Bringing convenience and accessibility to pet care
Joii is a the UK’s leading veterinary telehealth provider, with over 500,000 users accessing registered vets through its 24/7 video consultation service.
To remain competitive and meet growing demand for frictionless care, the business needed to evolve beyond its video-first model.
To remain competitive and meet growing demand for frictionless care, the business needed to evolve beyond its video-first model.
- Contributions
- UI & UX (App & Web)
- Research & testing
- Design system
- Prototypes
- User flows
- Team
The team comprised:- 4 Engineers
- 1 Product Manager
- 1 Product Designer
- Outcomes
- 51% Increase in consultations
- 90% Conversion rate
- 68% MoM growth
The Problem
Video consultations had helped Joii scale early adoption, but growth was plateauing.
Increasing competition and user feedback signalled that the current model didn’t serve all customer needs, particularly those seeking quick, casual advice or asynchronous care.
Our challenge wasn’t just to design a new feature, it was to reframe the user journey to reduce friction, unlock growth, and future-proof the platform. This meant identifying new communication formats that aligned with user behaviour and business goals around consult growth, user engagement, and care accessibility.
Increasing competition and user feedback signalled that the current model didn’t serve all customer needs, particularly those seeking quick, casual advice or asynchronous care.
Our challenge wasn’t just to design a new feature, it was to reframe the user journey to reduce friction, unlock growth, and future-proof the platform. This meant identifying new communication formats that aligned with user behaviour and business goals around consult growth, user engagement, and care accessibility.
Understanding drop off and communication preferences
Objective: To understand why users were dropping out of the video queue and not completing their consultations
Approach: Early on, we suspected that video calls might not be the ideal solution for all users. Before running surveys, we analysed drop-off patterns and benchmarked against other telehealth services, where chat had proven successful. Our survey with 136 users then confirmed that video calls were creating engagement barriers. Survey questions focused on comfort levels, perceived effort, and alternative preferences.
Approach: Early on, we suspected that video calls might not be the ideal solution for all users. Before running surveys, we analysed drop-off patterns and benchmarked against other telehealth services, where chat had proven successful. Our survey with 136 users then confirmed that video calls were creating engagement barriers. Survey questions focused on comfort levels, perceived effort, and alternative preferences.
Key insights:
Impact: These findings confirmed that video calls created a barrier to adoption, supporting the need for a low-effort and accessible alternative.
- 81% of users didn't want to wait long to communicate with a vet
- 72% would prefer a quicker, less intrusive way to communicate with a vet
- 35% stated they didn't have their pet with them at the time of the video
Impact: These findings confirmed that video calls created a barrier to adoption, supporting the need for a low-effort and accessible alternative.
Framing the right solution with lean thinking
During our definition phase we began with the Lean UX canvas and identified with the hypothesis prioritisation canvas that a live chat fell within the high perceived value and high risk category. Myself and the PM worked with the engineers, vets and stakeholders to align on feasibility and operational impact. By balancing user needs with business constraints, we ensured that our solution was scalable, sustainable, and desirable for both users and the business.
We began with qualitative research which we relied on to give us some confidence that users would find a chat service helpful. This was in the form of a survey where 94% of participants said that a live chat would be a useful feature to access.
Mapping the first steps
User flow
I mapped out the desired user flow and worked closely with the developers to ensure the flow matched their mermaid chart which they had created during their SPIKE. Once the squad was in agreement with the flow I was able to proceed with wireframing, and prototyping.
Wireframes
I began to align the flow with wireframes to ensure we weren't missing any considerations or user stories.
How we proved chat was the right move
Building our own live chat at this point was still risky, however we had laid the foundations to know that we were heading in the right directions to solve our problem. The final steps were to begin testing.
We made a two phase plan of low effort development experiments to test 1) User engagement & uptake and 2) Chat quality.
We made a two phase plan of low effort development experiments to test 1) User engagement & uptake and 2) Chat quality.
Experiment 1 - False door test
Objective: To evaluate user interest in a live chat feature for vet advice while retaining the video call option
Hypothesis: Users seeking veterinary advice are more likely to engage with a live chat feature, than the current video call option, due to its convenience and lower perceived effort on the users' side.
Results & conclusion:
Adding a chat option significantly boosted engagement and consult completion, driving revenue. Based on these results, we proceeded to phase 2 with a low effort chat implementation using a plugin provider.
Hypothesis: Users seeking veterinary advice are more likely to engage with a live chat feature, than the current video call option, due to its convenience and lower perceived effort on the users' side.
