About Novakid
Novakid teaches live, one-to-one English lessons to children in more than 50 countries. Its sales team is organised by region, from Turkey and Poland to Korea and the Gulf, and split between first sales after a trial lesson and repeat sales to families who already study.
Dashboards showed what happened, and sales still had to ask why
The repeat-sales team had dashboards, but nothing that told them when to upsell a family, who to call or what to offer. Repeat selling mostly ran on support macros for inbound chats.
The answers sat in a warehouse of more than 5,000 tables that only the data team could work. Every operational question, from campaign conversion to trial-to-purchase by country, joined their queue.
Results from different tools did not always agree. One referral measure came out at 1.7% in one tool and 6.91% in another, and nobody on the commercial side could tell which to believe.
“I just wanna know, like, who to upsell and what to upsell, basically.”
Sales Coordinator, Novakid
Novakid narrowed what the AI could see to tables it trusted
Before opening the warehouse to sales, Novakid asked for it to be cut down to well-described source tables, with most of the clutter removed. Each remaining table carries its business meaning, so a question about sales reaches the right data.
Novakid’s CTO and CRO then brought the commercial team onto it. Within two months it had gone from one technical user to the commercial team working in Claude every day.
Sales managers investigate their own numbers
Sales managers now take a question straight to the warehouse. Recent ones ask what happened to first sales yesterday, region by region, how one market’s September compares with last year, and how many students in the Gulf take five or more lessons a week.
Others go deeper into the funnel. One manager asked what share of families the phone team missed after the email, WhatsApp and voice-bot steps in the Polish market, and then asked the same question for repeat sales.
Sales ops forecast the month
A sales operations lead built a dashboard that reviews past periods and forecasts this month’s revenue per customer, and refined it through more than 25 versions.
The same lead built an accounts-receivable view by payment window and region, and rechecks any figure before it goes to leadership.
Every trial call starts with context
Before a call with a family who has finished a trial lesson, a salesperson sees talking points, a purchase probability and the objections to prepare for.
Call reviews compare reps’ conversations with those of the top performers, starting from 500 recorded conversations. For repeat sales, the aim is a list of who to call and what to offer, built from each family’s purchase history and lesson pattern.