AI news, data, and consequences
for marketing leaders
Section
Longer looks at the patterns behind the news cycle. Where the marketing AI shift is heading, argued from evidence.
A researcher revised a conclusion after hearing how shoppers made decisions. For marketers, an unanswered product question can call for a different fix than low awareness.
A Similarweb-based analysis finds uneven organic losses. Marketers need comparable websites and their own search data before attributing a decline to AI.
Microsoft's CEO argues that a firm's own AI tests become valuable knowledge. For marketers, the reasons behind an approval are worth keeping.
Kevin Indig's 12-site analysis gives marketers a reason to inspect engagement and tracking gaps before assigning an unexplained traffic increase to AI.
October's Demand Gen update revisits a September measurement announcement. Paid teams need to separate newly reported conversions from extra sales before changing AI campaign budgets.
A 66.6 million-response study puts search in a minority of free, logged-out answers. Content teams need to separate brand mentions, retrieval, and citations before allocating work.
A small software-buying experiment matched listicles, comparisons, and guides to different questions. Content teams get an audit to try, with substantial limits on the evidence.
Anthropic says test models submitted forms and accessed restricted data while trying to complete tasks. Marketing teams need to define which actions an agent may take without a person.
A proposed class action challenges McDonald’s pricing recommendations. Brand exposure starts when customers cannot tell who made the call, even if a person sets the final price.
NIQ reports 55% say AI influences grocery choices, while 27% recently used it to research or decide purchases. Keep each measure's definition in the marketing brief.
Lawmakers passed a disclosure bill for deceptive AI advertising. A signature would start a 60-day clock for teams to review creative, labels, and final placements.
A survey of 7,675 U.S. adults found search is already a common use of generative AI, while only 7.7% use agentic AI today. Marketers should plan for discovery before assuming agents own checkout.
Rodney Rambo had left Converse before a disputed campaign ran, but AI answer engines still attached him to it. For marketers, public company information has become a reputation-control system.
Profound found that brand mentions and citations change sharply when an AI answer engine is told who is asking. One visibility score can hide the buyer you need to reach.
Similarweb's August numbers landed this week and the consumer chart has barely moved in six months. The enterprise chart looks nothing like it, and most AEO scopes are written off the wrong one.
Semrush asked 2,338 US adults what chatbots did to their buying. 57.5% were talked out of a purchase, 57.51% were talked into one, and the deciding input sits on somebody else's website.
The Justice Department told a federal judge on September 1 that training on copyrighted text is fair use, and that licensing fees would mostly enrich the biggest archives. Yours isn't one.
Two studies landed last week measuring what January's unannounced update did. Fewer listicles hold top spots, the ones still standing get more traffic, and the strongest predictor is the date.
A study of 29.4 million profiles found retroactive editing is heaviest in exactly the job titles B2B software sells to. Orthodontists barely touch theirs. Your targeting sits on the edited half.
Neither company has confirmed the reported $12.9 billion deal, and Hugging Face turned down Nvidia's money a year ago. If you don't run your own models, this reaches you through your vendors.
Pew classified roughly 500,000 pages with the detector Substack uses. Commercial .com domains run 10 times the .edu rate, which is exactly where your content lives.
Cloudflare's CFO gave analysts the 1,000x figure everyone quoted and a caveat almost nobody carried. A Sequoia partner's read: the agent is now a customer nobody is serving.
Profound analyzed 24,135 answers. One Claude searches the web in 93% of replies, the other in 13%. They agree on one brand in five, and they read different pages entirely.
Promptwatch says 86%. Qwairy says 95%. A widely shared post says 15% to zero. Promptwatch also says it can't rule out a problem with its own data collection.
Shopify says AI traffic tripled without replacing search. Similarweb's data shows why: most AI-driven visits arrive as branded search, so the dashboard credits Google.
Ad executives putting AI into creative believe young buyers are comfortable with it. Three separate datasets say the audience, the platforms, and the detectors are all moving the other way.
The AMA tracked which marketing roles grew and which collapsed. Influencer marketer came first, up 18%. Marketers ranked that same skill 37th out of 37. That gap is the whole forecast.
Token prices keep falling and AI bills keep climbing. A model released on August 14 runs on a good laptop, and it changes which work is worth paying premium rates for.
A new survey of 2,000 US consumers says most would hand an AI their card. The same survey says four in ten verify every recommendation before buying.
Anthropic is embedding an invisible mark in Claude's text output worldwide. It signals that Claude touched the words, which is not the same as Claude writing them.
Scrunch tracked 3.5 million citations. The median one is half gone in 4.5 weeks, 3.4 on ChatGPT. The AEO retainer you were quoted runs 12 months.
42% of US shoppers say AI tools stretch out how long they take to decide. The company that measured it sells ads into exactly that window.
Forrester's own data says 68% of buyers arrive with a favorite, and that favorite wins 80% of the time. Your intent tools are working the smaller half of the market.
Orbit Media measured 29 million visits across 97 B2B sites. AI referrals converted at 1.91% against organic search's 0.71%, and accounted for 140,000 of those visits.
Across a year of our reporting, the pattern under every AI agent is the same. It turns your team's hard-won judgment into an asset someone owns. The only question that matters is who.
Companies are deleting the junior roles where marketers learn judgment while demanding everyone do senior work. You can't run an all-seniors team forever, and the bill arrives around 2028.
The marketing task that survives AI keeps turning out to be the same one: checking the machine's work. Growth teams are reclassifying into quality control, and the best auditor now wins the seat.
The industry fixed unmeasurable ad inventory once, in 18 months. AI search is missing the two things that made it work, so the measurement gap isn't closing. It's being sold.
HubSpot built a company teaching everyone to run the SEO blog. On its own blog it now reports AI Overviews sit atop half its keywords, and the small-team fix isn't a newsroom.
AI agent requests jumped 45% last quarter while almost none of it sent a human anywhere. Referral traffic, the number most teams still grade AI by, can't see a mention or a citation at all.
HubSpot's new 'Loop Marketing' says AI search broke the funnel and a four-stage loop replaces it. The shift is real. The tidy diagram selling it is the part to question.
Kevin Indig's latest data shows AI Overview visibility and Google rankings coming apart. The rank tools most SEO budgets are graded on can't see the channel replacing them.
AI marketing money compounds in one place and burns in another. The dollar you hand to the ad platforms works. The one you spend building your own agent doesn't. Budget accordingly.
Seven frontier models shipped in 78 days and one got pulled 13 days after launch. The safe move now is making your marketing work portable across any model, not standardizing on one.
Brands are chasing ChatGPT ads while AI agents reshape how people buy. eMarketer expects AI to drive $20.6 billion in US online sales this year, and agents shop from product feeds, not ad campaigns.
A publisher coalition including the BBC and the FT standardized how to track AI's use of content, down to when a model uses your work and credits no one. Any brand with owned media should watch.
The company spending $145 billion on AI just told staff its agents haven't kept pace. Marketing is betting budgets on the same promise, and only 12% of its AI pilots are fully in production.
The owned-media boom plus AI drafting will bury B2B in competent, sourceless content. The publications that win will be the ones that can prove what they print.
The news about AI and marketing now arrives from a dozen channels faster than any team can track. So I built a newsroom that reads all of it, and I edit down to what's worth your time.
Buyers self-educate in private, zero-click search eats the top of the funnel, and attribution sees none of it. AI didn't kill the funnel. It revealed which teams never built a demand engine.
Net Effect is The State of AI Marketing's Monday recap: how AI is changing marketing, the moves brands are making, and what to do next.
Trouble loading the form? Open the signup form .