The Hardest Part of One Streaming App Is the Data
Skydance says its streaming services will become one. The app is the part everyone will see. The decisions that last are underneath it: who the viewer is, what gets recorded, and where it lands.
Three streaming systems. One app. Five decisions that determine the data layer.
Teal is what the company owns and can take anywhere. Gray is what a vendor operates.
What’s Been Announced
The public record is short, and it is mostly about intent.
Mar 2, 2026
One service planned
David Ellison says Paramount+ and HBO Max will be combined once the Warner Bros. Discovery deal closes.
Jun 30, 2026
Pluto TV moves first
Its web app goes onto the unified platform. On August 4 the company said the rest would follow by the end of summer.
Oct 6, 2026
The deal closes
The combined company becomes Skydance and plans to merge its streaming products into a single service over time.
Next
Timing not announced
Ellison says one app will take time and points to a bundle first. Nor has the company said which platform it runs on, whether Pluto TV is part of it, or what happens to discovery+.
In June, Paramount began moving Paramount+, BET+ and Pluto TV onto one technology stack. The Next Web described the real purpose as a repeatable integration process for HBO Max.
This piece doesn’t guess at the app. It looks at what has to be decided underneath any version of it.
What One App Inherits
A single app inherits more than a catalog. It inherits years of data decisions, made by different teams for different businesses.
Several sign-in systems
Each service has its own accounts, profiles and device IDs. The single app needs one account key, and a way to bring history across without counting a household twice when it paid for more than one service.
More than one way of measuring viewing
The services were built by different companies and record plays, pauses and ads in different ways. Measures with the same name, such as starts, minutes watched and completion, are often defined differently.
Three business models
Subscription, subscription with ads, and free channels paid for by ads alone. Each needs its own events: entitlements and churn signals for subscribers, ad exposure for the ad tiers, channel tune-ins for free live TV.
Consent in many markets
HBO Max is sold in many countries, each with its own privacy rules and regulator. A choice a viewer made in one app has to hold when their data meets the other’s.
And one app may not mean one service. The Hollywood Reporter expects discovery+ to carry on as its own service, so the account key and consent rule need to reach the services that stay outside the app, not only those that move into it.
The app’s design will settle over a few releases. The data inheritance won’t settle on its own: whatever the data layer does first tends to become what everyone builds on.
Five Decisions That Lock In Early
Five decisions get made early, often by default, and are expensive to unwind.
| Number | Decision | What it settles | Lock-in moment |
|---|---|---|---|
| 1 | Account key | Which ID stands for a person across apps, devices and sign-ins, and how profiles and households hang off it | Locks inThe first release that signs people in to the single app |
| 2 | Consent rule | Where each viewer’s choices are checked, and how they travel with the data | Locks inBefore any one service’s data is combined with another’s |
| 3 | Event contract | The names, fields and meanings of every play, ad, page and purchase event, whichever app or device sends it | Locks inThe first reports and models built on combined data |
| 4 | Collector | The code in each app, site and TV build that sends those events, and where it sends them | Locks inThe first release that ships tracking for the combined catalog; TV apps update slowly |
| 5 | System of record | Where combined data lands, under whose schema, and who can query it | Locks inWhen the first combined dashboards and models go live |
The collector is the one most often decided by default. Whichever app the single service is built from brings its tracking with it, and every other service’s history has to be translated to match. Settle the event contract first, and the collector becomes a replaceable part instead of the decision that makes all the others.
What It Looks Like on Oracle
On Oracle, the data layer for one app would run in five steps. The parts that must stay portable sit outside any vendor.
Every source writes to one contract the company owns. Oracle runs everything from collection to activation, and could be replaced without re-instrumenting the apps.
- Collect. Oracle Infinity takes app, web and TV events: from tags on the web, and as server-side copies from apps and TVs.
- Unify. Oracle Unity resolves the account key into profiles and households.
- Land. OCI Object Storage keeps raw events and exports in the company’s own storage.
- Store and learn. Oracle AI Database and OCI’s AI infrastructure keep models next to the data they learn from.
- Activate. Unity audiences flow to Responsys, Eloqua and ad partners.
