Why Streaming Platforms Can Recommend the Perfect Film (But Game Stores Still Can’t)
Aleksander Kjeserud
Product Lead - Metadata & Taxonomy
@Gameopedia
There’s a reason Netflix, Disney+, or HBO can recommend something you’ll actually enjoy on a Tuesday night when you're tired, vaguely sad, and don’t want to think too hard.
It’s not magic. It’s not even a particularly clever algorithm.
It’s metadata.
The film and TV industry, especially streaming platforms, has spent years building a structured understanding of:
What content is
How it feels
Who it’s for
Games, despite being a larger industry by revenue, are still working with something far less structured.
Look closely, and that gap is actually a moat. The platforms that close it first will win on:
Conversion
Retention
How Streaming Learned to Describe Content
In the early days of streaming, recommendations weren’t exactly great.
Imagine watching Back to the Future and being recommended 2001: A Space Odyssey next. They’re both technically science fiction, but require completely different emotional states to appreciate.
The problem wasn’t the algorithm. It was the metadata.
Netflix became one of the most prominent examples of a platform investing heavily in human-led metadata tagging, an approach that is widely documented.
Using a 36-page training guide, paid taggers scored every title across dimensions like:
Romance levels
Narrative complexity
Emotional tone
This eventually generated nearly 77,000 unique micro-genre descriptors, including:
“Dark Suspenseful Gangster Dramas”
“Cerebral French Art House Movies”
Beyond tone and vibe, streaming metadata also captures:
Ratings (content descriptors and age guidance)
Runtime (time commitment)
Critic vs. audience scores
Awards and accolades
👉 The result: a structured answer to “What is this actually like to watch, and who is it for?”
What Game Metadata Looks Like Today
Open a typical game storefront and you’ll see:
Genre and subgenre
Community tags
User reviews
Price and screenshots
The structural problem isn’t that storefronts are doing nothing.
It’s that the current approach has a low ceiling.
Community-sourced tags are inconsistent
There’s no standard taxonomy
Interpretations vary across players
The same game can be tagged “hard” by one player and “casual” by another — not because either is wrong, but because there’s no shared framework.
What’s missing from game metadata:
Emotional context
Session expectations
Player intent and motivation
Ratings systems like ESRB or PEGI focus on content warnings, not experience.
They tell you what’s in a game — not what it feels like to play it.
The Biggest Gaps in Game Metadata (and What They Cost)
These gaps aren’t abstract. They create measurable problems across the entire discovery and commerce funnel.
1. Search Breaks When Metadata Is Thin
When a user searches for “cozy games,” keyword-heavy systems struggle to interpret intent.
Missing translations
Inconsistent labels
Poor query matching
👉 Result: users drop off or find irrelevant results.
2. Similarity Fails Without Deeper Context
Recommendation systems rely on shallow signals like genre.
Example: “Action”
One game → chaotic and silly
Another → slow and tactical
If metadata treats them as identical, recommendations will too.
👉 Result: “similar games” that feel nothing alike.
3. Emotional Tone Goes Uncaptured
Streaming answers:
Is this relaxing?
Is this intense?
Game metadata rarely does.
👉 Result: players can’t choose based on how they want to feel.
4. Session Length and Pacing Are Invisible
Films:
Runtime is explicit
Games:
No equivalent
Players can’t answer:
“Can I play this for 20 minutes?”
👉 Result: hesitation and abandoned sessions.
5. Product Pages Stay Thin at Scale
Without structured metadata:
Pages compete on price
Content becomes inconsistent
Information doesn’t scale
👉 Result: weaker conversion and unclear value.
6. AI Is Only as Good as the Data Behind It
AI systems rely on structured data.
Without it:
Recommendations degrade
Assistants give weak answers
Personalization fails
👉 Result: expensive AI, poor outcomes.
Why Streaming Got There First
Time
Film and TV have had decades to mature.
Structure
Films are:
Fixed
Linear
Consistent
Games are:
Interactive
Variable
Player-driven
That makes them harder to describe — but not impossible.
What Better Game Metadata Looks Like
A modern game metadata framework should answer:
👉 “What is this game actually like to play?”
That requires structured data across:
Emotional tone and vibes
Session profile
Pacing
Experiential complexity
This doesn’t replace genre.
It makes genre actually useful.
The Opportunity Ahead
This gap isn’t a small inefficiency.
It’s a structural weakness affecting:
Platforms → lost conversions
Developers → lost visibility
Players → frustration
Streaming proved the model works.
It just hasn’t been applied to games at scale.
The platforms that solve this won’t win because of better algorithms.
They’ll win because they finally gave those algorithms something worth working with.
