Experiments: Fake door Test
Joii: LLM Monetisation
Joii primarily operates as a B2B2C business, partnering with insurers to offer veterinary consultations to their users. However, in a changing market with tighter margins and increased competition, we explored diversified revenue streams, particularly D2C monetisation.

We had an existing Vet-trained LLM, already built for another initiative, making it a low-cost, quick-to-market candidate for experimentation. This experiment tested whether users would be willing to pay for AI-generated veterinary advice via a paywall experience.
The Hypothesis
We believed that if users are receiving helpful, trusted advice from a vet-trained LLM, then introducing a paywall after several messages will reveal if there’s a willingness to pay for continued access, giving us a signal on the viability of a D2C model.
What we did
  • Built a lightweight, vibe coded front-end (2-week build time) to host the LLM
  • Targeted high intent users via Google search and Meta ads (e.g. "dog vomiting")
  • Allowed users to freely engage with the chat, then triggered a fake paywall after a fixed number of exchanges.
Measured:
  • Engagement before paywall
  • Reactions to paywall (scroll, hover, exit intent)
  • Clicks through
Results
After 2 weeks we saw...

  • Users engaged well with the LLM: high completion of message flows, especially for symptom-based queries
  • When shown the chat limit card 14% of users clicked through
  • Only 2% entered their email address
Iterations of wording and position didn’t materially affect results
Takeaways
  • Users clearly valued the chat, the interaction data suggested strong interest in AI-led triage
  • Value did not translate into D2C willingness to pay
  • Despite several iterations of the test, we saw no traction worth pursuing
  • The LLM was useful, but monetisation through consumer paywalls wasn’t viable
Previous case study
Pets at Home - Deliver from Store
Next Experiment
Pets at Home - Mini Basket
Made on
Tilda