#CES2025: Eight Key Takeaways

CES is an annual trade show organized by the Consumer Technology Association (CTA). Held each January at the Las Vegas Convention Center, the event features presentations of new products and technologies in the consumer electronics industry.

In addition to the trade show, CES also features over 300 conference sessions with over 1,100 experts offering insight into the latest trends. This year, more than 150,000 people were expected to attend, making it the largest tech industry event globally.

I have been attending CES for more than a decade, and it's one of my favorite conferences of the year. Here are my key takeaways from this year's event.

1. Cannes in the Desert

I have attending CES for more than 10 years, and until recent years it was mostly a trade show focusing on consumer electronics and gadgets, attracting plenty of geeks interested in finding out about, and playing with, the latest trends and toys. My company would typically take brand clients on show tours as well as host an annual innovation networking event in the Barbershop in the Cosmo.

In recent years, by contract, CES has evolved into more of an advertising / AdTech conference that runs alongside the trade show. That trend accelerated after COVID, and CES now features more than a 1,000 thought leaders and analysts who travel to Vegas in early January to attend the many sessions and seminars, learn about new trends, or simply hang out in the Chandelier Bar to mingle with the who's who of the advertising world.

Someone I know recently referred to CES as "Cannes in the desert," which is fitting. This past week, I ran into much of the same cast of characters I would expect to see at RampUp, Possible, or Cannes-Lions, and the online chatter in the run-up to the event was ubiquituous, so I guess you could say the transformation is complete.

2. ESG is Dead... Sustainability is Not

Advertising in general and AdTech in particular has come under a lot of fire in recent years for wasteful and unsustainable practices, and frankly much of it is deserved. According to AdTechRadar, bid duplication wastes up to $20b annually, and only 14% of bid requests are unique. Not only are publishers sending out the same bid to multiple SSPs, but many SSPs are also submitting the same bid multiple times to improve their chances of winning. This requires a lot of energy and needlessly increases the advertising industry's carbon footprint.

When you look across programmatic ecosystem, there are many other glaring examples of waste. Believe it or not, each ad impression generates roughly one gram of carbon waste, and the cumulative emissions produced by the ad industry rival that of the airline industry. The Association of National Advertisers put out a report recently that 15% of ad spend is wasted on MFA sites, said AdExchanger . Plus there’s anywhere from 10-20% non-human traffic. All of this takes storage and processing power and generates a huge carbon footprint with no real upside for brand marketers.

From eliminating MFA sites to reducing ad fraud, from paring back bid duplication to cutting out middle men, there is simply a lot of opportunity to do get the same or better results with less. As an industry, we can and should do better.

Keeping with this theme, on Tuesday morning I was interviewed by David Savage from Nash Squared in their studio on the #CES2025 convention center floor. We talked about sustainability in AdTech, innovation, privacy, and a bunch of other great topics. I’m excited to see the documentary when it comes out later this year and presented to Parliament in the UK.

3. MTA is Dead... Incrementality Testing in Alive

Incrementality was a word I must have heard at least two dozen times over the four days. This is because due to creeping signal loss and ever-more-stringent privacy laws, user-level multi-touch attribution (MTA) has been dead on arrival for quite some time. In its place, marketers have been testing out a variety of measurement techniques, for example cohort analytics, and Marketing Mix Modeling has been making a strong comeback with its top-down (as opposed to bottoms-up) approach.

Incrementally testing has become a hot because in an increasingly fractured media environment, brands grapple to understand the effectiveness of their advertising campaigns and avoid showing ads to consumers who are going to make a purchase anyway. Incrementality testing is a statistical measurement methodology that compares the behavior of users who were exposed to an ad campaign against those who were not. The goal of incrementally goal is to determine if the campaign drove incremental conversions or if those conversions would have happened regardless.

Because retail media environments are often complex with various influencing factors, accurately measuring incrementality is crucial for brands to optimize their ad strategies and allocate budgets effectively. Moreover, other marketing activities or market trends can influence sales, making it difficult to isolate the campaign's impact. Unlike traditional MTA and other attribution models, incrementality testing helps isolate the impact of a specific ads, allowing brands to understand if they are driving new customers or simply influencing existing purchase behavior.

