Past Sessions

• Event: MAU Vegas 25
• Date: Thursday, May 21, 2025
• Speakers
  • Omar Karim, Director, Brand and Video Demand Products, Amazon Ads
  • Ahmad Anvari, Sr. Manager, Performance Ads Product and Engineering, Amazon Ads
• Estimated read time: 6-7 minutes

Quick Read Summary

Modern advertising has a connectivity problem, not a creative problem.

Customers are moving across screens, platforms, and contexts faster than most measurement stacks can keep up, which is why relevance and incrementality feel harder to prove year after year.

Amazon Ads frames the solution as a three part system, supply that spans owned properties and the open internet, measurement that can connect touch points, and AI that turns fragmented signals into addressable opportunities.

For mobile growth leaders, the practical shift is to treat signal capture, identity constraints, and cross device measurement as product inputs to your media plan, not reporting chores that happen after spend.

The opportunity is scale with fewer blind spots, but only if you define what you are optimizing for, wire measurement early, and use automation in ways that preserve control over brand safety, frequency, and learning.

The new attention economy is multi screen, multitasking, and measurable only by design

The headline numbers in this session are less interesting as trivia, and more useful as a diagnosis. Omar Karim points to behavior that most marketers recognize instinctively, phones are in hand while streaming, and shopping decisions are happening in parallel with entertainment. He cites 210 million US adults using their phone while watching streaming TV, and 56 percent of Americans shopping while binge watching. He also cites a daily media consumption figure of 12.4 hours, with a prediction that we are approaching peak media consumption.

The implication is not simply that people are distracted. It is that customer intent is distributed across moments that do not share a single identifier, a single device, or a single channel owner. In that environment, brands compete for the same slices of attention, and measurement becomes the constraint that defines what you are willing to buy.

Karim layers in the core friction points, media fragmentation, more than 20 touch points in a purchase journey, and the widening gap between what marketers run and what audiences perceive as relevant. He cites that only 9 percent of viewers self report seeing a relevant ad daily, and that 62 percent of marketers struggle to measure marketing impact.

If you are a mobile growth leader, the takeaway is blunt. Any strategy that assumes a straight line path from impression to install to purchase is not just outdated, it is structurally incapable of proving value when the user journey is spread across screens.

Supply is not just reach, it is the environment where signals are created

Amazon’s pitch starts with supply, but not in the usual sense of more placements. Karim describes an ad supported content network built from Amazon owned and operated properties, including Prime Video, Fire TV, Twitch, and Alexa, then extended through third party publishers such as NBCUniversal, Paramount, Hearst, and Fox. He frames the goal as brand safe environments and “quality connections” across the journey, not reach for its own sake.

He also makes a bold reach claim, engagement of nearly 90 percent of the US advertising audience through authenticated Amazon Ads interactions and third party publishers. Whether you take that as a planning anchor or a directional signal, it supports a strategic point, reach matters less than the ability to connect what happens in different environments without breaking privacy expectations.

The useful way to interpret this is as an operating model:

• When you buy supply that is fragmented by design, you need measurement that is interoperable by design.
• When you buy supply that is brand safe and authenticated, you can shift budget from probabilistic guesswork to higher confidence learning loops.
• This is why supply, measurement, and AI are presented as one system. Each part is only as valuable as its ability to carry signal into the next.

Measurement is the backbone, wire it before you spend

Ahmad Anvari brings the conversation back to the day to day reality for app marketers. He lists three challenges that will feel familiar to anyone operating under modern privacy constraints, privacy regulation complexity, multi device attribution difficulty, and fragmented audience data that makes optimization inefficient.

His proposed remedy is measurement first planning, implemented through mobile measurement partner integrations, then used to set clear campaign objectives and optimize with Amazon DSP. He describes a simple flow, establish measurement using first party app signals, define objectives, access inventory sources, then use measurement and reporting to understand engagement, performance, and app events.

The specific operational detail that matters is the integration layer. Anvari names partners including AppsFlyer, Kochava, Singular, and Adjust, with expansion to Branch and Airbridge. He ties that integration to real time CPI and CPA measurement, predictive modeling, and reporting that continuously informs strategy.

If you want a practical decision rule, use this one, if a channel cannot pass consistent event level signals into your measurement stack, it should not be treated as a performance channel, even if it delivers short term volume.

A short checklist to stress test your measurement readiness:

• Confirm your primary conversion events, and ensure they are consistently defined across platforms
• Validate that your MMP integration is live before scaling spend, not after
• Decide where you need deterministic measurement, and where modeled measurement is acceptable
• Align your reporting cadence with how quickly you expect the system to learn, and how quickly you can act

AI is not an add on, it is how you reclaim non addressable opportunity

The AI section is strongest when it avoids hype and stays anchored to constraints. Anvari argues that Amazon is “not new to AI,” and frames the principle as using AI to deliver results rather than using AI for its own sake.

