Who really pays the format tax? Inside FAST's metadata delivery problem

In Amagi's June 2026 AIRTIME webinar, From Metadata to Monetization: How AI closes the gap in media operations, FASTMaster Chief Analyst Gavin Bridge sat down with Amagi's Vijay Sankar to unpack the findings behind the AIRTIME Report June Edition. The conversation moved past the usual "FAST is growing" narrative into something more operational: what metadata actually costs media and streaming owners, and where AI genuinely helps versus where it just moves the problem around.

Here are the four moments from the panel that stood out:

• What the "format tax" is and why it costs more than creating metadata itself
• Why channel operators and aggregators absorb most of that cost — not content owners or platforms
• The exact line AI shouldn't cross when generating and delivering metadata
• How much of the industry says reformatting metadata is its biggest operational

Let's dive in.

What is format tax?

Bridge: So to me, when you ask what is a format tax — why is reformatting metadata maybe more painful than the cost of metadata itself — it's because the format tax starts with that. It's a recurring cost of taking your title's metadata and recutting it to satisfy every platform's bespoke spec. You know, we just mentioned that 80 — perhaps 80 fields for one platform. You can't just copy and paste that; put it into another platform. They each have their own requirements. There's no standardization, and you have to pay for that for every title, on every platform, and every refresh. We're sort of looking at things like field definitions, character limits, dimensions of the pictures, the thumbnails, aspect ratios — everything. There's no standardization.

You know, I think it was after World War II that there were standards applied to things like the size of a screw across the world, to sort of make things easier for everyone to do, and it sort of became like a universal set standard? Doesn't exist in streaming for this stuff, so it becomes quite frustrating if you're distributing out to multiple platforms and multiple places.

86% said reformatting is the biggest operational drag

Bridge: So, in our survey, 86% of people said that reformatting their metadata was the biggest operational drag. And where that gets interesting is that no one disagreed. They may have been neutral, or they may have agreed, but it's very rare to find something where no one says, I disagree with this. So we could tell that it's a real pain point for the industry there. And it's ironic, because metadata itself isn't expensive. You know, if you look at the results, 39% of people said that metadata costs too much for its value. That's actually only a minority. So the industry's effectively telling us: we're fine to create metadata, we're drowning in delivering the metadata.

READ: AIRTIME Report March 2026 — AI in media operations

Who absorbs the format tax?

Bridge: For the most part, everyone's absorbing it to a point, but I think that channel operators and aggregators are kind of getting squeezed the most by this. We could break it down into its constituent parts. Content owners will pay some part of this tax, but they're mostly exporting it through, unless they own the channel. They could create metadata for their content and include that when it's licensed out, but it might just be good enough, and someone else further downstream has to finish that job. So it's the end deliverer who really pays that cost.

I think platforms absorb it the least, and that's by design. They're writing the spec, they're not processing it, they're not being a true partner, I guess, if you want to put it through a lens. And they're not saying, give us the stuff, we'll format it for you — no, this is our standard, we know it's different to these guys, if you want to be here, you're going to have to do it. So they obviously have other costs, which I know they pass on, but they're not really taking that one in.

I think it's really the aggregators and channel operators who are absorbing it, because they can't push it. It's like, you either do it, and you have a channel, or you don't do it, and you don't have a channel. They can't force anyone to standardize it. They can't say to a TV OS that has channels, you need to be in line with these guys, because we've already got our channel there. It's all on them.

And we're seeing in the data — 64% of people said that they're basically being squeezed on both sides at once. So it's definitely, I think, primarily falling on the channel operators.

Where should AI autonomy stop?

Bridge: Anyone who just takes a completely hands-off approach and publishes without checking is going to probably run into issues. It's a good tool, but it's not perfect, and it will require some checks. So I'd say, let's say it does 95% of the job, and you have to fix 5% — that's a massive win. So I think right now, today, the AI runs the plumbing of making everything, and you are the person who checks the meaning.

Autonomy stops where a mistake could move from being a process error to a judgment error. So we want to make sure it hasn't pulled in the wrong description. There are shows that have multiple names, series that share the same name. We want to make sure it's getting the right stuff, and not just inventing stuff, as AI is prone to do.

But we definitely — you want it to run on the grind, the high-volume, deterministic, checkable work that AI is really good for — so reformatting, resizing artwork. You want to keep a human doing the subjective, the risky tone and vibe tags, a synopsis voice — anything that could be an ambiguous genre call.

I think you could probably argue that some of what some platforms have as kids' channels — no kids are actually watching a TV show from the late 80s or the early 90s. Very few children are going to watch that; they've got many more things in line with what they want to watch. Maybe that shouldn't be — that could be an ambiguous call for a person to make. Should this be under kids and family, or should it be under retro? Quite honestly, why wouldn't you put it under retro? You could get ads. But that's exactly where a machine couldn't do it, but a human could.

So I think what we want to do now is let AI own the pipes, own the gritty work, keep people on the things where you really need that spark of intuition. Anything that's regulated — some platforms have a watershed, some don't. You want to make sure that if this is being prepared for someone who has it, you need to clear that this movie isn't going to have swearing, explosions, or graphic violence on in the daytime. And for the ones that do allow it, you don't have to worry — but anything that could blow back, you definitely have to still retain the human element on."

Metadata isn't the expensive part of FAST operations — delivering it is. And that cost doesn't fall evenly: operators and aggregators carry most of it, largely because platforms have no incentive to standardize. AI can absorb the mechanical grind of reformatting, but the panel was clear that it can't absorb judgment calls without a human checking the output.

Want to explore more?

• Watch the full on-demand webinar here.
• Get all the insights into our metadata pulse survey and exclusive insights from Gavin Bridge in the June edition of Amagi AIRTIME Report.
• Discover how Amagi's team can help streamline your metadata and content distribution workflows.

FAQs:

What is metadata format tax?

Metadata format tax is the ongoing operational cost of reformatting the same metadata to meet the unique requirements of different streaming platforms. Because there is no universal metadata standard, media companies must repeatedly adapt fields, artwork, character limits, and other metadata elements for every platform they distribute to.

Why is metadata reformatting important?

Metadata reformatting is important because every streaming platform has its own specifications for how content information should be structured. Properly formatted metadata helps ensure content is accepted, discoverable, and monetized across FAST channels and streaming services.

Who is responsible for metadata in FAST?

Metadata responsibility is shared across the FAST ecosystem, but channel operators and aggregators typically handle the most work. While content owners provide metadata, operators often need to enrich, validate, and reformat it to meet each platform's requirements before distribution.

Can AI automate metadata management?

Yes, AI can automate many metadata management tasks, including generating descriptions, tags, genres, and reformatting metadata for different platforms. However, human oversight is still needed for editorial decisions, compliance, and quality control to ensure metadata is accurate and appropriate.