Results & conclusion:
- 118% increase in total CTA clicks (Variant B)
- 35% of users in Variant B chose chat - 2X more than those selecting a video call
- 89% increase in completed consults (vs 63% in control)
Adding a chat option significantly boosted engagement and consult completion, driving revenue. Based on these results, we proceeded to phase 2 with a low effort chat implementation using a plugin provider.
Experiment 2 - Trial chat
Overview: Building on our previous A/B test, we set the two-CTA setup (video call + live chat) as our control (Variant A). In Variant B, we replaced the live chat CTA with "Chat to an AI Vet" to gauge user interest in AI-driven consultations as the company explored investing in an LLM for triage.
Objective: Understand how users and vets interact with chat, including common user concerns, chat duration, and vet capacity. This test focused on learning rather than validating AI chat.
Hypothesis: Giving users the option to live chat with a vet, alongside the video call option will drive additional consults.
Results & conclusion:
Explicitly mentioning an AI vet reduced engagement and completion rates for both CTAs. This insight guided our next steps in refining AI chat implementation as part of a future phase and informed our next hypothesis: AI should support, not replace, human consultation.
Objective: Understand how users and vets interact with chat, including common user concerns, chat duration, and vet capacity. This test focused on learning rather than validating AI chat.
Hypothesis: Giving users the option to live chat with a vet, alongside the video call option will drive additional consults.
Results & conclusion:
- CTR: Control variant outperformed Variant B (Chat CTA +5%, Call CTA +2%).
- Completion Rate: Higher in the control (Chat +8%, Call +6%).
- Insights: Users preferred human chat over AI, leading to higher engagement. Most chat inquiries were non-urgent, with an average chat time of 7 minutes, and vets managing 3-5 chats simultaneously.
Explicitly mentioning an AI vet reduced engagement and completion rates for both CTAs. This insight guided our next steps in refining AI chat implementation as part of a future phase and informed our next hypothesis: AI should support, not replace, human consultation.
Experiment conclusion
After relaying the results of both experiments to stakeholders we were all in agreement that we had validated a live chat enough to warrant the development time needed to build it. We were up against a small time frame of two months to fully design, build and ship the live chat.
What we shipped
Our data and user personas indicated that the majority of our users were on iOS and regularly used messaging platforms like Facebook Messenger and WhatsApp. To reduce cognitive load and improve adoption, I intentionally designed the chat interface to mirror familiar patterns from these everyday apps, ensuring the experience felt intuitive, approachable, and low-friction from the start.
Landing Screen
I received a brief to redo the landing screen which was also the entry point into the chat. The landing screen at the time was quite jarring and limiting, it was a screen which I inherited so it was great to update the UI. At the time we were in limbo with our branding, with some stakeholders wanting to hold off bringing the branding more into the UI because there were talks of rebranding. However after presenting the screen to the C-Suite, we were in agreement that we couldn't afford to stay with a limited screen when this was our landing screen, which should grab the user's attention and drive our main revenue stream.
Trade offs and constraints
With just two months until a code freeze, we faced a hard deadline and missing it would delay launch by six weeks. To accelerate delivery, I opted to use the provider’s chat component library rather than designing custom components from scratch. This decision enabled faster implementation while giving me time to focus on refining the rest of the interface and ensuring visual consistency across the product.
Given the constraints, we prioritised critical functionality such as real-time messaging and status updates, over non-essential features like vet reassignment logic and treatment plans. This allowed us to ship a stable MVP on time without compromising the core experience.
Given the constraints, we prioritised critical functionality such as real-time messaging and status updates, over non-essential features like vet reassignment logic and treatment plans. This allowed us to ship a stable MVP on time without compromising the core experience.
The results
51% increase in consultations
The barrier to entry into a consultation was completely lowered as we introduced a more accessible means of communication with a vet
90% conversion rate
Across our users with access to the live chat
68% MoM growth
This figure is predicted to grow as we roll out marketing initiatives
Next Steps
Post launch
With chat adoption validated, we focused on improving continuity. Our first problem to solve was to allow vets to switch between chat and video seamlessly - this meant that we could provide extra caution and care with our consultations and allow a better experience which left users feeling confident with our care.
What's next for Joii chat?
While chat was a key first step, the future of pet telehealth lies in AI-assisted triage, and asynchronous messaging. As Joii expands, there’s an opportunity to explore predictive vet care, using chat history and AI to preemptively address common pet health issues before they escalate. This positions Joii as an innovator in proactive pet healthcare, beyond reactive consultations.
Next case study
Pets at Home - Deliver from store