4. Hyper-Personalization Gets Real

Personalization has always been an aspirational goal for marketers because it works. Research shows that personalization can also lift revenues by up to 15 percent and increase marketing ROI by up to 30 percent. But personalization is difficult, and?63% of digital marketing leaders?say they still struggle to achieve it.

Content has long been the missing link when it comes to making personalization happen—the proverbial rubber meeting the road—and besides bad data the main reason why brands struggle to personalize at scale. It's one thing to come up with a strategy to send out dynamic creative with dozens of variants of creative, but another to have a design team create them, push them through various workflow and approval processes, and load them into various MarTech / AdTech platforms for activation.

The rise of GenAI has changed everything with its ability to churn out new or revise existing creative. It’s estimated 46% of marketers deployed some sort of GenAI for content production by EOY 2024. Solutions like Adobe Firefly are now being used by brands to create new graphics, modify images, or produce text effects. Recent GenAI releases by Google Gemini, SORA, and DALL.E, are making it easier for anyone to create images, text, video, or music.

Turning back to personalization, "hyper-personalization" refers to the use of real-time data, AI, and machine learning to create highly personalized experiences for individuals. Unlike basic personalization, hyper-personalization not only works in real time, but goes deeper by incorporating numerous data points to deliver a truly 1:1 customer experiences using GenAI to develop and deploy new content on the fly.

Last Fall, Amazon Web Services (AWS) announced a new capability is powered by Amazon Bedrock to craft personalized messaging, content, and experiences in a single interface. The GenAI-based process follows is fueled by the unified customer profiles to drive 1:1 personalized engagement.

Benefits of Hyper-personalization:

• Improved customer engagement: Highly relevant content can lead to increased customer interest and interaction.?
• Higher conversion rates: By presenting targeted offers, brands can drive more sales.?
• Enhanced customer satisfaction: A more personalized experience can lead to better CX and higher customer loyalty

5. Amazon Selling AdTech

Everyone in advertising knows that Retail Media Networks (RMNs) are red hot. As of October 2024, there were around 220 RMNs worldwide, and this this number is constantly increasing. Retail media will make up one-fifth of worldwide digital ad spend this year, and with a growth of more than 21%, Retail Media is growing faster than almost any other form of advertising.

At this year's CES, Amazon Ads—the biggest RMN with 74.2% of retail media ad spend in the US—announced a new ad product called Amazon Retail Ad Service which enables retailers to use Amazon’s homegrown AdTech tools to sell product ads appearing on their ecommerce sites. "The idea is to pitch retailers the same performance-heavy ad technology and machine learning models that Amazon offers advertisers," explains AdWeek.

I think this product will gain a lot of traction with marketplaces, delivery apps, and retailers who don't see Amazon as a direct competitor. It's also possible some mid-cap retailers who currently work with Criteo or Citrus might decide to try it out.

Overall, this product aligns well with Amazon's strategy of letting retailers control which parts of their ad businesses use Amazon’s technology, most of which was designed in-house. This is the latest smart move by Amazon as they continue their strategy to become a key player the space.

6. CTV, CTV, CTV...

Television advertising is big business in a state of flux... and was everywhere at CES. This shouldn't be surprising because as viewing behaviors shift from linear to streaming, the way advertising inventory is bought, sold and measured is changing dramatically as the industry’s decades-old model is disrupted.

This year, advertisers in the US will spend more than $28 billion on Connected Television (CTV), growing 18.8% YoY, says a report. Advertising revenue is growing in lockstep with viewership. This past summer, according to Nielsen, CTV reached record usage with 40.3% market share in June 2024, trouncing cable and broadcast with 27.2 and 20.5%, respectively.

Moreover, as CTV has risen, linear has dipped: Between 2021 and 2024, linear ad spending had a negative 3.5% CAGR, and upfront TV buys had a negative 2.7% CAGR, reports eMarketer. Looking forward, overall TV ad spending is fairly healthy, projected to grow by over $12 billion between 2023 and 2027, but nearly all growth is projected to come from streaming, not linear.?

Recent news items include The Trade Desk’s (TTD) announcement of its upcoming CTV operating system, Ventura, to be launched in the second half of 2025. TTD, operates one of a leading demand-side platforms (DSP) that manages inventory for, and allows brands to do ad buys in, a variety of ad formats, with CTV making up nearly half of the company’s revenue.

