TV’s attention layer: Why discovery, not catalogue size, wins now
Partner content
Discovery is not a “nice-to-have” layer on top of TV content; it is increasingly the product, writes NAGRAVISION’s Tim Pearson.
TV doesn’t lose attention because it lacks content. It loses because choosing what to watch is often slower, clunkier, and more fatiguing than it is on social video.
That’s the uncomfortable comparison most TV platforms now face: not other broadcasters or streamers, but the infinite scroll. YouTube, Instagram, TikTok and the rest don’t just compete with TV for time; they’ve also re-trained audiences to expect frictionless “next best” decisions instantly, endlessly, and with minimal effort.
The paradox of abundance: more choice, worse outcomes
The industry has spent a decade celebrating abundance, with more services, bigger libraries, and broader genre coverage. But abundance has created a new UX tax: decision-making.
In this context, “searching” isn’t only typing into a search bar. It includes browsing rows, switching apps, scanning home screens, backing out of titles, and looping through menus before a viewer commits to pressing play.
Industry reporting citing Nielsen and Gracenote has found that viewers can spend 10+ minutes per session trying to decide what to watch, and that a meaningful number abandon sessions when discovery fails. Exact figures vary by market, platform mix, and session type, but the direction is consistent: the decision phase is getting longer.
Those minutes aren’t harmless browsing; they translate into session abandonment, platform switching, and higher churn risk when finding something good feels easier elsewhere.
This is the paradox: platforms can add more titles and still lose attention, not because catalogue growth is bad, but because unstructured abundance without decision support increases friction at the exact moment viewers are deciding whether to stay.
“Algorithm > content” (and why TV should take it personally)
Social platforms won because they built the world’s most aggressive attention machinery: recommendation feedback loops, fast context-building, and interfaces designed to collapse the decision cycle.
That advantage is showing up in broader consumption trends. Deloitte’s Digital Media Trends research has pointed to younger audiences gravitating toward social video and, crucially, to the effectiveness of algorithmic discovery as a driver of that behaviour.
The implication for TV is blunt: discovery is not a “nice-to-have” layer on top of content; it is increasingly the product.
Introducing the “attention layer”
The attention layer is the part of the experience that defines whether a viewer finds something to watch or leaves before playback even begins. This layer directly influences whether a session converts into viewing time.
It includes:
* Search and navigation (including voice and remote ergonomics)
* Recommendations and personalisation
* Editorial curation (how content is organised and packaged)
* Context awareness (who’s watching, when, for how long, mood/intent)
* Availability and entitlements (whether the content is available to the user in that moment)
For many operators, there’s an added structural challenge. Distribution on connected TVs is essential, but it often comes at the cost of reduced control over the user experience.
When home screen placement, navigation frameworks, and key viewer data sit entirely with the TV platform, operators have fewer levers to improve discovery over time or build repeatable habits that drive retention.
When the attention layer works well, users quickly find something relevant to watch with minimal effort. The experience feels intuitive and responsive.
When it doesn’t, users are forced to search, scroll, and guess. This creates friction, and friction leads to shorter sessions and increased app-switching.
In practical terms, if finding content feels like effort, users disengage (regardless of how strong the content offering is).
LLMs as a TV concierge: from “rows” to intent
Traditional recommendation systems have been strong at pattern-matching: “you watched X, so here is Y.”
But the next leap is intent: what the viewer wants right now, in a specific context.
That’s where LLM-assisted discovery becomes genuinely interesting, not as a gimmicky chat box, but as a way to transform the interface into an “answer engine” for video:
- “I’m all caught up on seasons of Master Chef. I want to watch a cooking competition that is similar”
- “A film like Heat, but lighter, and available on services I already have.”
- “What should I watch to understand the news story everyone’s referencing?”
- “Find me a series I can watch with my teenager that isn’t cringe.”
This “conversational narrowing” helps viewers articulate preferences they can’t easily express through menus and rows. It’s also a chance for TV providers to become a trusted source of TV-related knowledge: what it is, why it’s relevant, where it’s available, and what to watch next.
Guardrails: the hard part that determines whether this works
If LLMs are going to sit in the attention layer, TV platforms need to treat accuracy, safety, and trust as core product requirements, not PR talking points.
A few non-negotiables:
Recommendations must reflect real availability. Recommendations need to reflect what the user can actually watch, based on rights and entitlements. Nothing kills trust faster than recommending titles viewers can’t access.
Metadata integrity becomes strategic. Discovery depends on accurate catalogue, rights, and availability data.
RAG over vibes. LLMs should be grounded in platform-truth sources (catalogue + rights + user entitlements), not generic knowledge
Bias and editorial responsibility. A conversational system can accidentally over-amplify what’s merely popular, under-serve niche audiences, or create echo chambers. TV platforms need clear principles for diversity, prominence, and transparency.
Privacy boundaries. Intent systems are powerful because they’re contextual; that also makes them sensitive. Users need clarity and control.
TV is not search or social — errors here impact trust, rights, and revenue.
What success looks like: measure the attention layer like a business
If discovery is the battleground, platforms need to instrument it like one. Success metrics can be refreshingly practical:
- Reduced time-to-play (browse time before first content starts)
- Higher session starts per visit
- Higher completion and return rates
- Lower “abandon without play” sessions
- Improved retention and reduced churn
In other words: fewer wasted minutes, more started sessions, more satisfied viewing.
AI-ready UX foundations: where platforms should invest now
LLM-assisted discovery won’t compensate for weak foundations. The winners will be platforms that treat discovery as an integrated system in which search, recommendations, editorial curation, and conversational inputs reinforce one another.
High-level priorities that consistently matter:
Metadata enrichment and normalisation (to support accurate intent matching)
Unified catalogue views across live, VOD, apps, and bundles
Identity and entitlement awareness (so suggestions are actionable)
A consistent UX language across devices (especially connected TVs)
Closed-loop learning (measure what gets played, skipped, saved, abandoned)
This is also where “experience platforms” (rather than single-purpose streaming apps) become strategically relevant: the attention layer must work across content types, partners, and business models, not just within one catalogue.
The takeaway
What becomes clear is that larger content libraries alone are no longer enough. Social video created the attention war, not streaming. The streaming platforms that succeed will be the ones that reduce the effort required to decide what to watch.
That’s ultimately what the attention layer controls. And for operators, the question is not only how to improve it, but whether they retain enough control over the experience to shape it in the first place.
Catalogue size gets you in the game. The attention layer decides who stays.
Tim Pearson is vice president of global product marketing and strategic alliances at NAGRAVISION