Broadcast Media Asset Management: Build or Buy?

When is off the shelf genuinely the right call here?

For the plumbing, always. Storage handling, proxy generation, transcoding, search indexing and editorial integration are heavy engineering that Dalet, Vizrt Viz One, Avid MediaCentral and Tedial do every day. Rebuilding that to arrive at something you could have configured is the most expensive mistake available in this category, and it is more work than it looks from outside.

For the whole thing, buy if you are a production company or a single channel broadcaster with a few hundred terabytes, conventional delivery requirements and a manageable version count. Configuration in these products stretches considerably further than most buyers actually use. Before assuming you have hit a ceiling, take your three most awkward assets, a compliance edit with a described audio track and a partner specific runtime cut among them, and ask a vendor to model them live.

Buy, too, if your archive is not commercially active. An archive you pay to store and never sell does not justify a metadata programme. The build case in this category rests on exploitation: finding material, proving you can use it, and delivering it in the format a buyer accepts. Without a sales motive behind it, the honest recommendation is to spend less on software and more on deciding what to delete.

The one thing to insist on regardless is an export. Ask your vendor for a documented, tested metadata export before you renew, not when you are leaving. You are keeping records for material that will outlive several generations of software, and portability is the difference between an asset and a future migration problem.

When does a custom build actually pay off?

The build worth doing is a layer, and it pays off when the MAM has quietly become the system of record for meaning rather than for media. Two or more of the following make the case.

• Your house schema exceeds what the product will model, and the overflow lives in spreadsheets. Your genre taxonomy, production numbering, commissioning references, territory codes and compliance categories encode how your organisation thinks, and every change to them currently arrives as a professional services engagement.
• Version relationships are encoded in file names and you have had a delivery rejected because of it. One programme is a master, a textless version, a compliance edit, several audio configurations, subtitle assets in multiple revisions, territory specific cuts and a promo. A naming convention is not a model, and it decays the first time a freelancer joins for a busy fortnight.
• Rights information never reaches the editor making the usage decision at four in the afternoon. They cannot know an archive interview was cleared for original transmission only, because that fact lives in a contracts system the media tools were never given.
• You deliver to more than about five partners with distinct specifications, and rejections arrive by email three days later.
• You are migrating from a legacy catalogue and need provenance preserved rather than flattened.

From Digital Heroes delivery experience, the first release covering your house schema, the version model, unified search across storage tiers and a media operations interface runs $100,000 to $250,000 over 16 to 24 weeks. A full platform adding rights aware access, automated partner delivery with pre delivery validation, archive migration in waves and machine generated metadata runs $300,000 to $800,000 phased over 12 to 24 months.

How do they compare on the things that matter in this industry?

Schema ownership. Every product supports custom metadata and none of them knows your schema, because your schema is organisational rather than technical. The practical test is whether adding a field your sales team needs is a Tuesday afternoon or a purchase order. A layer you control maps outward to exchange standards such as EBUCore and PBCore when a partner needs them, rather than letting a vendor model decide what you are permitted to record.

Version modelling. This is the decision with the widest reach in the entire project and the usual reason a second attempt happens two years later. Ask any product to represent a compliance edit that is later re edited, with its own audio configuration and territory validity. If the answer is a folder of related files, delivery selection stays a judgement call rather than a query.

Rights at the point of use. Search results should carry usage status, and adding restricted material to a project should raise the restriction and name who to ask. Where the position is unknown the system must say unknown rather than implying clearance. That single design choice is small in code and large in consequence, and it is the capability broadcasters most often say they wish they had bought.

Delivery profiles. Each partner wants a specific wrapper, audio mapping, subtitle format, naming convention, metadata sidecar and checksum manifest. Holding each as a versioned profile with pre delivery validation turns a rejection into an exception caught in house, and a partner specification change into one profile update instead of retraining a team.

Restore behaviour. A request that triggers retrieval from a tape library or deep object storage needs queuing, batching, visible cost and an honest message that a clip arrives in four hours. Designs assuming instant storage do not fail with an error, they fail with a retrieval bill nobody attributed.

What does total cost of ownership look like at your scale?