The more concrete claim is about addressability. Karim states that without AI, brands would miss about 65 percent of previously non addressable inventory and anonymous traffic due to lack of identifiers. He then describes using shopping, entertainment, and streaming signals to convert that anonymous traffic into media opportunities, plus unified predictive models for frequency controls across the open internet.

He attaches performance outcomes to that approach, 21 percent more reach on average with 26 percent savings on media investments. Separate from that, Anvari cites outcomes tied to MMP integrations, including a 40 percent drop in CPA, and a 5 percent increase in purchase rate when combining machine learning with MMP integrations.

For growth teams, the important nuance is control. Anvari describes maintaining control over key optimization levers such as brand safety, supported by granular reporting and aggregate reporting, while automation reduces day to day management and shifts attention back to optimization.

A useful mental model is to treat AI as a translation layer, it translates partial signals into usable predictions, but your strategy still defines what “usable” means.

Four product bets that show where Amazon thinks advertising is going

Anvari names four solutions that reveal Amazon’s roadmap logic.

Amazon Audiences

He describes standard catalog audiences with behavioral based segmentation, including in market and lifestyle segments, then predictive models that combine Amazon first party signals, advertiser data, and third party signals to predict likelihood of interest across verticals including shopping, entertainment, gaming, travel, and automotive.

Multitouch attribution

He notes a multitouch attribution product in beta, designed to show how multiple touch points contribute to conversions, quantify upper and mid funnel impact, and connect streaming TV campaigns to outcomes like sponsored product performance or app install campaigns.

Ad relevance without identifiers

He frames ad relevance as delivering relevant ads regardless of the presence of ad identifiers, maintaining meaningful connections through the day, adapting strategy as available signals evolve, and converting anonymous inventory into higher value opportunities.

Performance Plus and Brand Plus in Amazon DSP

These are positioned as AI based solutions that combine advertiser conversion signals with Amazon first party and third party signals to create the opportunities that matter most, while preserving controls like brand safety and reporting visibility.

Taken together, these bets suggest a coherent strategy, win the future by making cross context journeys measurable, then use AI to make them optimizable.

The Amazon canvas, where supply meets experience

Karim closes the loop with what he calls the Amazon canvas, a 360 degree experience centered around Prime Video IP, designed to drive engagement, conversation, and conversion. He cites examples including a Twitch “watch with me” activation, an Amazon.com landing page activation tied to entertainment content, and interest in future IP offerings, including Fallout season two.

He also introduces Twitch Shoutout Ads, where creators display a clickable banner and talk about the brand or product, citing click through rates “upwards of 14 percent.”

On Prime Video, he references interactive formats such as interactive video ads, interactive pause ads, first impression takeovers, and interactive carousels. He adds that interactive features can include Amazon store data such as pricing, deals, reviews, and Prime shipping status, and he describes use cases for both Amazon sellers and non Amazon sellers, including a travel brand promoting a rewards program and destinations.

Most notably, he describes contextual native ads powered by AI, where paused content triggers context matched creative that is generatively produced, then hints at testing placement across the Prime Video home screen, instream, pause, and end credits, connected contextually.

The strategic point is that experience design is becoming the unit of value, not the isolated impression. If you can carry intent from content to commerce without forcing a context switch, you are closer to what “meaningful connection” looks like in practice.

Conclusion

The argument in this session is that relevance at scale is no longer something you buy with better creative alone. It is something you engineer, by choosing supply that creates usable signals, measurement that connects touch points, and AI that turns fragmented, privacy constrained data into addressable opportunity.

For senior mobile growth leaders, the stakes are clear. The teams that win will be the ones that treat measurement as infrastructure, automation as leverage with guardrails, and customer attention as a cross screen journey that must be designed, not assumed.

Speaker bios

Omar Karim, Director, Brand and Video Demand Products, Amazon Ads.  Omar leads Brand and Video Demand Products at Amazon Ads, building immersive ad experiences for gaming and TV. He launched Sponsored TV and previously served as Chief Product Officer at Engine Gaming and Media, with 25 plus years across media, mobile, web, and video products.

Ahmad Anvari, Sr. Manager, Performance Ads Product and Engineering, Amazon Ads. Ahmad leads performance ads product and engineering at Amazon Ads, with 20 plus years across advertising platforms. He has held leadership roles at Uber, helped launch ads on Instagram, led multiple Facebook ads and analytics initiatives, and previously worked on ads products at Google and experimentation at Yahoo.