its experimentation programme end to end: I defined the hypotheses, designed the variants, shipped them through Insider and read the results.
- My role
- Product designer, AI & experimentation: hypotheses, variants, analysis
- Company
- Alicorp · Digital Transformation CoE
- Product
- insuma.com.pe · B2B e-commerce, web and mobile web
- Goal
- Digital independence: customers ordering without a rep
01 — In 30 seconds
The floating banner
Challenge. High-potential promotions got lost in the product feed, so key categories like cleaning supplies barely converted.
Hypothesis. A persistent, dismissible banner will catch attention without blocking navigation.
How we show an offer matters as much as the offer itself.
Visual urgency
Challenge. Users browsed Flash Promos but put off the decision, and carts were abandoned.
Hypothesis. A visible countdown triggers loss aversion and shortens the path to purchase.
Urgency unlocked purchases people were already planning, in the categories they needed.
A home for every business
Challenge. Insuma serves bakeries, restaurants and cleaning services. A one-size-fits-all home put flour next to bleach and felt irrelevant to everyone.
Hypothesis. Tailoring the home to each business type, with the right products and how-to content, will raise engagement and direct sales.
The design was right the first time. The audience wasn’t. Personalisation only works when it maps to real intent.
Insuma Stories
Challenge. Customers leaned on their sales rep because they didn’t know what the platform could do.
Hypothesis. A pattern they already use every day, stories, can deliver bite-sized education and build trust.
Education can be a commercial lever, not just support content.
How I ran the programme
The experiments served three goals: digital independence (10,000 customers buying online against a goal of 8,000), stronger commercial communication (20+ campaigns built on patterns that had already proven value) and de-risking the Unified Product, the platform customers were migrating to, with 5+ experiments that validated its hypotheses first. Most tests ran in Insider with a holdout control group; when a promotion had to reach everyone, we measured before and after.