Curated Guide to Research
MediaPost's comprehensive guide to current research in the marketplace.
Search results for "media buying"
- Google scrapes profit from advertisers through heavy use of hidden search terms, as advertisers continue to see a decline in search query data.
- The more an AI model thinks, the more carbon it emits, a research study finds, and reasoning models can produce 50 times more emissions than concise ones. Large language models use tokens - words or parts of words - to convert them into a string of numbers processed by the LLM to come up with an answer.
- Google has added general availability for 2.5 Flash and Pro and is bringing custom versions into Search. Features such as reasoning and calculating tasks and costs will be important to marketers and developers using these platforms.
- Gannett became the first publisher on Wednesday to use Taboola's DeepDive generative AI search engine, but the tool seems to be missing several important features.
- Apple hosts its Worldwide Developers Conference next week. For advertisers this means gaining a better understanding of how desktop and mobile referrals are split and what this means for their media-buying strategies. For publishers it means that while AI search referrals are growing, mobile - the dominant traffic source - remains untapped. Mobile product share belongs to Apple, but Google dominates mobile web search.
- The official start of the fall television season is still about four months away. But for insiders at media companies, agencies, and advertisers (as well as television analysts like me), the upfront season has placed thoughts of September squarely into May and June. Of course, in today's video world, the very notion of a fall TV season is antiquated. Only the broadcast networks still adhere to seasons that begin in September and end in April or May (and promote "fall finales" like they're special episodes). Cable series tend to premiere throughout the year, and streaming platforms debut new shows and new seasons whenever they happen to be ready, which is not always in the same month as the previous season (and viewers often have to wait more than a year for the next installment).
- OpenAI's latest ChatGPT model disobeyed human instructions to shut down during a test by researchers, and I wondered "what if" AI refused to alter ad campaigns based on data the models believed could produce better results.
- As we head into upfront season, there's a reason you're hearing the term "performance media" now more than ever. Data - and the collection of it - has never been more robust.
- An analyst has made two predictions related to AI - that Google AI Mode will replace traditional search, and Apple will continue to use Google's engine as its default.
- One year after Google released AI Overviews, BrightEdge published data showing total search impressions rose more than 49%, highlighting a dramatic shift in how people seek and find information.
- AI, online ad serving, media buying, and search require a substantial amount of energy that companies have been trying to figure out how to support. Google's agreement with nuclear developer Elementl Power will help to develop three project sites for advanced reactors.
- Reference points line up everything. It is impossible to assess our situations in the world in the absolute. We need reference points: something against which we can compare our situations to see if they're good or bad, worse or better, proper or improper.
- Analysts have begun lowering ecommerce and paid-media industry forecasts as tariff policies persist and consumer sentiment continues to fall. U.S. ecommerce growth for 2025 is now projected at 4.5%, down from 7.5% -- and globally, excluding China, dropping to 6.5% from 9%.
- A survey by iSpot finds 16% of advertisers believe there will be declines in their budgets (1% to 49%), and 27% will see gains (1% to 49%) - while just 1% see their budgets climbing by 50% or more.
- Generative AI digital customer experiences are promising, but hallucinations remain a challenge with 49% of customers in a survey experiencing AI hallucinations leading to widespread skepticism about the tech.
- New research shows advanced AI models, including popular LLM agents, routinely lie in pursuit of their goals, raising new concerns about the reliability of their outputs.
- Replacing Hsiao will be Josh Woodward, who leads Google Labs and oversaw the launch of NotebookLM, the company's tool that turns text into a podcast-like show.
- The study is expected to presume the marketplace will consolidate from a multitude of contenders to two recognized ad "currencies," which is consistent with the history of the ad industry.
- Google has expanded access to its AI Mode search beyond Google One AI Premium U.S. subscribers. The lack of links to businesses will become a challenge for all when asking for detailed information. The feature provided links to general information, but not when it came to specifics in my searches.
- Is Google a habit? Findings from a study suggest many users stick with Google simply because they never explore other options. When they do, some choose to stay with the competitor, like Microsoft's Bing. DuckDuckGo and other privacy-based engines were not mentioned in the research.
- Rewards went to 660 security researchers who reported security bugs in Google's products throughout the company from its Vulnerability Reward Program. Microsoft also expanded its program.
- Campaigns that succeed in 2025 will come from advertisers that recognize the data is about permanent shifts in how media is consumed, not just election trends, according to LoopMe head of political Robin Porter.
- The online learning platform, which helps students understand homework, has filed a lawsuit against Google, alleging that use of AI-generated summaries in search engine results has "crushed" its website traffic and revenue.
- The AI-based browser will support the rise of agents for search. Agentic search is a type of AI system that can autonomously make decisions, take actions and adapt to changing situations with minimal human intervention,
- As AI becomes an integral part of the ad industry, quantum computing will become a necessary technology because of the speed and quantity of data collected, analyzed, processed and used. Microsoft said it has combined science and art to solve previously unsolvable problems.
- Perplexity became the latest AI company to release an in-depth research tool - following Google, which unveiled something similar for Gemini AI in December, and OpenAI early in February.
- The tool can be used to benchmark application domains such as ecommerce, advertising and social networks. It can handle up to billions of rows, schema complexity, and temporal evolution. For each dataset, Amazon can define relevant predictive tasks, such as estimating missing cell values.
- The evolution of online advertising and search has raised important questions about trust, privacy, and continued relevance of traditional performance methods across digital. Shoppable video content and niche platforms are predicted as top emerging channels for growth in the future - all tied to performance.
- Largo.ai, an AI-driven analytics platform for the film, TV and ad industries, has secured $7.5 million in investment. Stallone is among the backers of the company, which just closed its Series A financing round.
- Winterberry Group estimates investments in connected commerce and technology stacks will drive U.S. ad, marketing, and data expenditures to $585 billion.
