Andrew Melián

Helium 10 · 2023–24

Customer Intelligence

Mid‑market brands couldn’t justify our top plan. I turned a narrow data request into a purchase behavior analytics product and shipped the MVP in weeks.

Role
Lead Product Designer
Team
A PM, 5 engineers, a UI designer, data science, sales, and marketing
MoM engagement
+40%
Repeat visits
3.5×
Customer IntelligenceLTVLoyaltyPurchase behaviorCustomer journeySentimentDemographicsFiltersLifetime value$105.32$17.75$4.3624.25:1

The problem

Mid‑market brands needed more data to justify our top tier. Leadership wanted repeat-customer data fast, from a new team with limited engineering.

Our hypothesis: without paying for a separate tool, brands couldn’t tell whether brand-building was creating repeat customers.

How I led it

  • Chose a lean processWith pressure to show traction, few back-end engineers, and a real risk of building the wrong thing, I set up a lean process: test assumptions fast and validate the MVP before building it.
  • Reset expectationsPast interviews, surveys, and sales calls showed stakeholders were over-indexing on early findings. I used that evidence to reset what the first release should promise.
  • Reframed the workInterviews showed repeat purchases were just one question. Brands also wanted to know customer value, what else customers buy, and how they buy. We reframed the work from a repeat-customer dataset to customer intelligence.
  • Cut to an MVPBack-end work was the biggest lift, so the PM and I cut the experience to the most important views. We shipped in weeks, not months, and rolled out slowly to stress-test the data.

Key decisions

Proved feasibility myself

Data science was busy, so after confirming we had the data, I queried it in Snowflake and wrote Python scripts to combine it.

Tested with real data

We tested concepts with users’ own data, first in spreadsheets, then in a live prototype. They needed real numbers to judge it.

Data to insight to action

I grouped the metrics into LTV, loyalty, and cohort retention views that balance the big picture with the details. A UI designer took it to high fidelity with existing components.

Screens

Outcome

Mid‑market engagement grew 40% month over month, and repeat visits rose 3.5×. Customer Intelligence and our new APIs became the basis of our data-as-a-service strategy.

Links

Next case study

Get in touch