Kora Fashion · 2026
Building Kora’s sales engine
Background
Kora Fashion is an early-stage B2B SaaS start-up founded by two former Meta employees. Daniel Wellington is one of its pilot customers. I joined as the first external hire.
The problem
Kora sells virtual try-on and visualisation that sits right on the product page of online stores. Every customer has to run Shopify, and the brands we look for sell their own products directly: clothing, shoes, eyewear, paint, wallpaper, flooring and wall art. I call leads with a colleague on fixed days each week, and I’m responsible for sales in Norway.
Most of our time didn’t go to the calls. It went to working out who to call. For every company we had to check whether the store ran Shopify, whether it fit our target group, who made the decisions and which number we could reach them on. That took around ten minutes each, and plenty of them turned out not to fit anyway.
What I built
The first tool is a lead engine. It’s written in Python and connected to 10 APIs, among them public company registers in six countries and a few contact services. It checks every store for Shopify in two rounds and only approves it when it finds at least two independent signs. Then it removes duplicates and companies outside our target group, and finds the decision makers, phone numbers, revenue and a factual line about each brand. It only uses open and legal sources, and it moves slowly enough not to put any load on the websites it visits. It runs overnight. One night it came back with 234 new paint and wallpaper brands from around the world.
The second is the sales platform, which is where the team works every calling day. It ranks companies by how well they fit us, looking at things like segment, size and whether they already use a competitor. Before each call it shows an opening written by AI, and afterwards it drafts the follow-up email. Every call, meeting and LinkedIn contact is logged against shared daily targets, with a LinkedIn pipeline from connection to message and a chart of our progress. It works in three languages, and we also use it to train new people on the sales team.
What changed
The pipeline now holds more than 1,400 qualified brands in 15 markets, and we estimate the tools have saved over 200 hours of manual research. I still work on both of them every week. Starting without a playbook taught me that you can write one by building it, and these tools are mine.
Next project All projects