HomeDashboard

Ten questions, answered with public data.

An analytics command center. Every card is a real analysis with its question, source, period and query. I start with e-commerce and operations, then data, then the sales work that came first.

  • Public data, not my business
  • Data refreshed

    The analyses

      Full analysisE-commerce retail

      Which customers and products earn the money?

      A real UK online gift retailer, not my business. I use it as the template for every other analysis here: the question, the charts, something to move, the query, the caveat and what I would do next.

        Monthly net revenue

        November is the peak, both years.

        UCI Online Retail II · Dec 2009 to Dec 2011 · CC BY 4.0 · public data, not my business. The last month is partial: UCI lists the data as ending Dec 9, 2011.

        Customer cohort retention

        Share of each first-purchase cohort that bought again, by month.

        UCI Online Retail II · public data, not my business

        Move the assumption: trim the catalog

        Keep only the top products by revenue. See what the rest was worth.

        explorable

        78.5%of net revenue is kept with the top 20% of products

        The other 80% of products bring 21.5%.

        UCI Online Retail II · public data, not my business

        The query

        loading

        retail_orders

        The caveat

        Source and license

        What I would do next

        • Turn the Pareto and cohort views into a monthly report I refresh, not a one-off pull.
        • Look at what the one-time buyers have in common before I decide anything about them.
        • Go through the long tail of near-zero products one by one. A product with no sales is not always a bad product.

        On my own site I ask the same kind of question about pricing, search performance and conversion. See the case study.

        03Contact

        Want this kind of analysis on your own data?

        zacharybhenley@gmail.com Book a 30-minute call