A worked example. A broadcaster with roughly three petabytes across on premise storage, cloud object storage and a tape library, an existing MAM in production, eight partner delivery specifications and two legacy catalogues holding twenty years of records.

First release: discovery and schema workshops with archive, media operations, rights and sales $18,000, house schema with outward mapping to exchange standards $26,000, version model $34,000, unified search across three storage tiers with restore queuing and visible retrieval cost $32,000, integration with the existing MAM for proxies, identifiers and index $21,000, media operations interface $24,000, and migration wave one covering two collections and about 120,000 records $28,000. That totals $183,000. A broadcaster with one storage tier and no migration in release one lands nearer $110,000.

Adding rights aware access, all eight delivery profiles with validation, the remaining migration waves and machine generated metadata across commercially active collections takes total spend to roughly $520,000 to $680,000 over the following four to six quarters. Each partner profile after the first runs $5,000 to $14,000.

Running costs. Infrastructure for the layer is modest at $500 to $1,500 a month, because the heavy storage stays where it already is. Support and enhancement runs 12 to 18 percent of build cost a year. The lines that surprise people are deep archive retrieval and egress, which carry real per request and per gigabyte charges that nobody tracked before the system made them visible, and transcription compute, which should be modelled per hour of content rather than per asset.

Set that against the honest comparison. Take your MAM licence and support for a year and add the professional services line, because schema changes, partner onboarding and upgrades tend to arrive as services engagements and those are what a layer you control removes. Then add a year of redelivery fees, missed slots and operations hours. Then price the material you cannot sell because the rights position cannot be proved, which is inventory you pay to store and cannot monetise.

What does the hybrid look like, and when is it the honest answer?

The hybrid is the recommendation here, not a compromise. Keep the vendor for the plumbing. Stop treating it as the system of record for meaning. Your schema, your versions, your rights and your delivery profiles describe how the organisation makes money, and they belong somewhere you can change without raising a purchase order.

There is a narrower hybrid worth naming for organisations whose only visible wound is rejected deliveries. A packaging and validation service alone, holding each partner specification as a versioned profile and checking a package before it leaves, runs $60,000 to $120,000 over ten to sixteen weeks. It sits beside your existing systems, changes a rejection three days later into an exception caught in house, and does not touch your schema or your rights position. It stops the bleeding finance can see, which is often what funds the rest.

Two disciplines keep the hybrid honest. Take the first five partner profiles rather than all twelve, because establishing the packaging and validation framework is the messy work and each profile after that is a fraction of the first. And start machine metadata on collections with commercial demand rather than backfilling everything, storing machine output separately from human catalogued fields so nobody later confuses a guess with a fact.

The hybrid stops being honest when migration is the actual project. If twenty years of records in two legacy catalogues is what stands between you and exploitation, that is a workstream with its own budget and its own waves, not something a delivery layer absorbs.

Which should you choose, by operator size and stage?

Production company or single channel broadcaster, a few hundred terabytes. Buy and configure. Dalet, Viz One, Avid MediaCentral or Tedial. Test your three most awkward assets in a live demonstration before concluding you need more.

Multi channel broadcaster, delivery rejections are the pain, schema tolerable. Buy the platform, build the packaging and validation service at $60,000 to $120,000. Measure redelivery count before and after so the next phase argues for itself.

Around a petabyte and up, several storage tiers, rights in spreadsheets. Build the layer. First release at $110,000 to $183,000 depending on tiers and whether migration wave one is in scope. Get the version model on a whiteboard and reviewed by the people who deliver before any code is written. If media operations cannot recognise their world in it, redraw.

Rights holder or distributor monetising a deep archive. Phase to the full platform at $300,000 to $800,000, with rights aware access first, because unprovable rights are what stops material selling. Treat each migration wave as a priced increment with its own decision point, since the second wave teaches you what the first actually cost.

In every case, insist on a documented and tested metadata export before go live, and own the repository and infrastructure accounts in writing.

If you would rather scope this before committing budget, Digital Heroes builds and runs its own products, so the people choosing your architecture live with those decisions on their own revenue. You can take that specification to any other firm on your shortlist.