Interview: Tuned Global’s Con Raso Talks About Streaming Service Manipulation Detection
Tuned Global Managing Director and Co-Founder Con Raso reconnected with NxtNow Music for an in-depth interview to discuss one of the music industry’s most timely topics: Streaming Service Manipulation Detection (SMD).
After reading our Q&A with Con below, check out our earlier interview with this music tech innovator HERE and learn more about Streaming Service Manipulation Detection on the Tuned Global website.
What was the "lightbulb moment" when you realized the Streaming Service Manipulation Detection (SMD) solution needed to be a native part of the Tuned Global platform to make it accessible for rights holders and your streaming service clients?
The honest answer is that it was not a single moment – it was a gradual accumulation of signals from licensors, services and ensuring that the system was optimised to pay the artists based on real data. Rights holders are increasingly specific in what they expected platforms to demonstrate. They are not just asking ‘Do you have fraud detection?’. They asked: ‘Show us your thresholds’, ‘show us your audit trails,’ ‘show us how discounted plays feed back into royalty calculations’. It was clear that we needed a native solution.
What accelerated it for us was that we already had the data infrastructure to do this properly. We operate a data lake architecture – raw playlogs flowing into structured layers for analytics – and the signals needed for manipulation detection were largely already there: play duration, completion rates, device identifiers, IP addresses, geographic data, session patterns. The question was not whether we could build this, but it was why we had not formalised it sooner.
As manipulators start using AI to mimic human listening habits, how does Tuned Global’s "network effect"—having a bird's-eye view across so many different services—give you the edge in spotting these subtle patterns?
This is actually one of the most important structural advantages we have, and it becomes more significant as manipulation tactics grow more sophisticated. When you are operating detection within a single service, you are limited to the behavioural patterns visible within that service's user base. A coordinated manipulation campaign can be calibrated to look entirely plausible within that narrow window.
When you are aggregating anonymised, privacy-compliant signals across a broader set of services, coordinated patterns become much harder to hide. Subtle behaviours – the same device fingerprint appearing across services, IP clusters that show up in unusual configurations across tenants, play timing patterns that replicate too cleanly – these things become visible at the network level even when they are below the noise threshold for any individual service.
Critically, this does not just improve behavioural detection, it allows us to map risk at the content level. We can identify tracks or catalogues that are consistently associated with anomalous activity across services, even when that activity appears benign in isolation. That means we are not only spotting suspicious listening patterns, but also tracing where those patterns concentrate economically – where fraudulent behaviour is likely resulting in payments to the same underlying actor.
As AI-generated listening behaviour becomes more convincing at the individual account level, the cross-service view becomes the primary detection surface. Our roadmap explicitly accounts for this. We are building toward anomaly and predictive models that use aggregated, anonymised feature data across our client base. The more services participate, the more refined those models become (not just in identifying suspicious behaviour, but in proactively flagging high-risk content and revenue flows). It is not just a commercial benefit, rather a genuine technical advantage in the arms race against increasingly sophisticated manipulation.
Every artist wants fans who will listen to their new track 50 times a day or more. When Tuned Global monitors for "behavioural plausibility" and "repetitive listening," how do you ensure the system doesn't accidentally penalize the most passionate human fans while hunting for playback bursts?
This is a question we take seriously, because getting it wrong in either direction has real consequences. Penalising real fans undermines artist earnings just as surely as tolerating bot traffic does.
The key distinction in our framework is between detection and decisioning. Detection looks for patterns across multiple dimensions simultaneously: not just repetition, but the combination of repetition with invariability. A genuine fan who listens to a track 40 times in a day might still show natural variation – different skip rates, slightly different completion points, varied listening sessions across the day. Bots, by contrast, tend to exhibit an almost clockwork uniformity – identical play durations, consistent skip behaviour to the second, session cadences that do not vary regardless of time of day.
Where a daily play cap applies, the framework supports configurable thresholds – it operates as a decisioning layer, not a penalty. Plays beyond the threshold are moved to an exclusion ledger for royalty purposes, but the data is retained. This also means that in a review, we can examine what was discounted, why, and make a determination about whether the activity looks human. The framework is designed to be explainable and auditable precisely so that edge cases – the genuine superfan – can be reviewed rather than automatically dismissed.
Tuned Global is a leader in music tech for fitness, wellness and gaming brands. Does service manipulation look different in a workout app or an in-game radio station compared to a traditional streaming service? Also, was this framework built to handle those unique consumption "vibes"?
Absolutely, and this is something we were deliberate about. The consumption patterns in a fitness app or an in-game radio station are structurally different from a traditional on-demand streaming service, and a detection framework that does not account for that would generate a lot of noise.
In a workout context, for instance, high-repetition listening is entirely expected. Someone running three times a week might listen to the same playlist, with the same tracks, on roughly the same cadence, every week. Completion rates will be near 100% because the music is background to an activity, not the primary focus. Skip rates might be unusually low for the same reason. If you applied a generic streaming service heuristic to that behaviour, you would flag a large proportion of completely legitimate engagement.
Our framework accounts for this. Detection rules and thresholds are calibrated per service context, not applied as a single universal standard. What looks suspicious on a general-purpose streaming service may be entirely normal on a fitness or gaming platform, and the system is designed to reflect that. It also means rights holders receiving reporting from us are getting data that has been appropriately contextualised – not raw anomaly counts that do not consider the nature of the service.
You’ve noted that the music industry still lacks alignment on how manipulation is actually defined and enforced. By providing structured reporting to rights holders, would you say Tuned Global is setting the "gold standard" for transparency in future royalty reporting?
We are cautious about claiming a gold standard when we are also the ones pointing out that industry-wide alignment is still lacking. That said, structured, auditable, consistently applied reporting is clearly what the industry needs to move toward, and we think demonstrating what that looks like in practice is a meaningful contribution.
What we have built is a discounted plays ledger that functions as the single source of truth for royalty and chart exclusions. Every decision to discount a play references a specific rule version and a feature snapshot – so rights holders are not just receiving a number, they are receiving something they can understand, including monthly summaries breakdowns with reason codes and trends. That level of documented transparency is what licensing conversations are increasingly demanding.
The harder challenge is broader industry alignment on what manipulation is, how it is defined across different platform contexts, and what constitutes adequate enforcement. We see our structured approach as both meeting our obligations to rights holders and artists today and contributing to the kind of documented framework that a broader industry standard could eventually be built on. Whether that makes us the gold standard or just a useful reference point – I will leave that to the industry to decide.