Oracle’s case is that database, cloud infrastructure and customer applications come from one company, so models can sit next to the data without an export step. What it still has to prove is just as concrete: that Infinity takes the event volume of three large streaming services in near real time, that its viewing measures reconcile with the numbers advertisers and finance already trust, and that analysts find its day-to-day tools as capable as the ones they use now.
Getting There Without a Big Bang
The route is the one Part 4 lays out for a single site, applied one app at a time.
Agree the contract first
Write down the event contract and the account key before the single app’s first build, and map each service’s current events onto them.
Mirror
Send each app’s events to the new destination alongside its current measurement: through the tag manager on the web, and as a server-side copy for apps and TVs, so that, where an app’s configuration allows it, mirroring starts without a new release.
Prove parity on business measures
Compare starts, minutes watched and ad impressions per title per day, within a tolerance agreed up front. Raw request counts won’t match, and shouldn’t be expected to.
Move each app once
Cut over an app when its numbers hold. Bring the next service onto the single app with the contract already in place, so its data moves once, not twice.
That is slower than a cutover weekend. It is much cheaper than a year of reconciling two sets of numbers that both claim to be right.
Portable by design, Oracle by choice.
Two Guardrails
Portability
Own what you instrument.
The company should own its event contract, its account key and its warehouse schema, kept in its own repositories and documented outside any vendor’s tools. Then the pipes can change vendors without re-instrumenting every app. That’s the lesson running through this series: lock-in lives in what you instrument, not in the contract you sign.
Viewer privacy
Enforce consent once, before data combines.
Viewing records tied to a person are among the most regulated data a media company holds. In the US, the Video Privacy Protection Act creates liability for a video provider that knowingly discloses information identifying a person as having requested or obtained specific videos, with narrow exceptions (18 U.S.C. § 2710). Outside the US, each market HBO Max serves brings its own rules, GDPR among them. Choosing a cloud vendor doesn’t change those duties. What the data layer decides is where consent is enforced: in one place, before one service’s data meets another’s, and recorded with every event.
Where to Start
Before the single app’s first build, I’d want three things on paper: the account key, the event contract and the consent rule.
Write these before the first build
Account key
Who the viewer is, across every app and device
Event contract
What gets recorded, and what each field means
Consent rule
What may be used, where, and how the choice travels
They are short documents with long consequences, and everything else in this piece can follow from them, on Oracle or anywhere else.
Part 3 covers the infrastructure move, and Part 4 the tag that carries existing instrumentation across. If you’re working through a platform merger of your own, I’d like to compare notes: get in touch.
Sources
- Axios: Paramount to combine HBO Max and Paramount+, March 2, 2026
- The Next Web: Paramount is merging Pluto TV, BET+ and Paramount+ onto one tech stack, June 10, 2026
- GuruFocus: Paramount Skydance Q2 2026 earnings call highlights, August 4, 2026
- TechCrunch: Paramount closes historic Warner Bros. merger to form Skydance, October 6, 2026
- The Hollywood Reporter: What Happens to Discovery+ Under Skydance?, October 7, 2026
- The Next Web: Paramount closes Warner Bros deal as Skydance and plans a single streamer, October 7, 2026
- 18 U.S.C. § 2710, Video Privacy Protection Act (Cornell Legal Information Institute)
Start of the series
Oracle CX: The Next Platform War
Where Oracle can take share from Adobe in enterprise CX.
The series
Oracle CX: The Next Platform War
Where Oracle can take share from Adobe in enterprise CX.
Infrastructure-Native Behavioral Intelligence
Infrastructure as the behavioral record, with the client tag as its calibration.
Full Infrastructure Migration to Oracle Cloud
A 24-month plan for moving a streaming technology estate onto OCI.
Lock-In Lives in Your Tags
One Tealium tag that speaks Adobe and sends it to Oracle CX.
The Hardest Part of One Streaming App Is the Data
Five data decisions under a single streaming service.
You're reading this
Evaluating your CX or CDP stack?
Get vendor-neutral guidance
We advise across 15+ CX and CDP platforms, including Oracle, Adobe and Salesforce, backed by a directory of nearly 7,000 integrations.
Independent · Architecture-led · No platform affiliation
Disclosure: CdpHub is independent and is not affiliated with, endorsed by or sponsored by Skydance, Oracle, Adobe or Tealium. This piece draws on public sources only. It describes a possible architecture, not the company’s plans. Product names are trademarks of their owners.
