16 Updates - Q3 Earnings, AI Mode Ads & AI Influencers
Happy Friday.
October is done after today, which feels wrong on about four different levels. Two months left in 2026, and I still have goals sitting in "we'll get to it" mode. Honestly? Consider this your friendly nudge to pick one back up this weekend.
- Earnings week delivered the numbers that actually matter for your ad accounts, and Reddit's chart has a twist.
- Google quietly admitted AI Mode earns less than old-school search, then immediately rolled out more AI Mode ads anyway.
- The best-performing AI influencers on social media all share one habit: they hang out with real humans.
Let me get into it.
SOCIAL MEDIA
Reddit's Record Quarter, Pinterest's Shallow Growth, and Meta's AI Pulling Its Weight
Reddit cashes in. Reddit's Q3 shareholder letter showed ad revenue up 54% to $847M, driven by higher prices and more impressions. Video ad revenue more than doubled, which tells you advertisers are shifting budgets into short-form video even on text-heavy platforms like Reddit.
The catch: in October, search referrals fell by over 1M per week on average, and Reddit warned that daily users could drop by a low-double-digit percentage from 130M in Q4. That is Google's AI Overviews eating into Reddit's free traffic, and it is happening to everyone with search-dependent inventory.
Reddit's answer is to push logged-in-only access, with tests rolling out outside the US this quarter. Fewer people may see your ads, but logins hand Reddit more targeting signals. My take: if Reddit traffic converts for you, watch your CPMs closely as the audience shrinks and targeting tightens.
Pinterest grows wider, not deeper. Pinterest reported Q3 results:
- Monthly users: a record 661M, up 10%
- Revenue: $1.174B, up 12%
- Revenue per user: $1.80, up just 1% and down 7% in Europe
Its Q4 guidance of 8 to 12% growth tops out at Q3's pace. Nearly all the user growth came from outside the US and Canada, where Pinterest earns far less per user. If you sell mainly in North America, note that Pinterest's audience there grew just 4%. I think this is the story behind most "record users" headlines: the platform is growing where the money is not.
Meta's AI is pulling its weight, says Meta. Upgrades to Meta's ad ranking system, the algorithm that picks which ads people see, lifted Facebook clicks 8.3% and conversions 15.7% in Q2. That is a meaningful jump from one ranking update, and it explains why some of you saw performance shift without touching a setting.
The boost is automatic, but your goal, data quality, and how much you let the AI optimize still shape results. My recommendation: if you have been fighting Meta's automated placements, this is more evidence that feeding it clean conversion data beats micromanaging placements.
Ads Manager's new Breakdown shows which creative did the work. Meta added a new Breakdown option in Ads Manager showing how each image, video, and text variation performs within an ad. This is the granular creative reporting many of us have wanted for years.
It skips collection and carousel displays, plus some custom conversions like Donate and Subscribe. Still, for standard image and video campaigns, start checking which variation actually drove the conversions instead of guessing.
My take on this section: the platforms with AI-baked ad systems (Meta) are improving advertiser outcomes automatically, while the ones fighting Google's AI traffic squeeze (Reddit, Pinterest) are restructuring access and leaning on cheaper international users. Plan your 2026 budgets accordingly.
AI MARKETING
Google Admits AI Mode Earns Less, Then Rolls Out More Ads Anyway
More chats, less cash. On Alphabet's Q3 earnings call, Google business chief Philipp Schindler said booming AI Mode queries weighed on Search growth. His reason: AI Mode's "lower monetization." Translation: those searches earn Google less money, for now.
Alphabet's revenue still grew 25% to $127.6 billion, but ad growth looked tamer:
- Search and other: $63.35 billion, up 12%, down from 17% in Q2
- YouTube ads: $11.21 billion, up 9%
- Network ads: $7.46 billion, up 1%
Search barely moved from Q2's $63.27 billion, even as CEO Sundar Pichai said query growth, commercial queries included, hit a five-year high. Google blamed currency swings most, and expects them to hurt slightly more in Q4.
AI Mode's "lower monetization" is the line to remember. More queries are not automatically more revenue when the new query format monetizes worse. If you sell ads in any form, this gap is worth tracking quarter by quarter.
The fix is in: more ads. Google's ads in AI Mode now run in 10 more countries, including Australia, Canada, and India. Brazil, Mexico, and the UK are among the Q4 additions. In the US, new multiturn ads also show on follow-up questions within an AI Mode conversation.
Search, Shopping, and Performance Max campaigns are eligible for AI Mode by default, with no separate opt-out. Negative keywords still apply, so review yours. My recommendation: check where your Performance Max spend is landing, because it is now flowing into AI Mode without you choosing it.
Paid, meet organic. Google is also testing Shopping ads inside AI Mode's organic product carousels, under "Sponsored" and "All products" headings. A Google spokesperson confirmed the test.
