Pickx TV Ai buddy - Killing a feature before it killed our budget !

Context & Business Challenge

In a rapidly evolving media landscape, Proximus wanted to explore how artificial intelligence could enhance the TV user experience. The hypothesis: a conversational AI assistant — the Pickx AI Buddy — could help users discover relevant content faster, thus increasing engagement and retention on the Pickx platform.

But with TV usage patterns shifting and AI still facing adoption challenges, the real question was: do our customers really need this tool? And more importantly: should we build it now?

My role

I led the end-to-end validation track of the AI Buddy concept:

  • Briefed internal stakeholders and external research partners (CMI, Profacts)

  • Designed the research plan and user testing frameworks

  • Synthesized qualitative and quantitative insights into actionable business recommendations

  • Concluded the project with a decision framework that helped the company save over €200,000 in development costs

Methodology

We structured the project into 3 main phases:

1. Exploration & Concept Framing

  • Defined business objectives: improve user engagement, differentiate the Pickx experience, and promote inclusive navigation

  • Designed a full-feature AI buddy concept, supporting voice & text commands, content recommendations, recording suggestions, and seamless cross-platform integration.

2. User Validation Phase

We implemented a dual research approach:

  • Qualitative Interviews (Explorative) conducted via CMI to assess:

    • Expectations around AI-based recommendations

    • Preferred platforms for interaction (mobile vs. TV)

    • Definition of a "premium" TV experience

    • Key decision drivers for adopting new digital features

  • Quantitative Survey

    • Measured usage intent, friction levels, and concept evaluation against industry benchmarks

    • Explored segmentation: novelty seekers vs. product seekers, AI chatbot familiarity, and content discovery behavior

3. Impact Modeling & Business Recommendation

  • Benchmarked alternative solutions (like 3SS POC integration)

  • Built a decision matrix combining friction level x usage intent

  • Recommended not to proceed with standalone AI Buddy at this stage due to:

    • Low perceived value from users (appeal 12%, usage intent corrected at 6%)

    • Lack of strong content discovery friction (only 21% struggled)

    • High development cost vs. uncertain ROI

Key Insights

  • Users like recommendations, but don’t necessarily want a chatbot for it.

  • Simplicity and discoverability are valued more than AI-driven novelty.

  • Mobile experience matters — but only if deeply integrated with their routines.

  • AI is still met with skepticism, especially by older or less digitally confident audiences.

Outcome

Decision: Do not proceed with feature development (saved ~€200,000 in cost)

Outcome: Redirected resources toward integrating a simpler, lower-cost solution with better product-market fit (via 3SS)Pickx Buddy.

Added Value: Strengthened our UX research capabilities, clarified digital strategy priorities, and avoided a costly misalignment with real user needs.

What I learnd

  • The most innovative solution isn’t always the most impactful.

  • Grounding product decisions in evidence-based validation is the fastest way to save both money and reputation.

  • Sometimes, killing a feature is the best form of user-centricity.