KRY + Dot
KRY is one of Europe's leading digital healthcare platforms, offering digital and physical healthcare services across Sweden, France, the UK, and Norway.
€800K/yr
revenue opportunity identified from a single open-ended question
10 min
to identify, analyze, and size the opportunity
20+
hypotheses tested simultaneously across dimensions
“We could have cut the funnel by age, by gender, by cohort, by dozens of dimensions. We just never had the time to do it systematically. Now we can.”
Claire Bertrand
Data and Analytics Lead, KRY
The Problem
KRY is one of Europe's leading digital healthcare platforms. Beyond telemedicine, KRY operates physical healthcare facilities in Sweden, offering a broad range of in-person consultations alongside its digital services across Sweden, France, the UK, and Norway. Thousands of consultations daily, across four markets, physically and digitally.
Like most companies that have reached a strong product maturity, KRY needs to find less obvious opportunities for growth. Some of them consist in fine-tuning the growth engine, improving conversion at specific funnel steps, for specific patient segments, in specific markets.
The challenge: fine-tuning requires exhaustive exploration across every combination of dimensions. And that requires time no team has.
- The Combinatorial Problem
- Break the funnel down by age, gender, specialty, market, booking type, time of day. The number of combinations to test grows exponentially. Analyzing all of this manually takes time data teams don't usually have.
- The Resource Constraint
- KRY had more data points than ever, but a smaller, leaner team. Testing a single hypothesis manually took 2 to 4 hours. Systematically testing all of them was simply not feasible.
- The Realistic Alternative
- Without a scalable way to explore the data systematically, most of these growth opportunities would remain undiscovered, buried in combinations no one had the bandwidth to test.
Evolution
Scaling Hypothesis Testing Beyond Bandwidth Constraints
- Manual Exploration
- Prioritized by Bandwidth
Even with advanced analytics teams, manual hypothesis testing means prioritizing. The highest-value questions get answered; the long tail doesn't. - Dashboard-Driven
- Known Metrics Tracked
Revenue funnels mapped, KPIs monitored. Great for tracking what you already know to look for, but dashboards don't surface what you're not already measuring. - With Dot
- Systematic Exploration
The same analytical rigor, applied to all hypotheses in parallel, not just the ones that made it to the top of the queue. Unknown unknowns surfaced and sized in minutes.
The Setup
KRY already had clean, structured data in their warehouse and a well-defined analytical framework. The setup was about connecting Dot to what already existed, not building anything new.
- 01.
- Connect the Semantic LayerDot connected to KRY's LookML, the semantic layer behind Looker that already held the funnel data and consultation events.
- 02.
- Define Data SourcesSelected which Looker explorers would serve as data sources for the analysis.
- 03.
- Sync Context & DocumentationPulled metric definitions, business term glossary, and explorer documentation. All the context behind doing analytically sound work.
- 04.
- Provide the Analytical FrameworkFed Dot the revenue engine methodology, the team's analytical playbook for understanding what drives growth.
- 05.
- Ask the QuestionOne open-ended prompt: "Where are the growth opportunities for France?" Dot handled the rest.
The Discovery
Claire connected Dot to KRY's existing Looker semantic layer, synced the metric glossary and business term definitions, and provided the revenue engine framework her team had built.
Then she asked one open-ended question: "Where are the growth opportunities for France?"
Dot analyzed a full year of data. It tested roughly 20 different hypotheses across multiple dimensions (age, specialty, funnel step, market, booking type) and surfaced three recommendations, each sized in euros.
Findings
Dot tested ~20 hypotheses and surfaced the three with the highest impact. Only the signal, not the noise.
Elevated Cancellation Rates in a Specific Specialty
Dot flagged unusually high cancellation rates in one specialty. Rather than treating it as a conversion problem, Dot investigated the underlying booking patterns and confirmed the elevated rate was structural, not a leak. Useful context for the team, not urgent to fix.
Unexpected Growth in One Specialty
Dot highlighted unusually strong growth in one specialty. While not directly actionable, it raised important questions: what was driving it, and could the pattern be replicated across other specialties?
An Underserved Patient Segment Worth €800K/Year
A specific patient segment was converting at a significantly lower rate than others at one particular funnel step. They arrived at the booking flow but didn't find what they needed.
Dot dug deeper: among patients in this segment who selected the catch-all option, a large share turned out to need a service that simply wasn't listed. The symptom was missing from the flow entirely.
The tip of the iceberg: patients who selected "other" despite not finding their symptom. The hidden part: all the patients who simply dropped off without clicking anything.
Why It Worked
- Shared Language
- The revenue framework and metric glossary created a shared foundation. Dot didn't produce generic statistical findings. It reasoned using KRY's own growth model and business terminology.
- Transparent & Auditable
- Every recommendation backed by simple, inspectable data. Not a black box. Just good ideas based on clear datasets. Quickly convincing that there's no people-pleaser bias or number-tweaking.
- Business Semantics
- Dot understood that age isn't just another variable; it's fundamental to healthcare delivery. The semantic layer of the business made recommendations relevant and credible, not just statistically significant.
- Democratized Analysis
- Previously, exploring a hypothesis required Looker or SQL proficiency. Now anyone (ops, clinicians, PMs) can test an idea. The data team's formalized context becomes a shared asset, not trapped expertise.
The Core Shift
Before Dot, KRY's data team could test the highest-priority hypotheses. With Dot, they can test all of them, across funnel steps, dimensions, and segments, in the time it used to take to explore one. The problem was never the data or the framework. It was always a resource problem. Dot solved it.
"It's so great to think that someone with a different gut feeling, who had no tools to dig into that idea before, can now come up with their own business case. There's no barrier of how good you are at Looker. That's bringing more democracy to the process."
Claire Bertrand, Data and Analytics Lead, KRY
The Numbers
- €800,000/year
- Revenue opportunity identified from a single open-ended question
- ~10 min
- To identify, analyze, and size the opportunity
- 20+
- Hypotheses tested in parallel before surfacing the top 3
- ~1 month
- Total setup time from first connection to systematic exploration
- 100x faster
- Hypothesis testing compared to manual analysis
Get your AI Analyst now.
Start for Free