Joyn had €15M a year to spend on Samsung Smart TVs and no reliable way to tie that spend to paying subscribers. A server-to-server pipeline through Adjust closed the gap.
Role Architected the server-to-server event pipeline, the Samsung Ads attribution setup, and the revenue reporting on top
Marketing wanted to spend on Samsung TVs the way it spends on phones: bid by bid, tuned on real subscriptions.
Samsung Tizen TV was one of Joyn's largest untapped acquisition channels. The €15M budget only pays off if each campaign can be judged on downloads, streaming and paid sign-ups.
TVs have no browser cookies and no phone-style ad IDs. Samsung uses its own identifiers, so the usual tracking set-up has nothing to match on.
Hardware limits, a closed ad network, and revenue that arrives after the ad.
Embedding a tracking SDK in the TV app slows it down and adds upkeep. Startup speed and playback stability had to stay intact.
Samsung Ads is a self-attributing network: it matches installs to its own ads. To optimise bids it needs Joyn to send back what happened after the install.
Installs and first opens are easy to count. Paid subscriptions and retention come later and sit in a separate database from the ad data.
Four steps: identify the TV, feed Samsung, respect consent, then join every ad to the revenue it produced. The track turns purple where the data moves from collection to answering the ROI question.
At app launch the TV's own advertising ID and its ad-tracking opt-out flag are read and linked to the Joyn account. Events then travel server to server into Adjust, signed with a security token so spoofed requests are rejected.
Adjust S2S · Samsung TIFA / PSIDSamsung Ads was switched on as a self-attributing network inside Adjust. Installs, engagement and subscription events now flow to Samsung in real time, so its bidding and targeting learn from what Joyn actually cares about. A retry queue keeps events from being lost when the network drops.
SAN postbacks · Retry queueTV-specific moments such as first open, registration and 30 minutes of streaming were mapped to standard Adjust events. User consent choices travel with the data, so GDPR and Samsung's interest-based-ads opt-outs are respected automatically.
Event taxonomy · GDPR consentPostbacks, click and impression logs and conversion data are pulled out through Amazon Feeds and loaded into Snowflake. There they are joined to the subscription database on the Joyn account ID. Tableau dashboards show campaign ROI, trial-to-paid conversion and 30, 60 and 90-day retention for Samsung spend.
Snowflake · Tableau · Amazon FeedsMarketing launched the entire €15M CTV programme on Samsung screens, with each campaign judged on subscriptions and retention rather than installs alone.
Smart TV ad exposure is linked directly to trial starts, paid activations and retention.
Campaign-level ROI in Tableau and live feedback to Samsung let teams shift bids toward what pays back, bringing down CTV cost per customer.
Because collection lives in the backend, the Tizen app keeps its startup speed and playback stability.
Consent states are passed along with every event, so opt-outs are honoured without manual checks.
Let's talk
Tell me your biggest data or AI bottleneck and I'll send you how I'd approach it — the same way I worked this problem. No résumé dump, no pitch. Just a useful conversation.
Start the conversation →