This blurs the paid-organic line right where shoppers compare products. For ecommerce advertisers, this means your product feed data is doing double duty, and keeping it clean matters more than ever.
And most roads lead to Mountain View. A recent report found 57% of AI Mode's product-answer links led to Google Shopping pages. It counted links, not clicks, but with ads placed beside them, where they point matters more.
Just remember who pays for Google's bigger boat. I think advertisers need to ask whether AI Mode becomes another toll road where Google owns both the traffic and the checkout.
My take on this section: Google has the classic innovator's dilemma with AI Mode. Queries are booming, revenue per query is not, so the answer is more ad placements, more countries, more Shopping links. As an advertiser, get your negative keywords and feed quality sorted before this scales.
INFLUENCER MARKETING
The AI Influencers That Actually Convert All Have Real Friends
Table for one, please. A new study, covered by Science Says, analyzed 9,500+ AI influencer posts and ran four experiments. Only 15% of AI influencer posts show the influencer interacting with someone. Bad call. Posts that did earned 20.7% more engagement, an average of 290 extra likes or comments.
Friends close deals, too. People were up to 17.5% more likely to choose the product an AI influencer promoted when the post showed it engaging with a real person. Selling activewear? Send your AI on a run with a human.
Extras and robot buddies do not count. The boost disappears when a person just stands next to the AI influencer, and when the companion is another AI character. They have to actually do something together. That is a nuance most brands will miss.
Why the coffee date works. People know an AI cannot experience a product in real life, so they trust it less. Watching it do everyday things, like grabbing coffee with friends, makes it more relatable and more trustworthy. The effect grows stronger for less human-looking influencers, such as anime-style characters.
Sometimes fake is the point. Selling fantasy video games? Looking artificial may work in your favor. And on average, human influencers still out-engage AI ones.
The disclosure rules are here. Since June, New York has required advertisers to disclose AI performers that look human, and California's version kicks in January 1. My take: label the AI, and let its human co-star build trust. Passing an AI off as a person is getting legally risky.
My take on this section: if you use AI creators in your content, the formula is simple. Give them a human friend, show them doing something together, and disclose that they are AI. That 20.7% engagement lift is free performance for a disclosure label you have to add anyway.
DATA STORIES
ChatGPT's Day Job Is Information Lookups, and It Is Mostly Personal
ChatGPT has a work life, but its personal life is busier. New data on how people use ChatGPT shows that looking up information now edges out writing as the top use. Roughly 76% of those lookups have nothing to do with work. Practical guidance is 74% personal, and self-expression is 91% personal.
Only one category leans toward work: technical help, at 57%. Writing is close to an even split, with 49% for work.
People now ask ChatGPT what they used to ask Google, like what to buy or how to fix something. Those personal, high-intent moments are where ChatGPT's new ads, now arriving through Amazon DSP, will show up.
If you sell to consumers, my recommendation is simple: ask ChatGPT the questions your buyers ask. Fix any wrong or missing answers about your products before you pay to appear next to them. That is the starting line.
My take on this section: the consumer purchase journey is quietly shifting from search bars to chat boxes. If your product answers are wrong inside ChatGPT, no ad budget fixes that. Audit your brand's AI answers the same way you used to audit your Google listings.
QUICK UPDATES
YouTube: Soon, the cliffhanger does the selling. YouTube will let mobile viewers join a channel membership mid-series to keep watching members-only Shorts episodes. Microdramas, short vertical dramas, passed 6.5B views in the first half of 2026, and 40% of episode views come 30+ days after release, so sponsored episodes keep paying off.
Amazon: Q3 ad revenue grew just 11%, to $19.64 billion, blamed on Prime Day's move into Q2. Ads Agent, its AI ad platform, now lets all US advertisers run sponsored ads, display, video, and streaming TV as one Full-Funnel Campaign. You set the inputs, and AI does the rest.
Adobe: Adobe found AI referrals converted 90% better on US retail sites in September, brought 83% more revenue per visit, and grew 71% year over year. Adobe expects a 159% jump on Thanksgiving. Before BFCM, check what your AI visitors buy.
Shopify: Your newest media buyer works for Shopify. Campaign Autopilot takes your budget and guardrails, picks channels, splits spend, and shifts it toward what works. It is now in every merchant's Growth tab and covers most ad channels. Set those guardrails before BFCM.
Microsoft: Microsoft's search and advertising revenue, minus what it pays partners for traffic, grew 12% last quarter. But that line now includes LinkedIn's ad business, so you can no longer track Bing's ad growth separately. Need a Bing benchmark? It just got blurrier.
Google: Manage a restaurant's Google listing? You can now switch off individual ordering apps, like DoorDash or Uber Eats, in the Business Profile dashboard, for pickup, delivery, or both. Every provider shows by default, so audit yours, but keep the apps your diners already subscribe to.