TTD says it’s creating a CTV operating system to streamline the ad buying process for brands while simultaneously making it more profitable for TV manufacturers. As a side benefit, any first-party data captured by Ventura could presumably be used to help with ad targeting and measurement across the TTD’s DSP.

Another recent announcement was made by Mediaocean, the maker of the popular Prisma workflow software used by the vast majority of media agencies to support planning and management of TV media buys. Innovid, the company that Mediaocean acquired, offers software to create, deliver, measure, and optimize linear and CTV advertising. Innovid’s leading product, Harmony Direct, enables media buyers to execute PG deals without using a DSP or SSP.

7. Monetization for Publishers

As publishers continue their inexorable shift to digital, more and more organizations are leaning on programmatic ads to generate revenue for online ad slots their sales teams are unable to sell. Compared to advertising sold directly to brands, programmatic brings in far lower CPMs and lower margins. According to Outbrain.com, CPMs for programmatic ranges from $0.50 and $2, which is frankly low.

Moreover, margins are lower because, as eMarketer reports, demand-side platforms (DSPs) and supply-side platforms (SSPs) each take about a 10% cut of ad spend, and collectively, receive about 20% of the money advertisers send to publishers when they purchase inventory. Granted, direct PMP or PG deals can have higher CPMs, but given these troubling dynamics it's not surprising publishers are seeking new ways to monetize their most important assets: 1P Data.

As the CTV market grows and matures, it’s becoming clear that publishers have a powerful asset they can monetize: first-party data. While publishers are not allowed to send data containing PII in bidstream directly to DSPs, recent advances in Privacy-enhancing technologies (PETs) offers an interesting opportunity for publishers to boost monetization efforts using this value asset.

To use data with an SSP to create deal IDs, a publisher essentially defines specific targeting criteria based on their audience data within the platform, then generates a unique “deal ID” that encapsulates those targeting parameters, allowing buyers with access to that deal ID to bid on the relevant inventory through their DSP, creating a PMP where only pre-selected buyers can access certain ad spaces based on the provided data.

Leading publishers like Disney’s are already using a data clean room to anonymize identifiers for identity resolution, allowing them to match and analyze data from different sources. Disney uses its clean room combined with an audience graph to create synthetic user IDs, which are provided to VideoAmp to perform a data overlay to report on reach and frequency across Disney’s linear and streaming properties.

It’s easy to see a future in which a CTV data clean room is used to match a publisher’s first-party identity graph, containing viewing data or demographics, with a partner’s second-party data to target consumers without compromising privacy. The matched data creates a custom audience in the clean room, from which the advertiser can create targeted audience segments.

8. Agents Galore

Across the world of tech, nearly every vendor had some sort of AI Agent (or Agentic AI) offering to present at the conference. Everywhere you go, there was talk about how AI agents can driving innovation by automating complex workflows, enhancing decision-making, and delivering personalized experiences at scale.

Nvidia in particular had five important AI Agent announcements including a new set of Blueprints to help build AI Agents to analyze data, distill insights, reason, and act, along with a new video search and summarization Blueprint that integrates with the NVIDIA AI Enterprise software platform, presumably paving the way for the creation of AI Agents that perceive visuals.

While Agentic AI is interesting and clearly offers a lot of potential, the technology also raises a slew of ethical and societal questions. For example, should humans have an expectation for human interaction? Or assuming agentic AI is able to automate many mundane or rote tasks, what happens to people if when these entry-level jobs disappear?

Further, Perplexity recently caused a stir when they discussed their future vision in which brands advertise to AI agents instead of people. In this vision, AI agents, not humans, become the target audience for digital advertising. “The user never sees an ad,” exp0lained Aravind Srinivas, Perplexity's founder. “The different merchants are not competing for users’ attention; they’re competing for the agents’ attention.”

It's admittedly an interesting vision, but potentially calls into question (or compromises) the role of an "agent." Knowing one of the main benefits of AI is it does mundane tasks much quicker and more efficiently than humans can, why would an AI agent need to see ads to, say, plan a trip for me when it can locate, review, process, and organize virtually limitless information very quickly? Assuming the vendors are able to make their information publicly available for AIs to index, that’s all it should take.

But maybe or definition of advertising simply needs to change when it comes to AI agents, and publishers end up paying some kind of "fee" to have their content prioritized for review - not exactly an ad in the way the way humans view them.

Up next, RampUp in San Francisco!