How Streaming Teams Use Nexa to Get Answers From Video, Ad, and App Data in Real Time

After every big premier or live event, streaming teams ask three questions: how did it go, why did quality drop, and what did it cost us in ad revenue? Answering them used to mean exporting several reports and joining them by hand, so the answers often arrived after the decisions they were meant to inform. 

Nexa, the AI analyst inside Conviva, does that joining for you. You ask a question the way you would ask a colleague, and it pulls from your video, ad, and app data in one pass to return an answer you can act on.

One event, start to finish

Here’s what a unified timeline in Nexa looks like before, during, and after a big premier or live event.

Before the show: the report is ready before the data

An Asian broadcaster's live-events team knew its executives would want a full readout the moment the event ended. Days ahead, before any data existed, the team had Nexa build the entire report. On event day, it filled itself in and went straight to the executives. A Spanish-language broadcaster did the same for a multi-week event, and its report grew from 16 broadcasts to 54 without being rebuilt.

During the broadcast: finding the cause while people are still watching

At a European broadcaster, a high-profile opening broadcast ran at a Streaming Performance Index (SPI) of 58.3, with high rebuffering on about 19% of minutes. With a CTO call coming up, the team asked Nexa to analyze the event across every platform and device. Nexa traced the drop to CDN latency on 4.57% of minutes and to 4xx errors from a single host, and ranked the viewer groups hit hardest. The team walked into the call with the cause in hand, a diagnosis that usually takes hours.

After the fact: what the event cost in ad revenue

This question used to take days, because it needs video and ad data side by side. A global sports streamer now asks it during every event. On one event day, Nexa found 528 thousand unfilled ad breaks out of 18.17 million attempts, with mid-roll accounting for 95% of the gap. Two causes could be fixed before the next event, worth $50,000 to $60,000 across a multi-week event series at typical CPMs. Closing the larger structural demand gap would bring the total toward $400,000 to $500,000.

A follow-up asked how many of the ads that played actually finished. Smart TVs, the team's highest-CPM inventory, completed ads at 71.81%, about 9 points below average, and Nexa named the specific creatives completing below 63%. That gave the team a short list of fixes on the screens where each ad is worth the most.

The week after: fixing what the show exposed

One streamer with more than 15 CDNs and per-country routing policies had Nexa build a single heatmap covering 565 million plays across 10 countries, 10 CDNs, and six quality metrics, giving it one view to base routing decisions on.

A video operations team used one reusable, scheduled prompt to validate a hotfix across CDN, ISP, device OS, and player version. It caught the video start failure rate doubling and traced it to a single device and ISP combination. A European public broadcaster used a scheduled prompt the same way to replace its manual 37-channel reporting cycle.

One conversation instead of three reports

Each of these outcomes comes from the same shift: one Nexa conversation spans video quality, ad delivery, and app experience, so teams stop stitching together exports from separate tools. Teams can paste in a brief of up to 10,000 characters, get back a finished PDF or HTML report, and send it on a schedule. Nexa also answers in the language it is asked in, and customers have used it in Arabic, Korean, Portuguese, and Spanish to reduce hours of translation and dropped context.

What's next: Nexa as an agent your agents can call

More teams are building their own AI agents for support, engineering, and operations, and those agents need answers about the viewing experience too. That's where Nexa is headed next. 

Through MCP, your own agent will be able to query Nexa directly, with no one logging into a dashboard. Raw data access alone still leaves your agent guessing which dimensions to check. Nexa already has the full video, ad, and app context on your sessions and knows how to work through a diagnosis, so your agent gets an answer back instead of a pile of data. Picture a viewer messaging support because their stream keeps buffering. Your support agent asks Nexa what happened on that viewer's sessions, Nexa traces it to the root cause, and the agent replies in the same chat with an explanation the viewer can actually understand. Your agent brings the context Conviva will never have, like tickets and runbooks, and Nexa brings the full-census evidence, for an answer neither side could produce alone.

Get started with agentic analytics

Nexa is available in Pulse today. To try this on your own data, start with one of these prompts:

Run an ad-inventory fill analysis for [date or event window]. Show total ad breaks, unfilled breaks, and unfill rate by ad position. Separate quickly fixable causes from structural demand gaps, rank causes by share of unfilled breaks, and estimate recoverable revenue at our typical CPM.

Validate my new app version [new version] against [old version] over [date range]. Check CDN, ISP, device OS, and player version, scan SPI, Plays, VSF, and CIRR for anomalies, drill into anything flagged, and give a verdict: safe to proceed or investigate first.

To add app data alongside video and ads, or to set up a walkthrough for your team, contact your Conviva account team or email support@conviva.com.