13 Updates - Meta TikTok Ad Ban, YouTube Rule Change & Trading Down
Happy Monday.
The platforms spent the weekend redrawing the map. Meta banned its biggest rival from buying ads, Google changed what counts as a YouTube engagement, and new data shows shoppers are not quitting - they are just spending smarter.
Here is what caught my eye today:
- If you run TikTok Shop traffic through Meta ads, you need a new plan this week.
- Your YouTube engagement numbers are about to change, and old reports will not match new ones.
- Three small words in a product title lifted buying intent by 9.5% in a new study.
SOCIAL MEDIA
Meta kicks TikTok off its ad platform, plus agent protocols, teen ad limits, Threads Gems, and LinkedIn's AI filter
It is like Monopoly with house rules. Meta just removed its biggest rival from the board - and told everyone else to play by its AI shopping vision.
Meta banned TikTok from advertising on its apps. The ban covers the US, Canada, Japan, and four other markets, and it includes third-party ads that link to TikTok or other ByteDance properties. If you have been sending Meta traffic to TikTok Shop, that route is now closed and you need to reroute.
This is a power move, plain and simple. Meta is not just competing with TikTok for attention anymore - it is cutting off TikTok's access to Meta's ad inventory entirely. My take: if TikTok Shop was part of your funnel, rebuild that path this week, because this ban is unlikely to reverse.
Meta and Sierra want to set the rules for AI shopping agents. The two companies are designing an open protocol so websites can authenticate AI agents and see what they do. Amazon already blocks these agents outright.
The catch for marketers: an AI agent that shops for your customer might skip your upsells, bundles, and display ads entirely. I think this is worth watching closely - if agents become the buyer, your product page needs to sell to software, not just people.
Florida wants Meta to stop showing teens third-party ads. The state's attorney general asked a judge to bar all third-party advertising to under-18s on Meta's apps. The request also includes age checks, no accounts for under-14s, a 60-hour monthly usage cap for teens, and limits on what it calls addictive features.
No ruling yet, and Meta calls the move meritless. But I would not ignore this - state-level pressure on teen safety keeps growing, and any ruling here would reshape how you target younger audiences on Meta.
Threads now hands out Gems for original takes. The platform auto-awards Gems to public posts and direct replies that spark real discussion with an original point of view. Crossposts do not qualify, and it is US-only for now.
This is Threads trying to buy quality conversation with status instead of reach. My recommendation: if you post on Threads, lead with a genuine take rather than recycled content - that is literally what the algorithm is now paying for.
LinkedIn showed how it catches AI slop. LinkedIn's Tim Jurka explained the system: user reports train large "teacher" models, which then train smaller detectors. LinkedIn now checks every post shown beyond your immediate network, at 94% precision.
That last number matters - this is not a sampling system, it is the reach you actually post for. I think the message is clear: generic AI-written posts will quietly die outside your own network. Write like a person or do not bother.
GOOGLE ADS
YouTube engagement now needs 10 seconds, plus bidless video reach, DV360 location upgrades, and coupons that track to the store
The skip button on a YouTube ad appears at 5 seconds. Until recently, that was also enough for Google to count an engagement. Not anymore.
YouTube in-stream ads now need 10 seconds watched to count as an engagement. A click still counts, as does a full view of a shorter ad - but the old 5-second standard is gone, which means your historical engagement and cost-per-engagement numbers will not line up with new ones.
Also worth checking: Demand Gen's new "Overlay on videos" setting puts your text assets directly on your videos, and it appears to be on by default. My recommendation: before your next report goes out, check that setting and add a footnote explaining the 10-second switch, or your numbers will confuse everyone.
Video reach campaigns can now run without a bid. Google is beta-testing bidless video reach campaigns, where it adjusts bids automatically to spend your full budget.
Google does warn that narrow audiences paired with big budgets can push CPMs up. I think this is worth a test for broad awareness pushes, but keep a close eye on efficiency - handing Google a blank check on bids has a history of expensive lessons.
DV360 bulk uploads get smarter about location. With SDF version 11, bulk uploads can now target people near a business chain's locations. Older versions before v10 stop working in January 2027.
If your team still uploads with old SDF versions, put the upgrade on the calendar now. January 2027 sounds far away until you realize how many bulk sheets your team touches.
Performance Max can now send clicks to your Google Business Profile. A new setup option lets PMax campaigns point clicks to a Google-hosted Business Profile instead of your website, which may shift conversions toward calls and directions.
My take: this is built for businesses where the phone call or the store visit is the conversion. If that describes your clients, test it - but watch attribution closely, because the conversion mix will look very different.
Google is testing coupons that follow shoppers into the store. The Local Coupons pilot lets people save deals they find on Search straight to their mobile wallet and redeem them in person. Retailers only pay when a coupon is actually redeemed.
This finally ties a search ad directly to a store sale, which has been the holy grail of local measurement for years. I think local retailers should be first in line for this pilot.
PRODUCT MARKETING
Three extra words made 9.5% more people want to buy
Your product page says "windbreaker." Your customer is thinking "something for the scooter ride to work." When those do not match, fewer people buy.
Use-specific product names lifted purchase intent by up to 9.5%. Researchers at Dalian University of Technology ran four experiments comparing broad descriptions against use-specific ones. "Sleep-aid scented candle" beat "scented candle" - buyers were up to 9.5% more likely to say they would purchase, and rated the product 9.7% more useful.
The researchers' warning is worth hearing: this measured stated intent, not actual checkout data. Still, renaming a title costs almost nothing, so it earns a test. My recommendation: start with the titles that meet searchers - shopping ads, marketplace listings, and your own site search results - because the lift was biggest for people actively looking for a specific product.
One more sharp detail from the study: do not overdo it. "Traveling scented sleep candle for moms of newborns" is precise and nearly unsellable. Name one clear job the product does, not one tiny persona. And consider saying who it is NOT for - ruling people out makes the right buyer more confident.
DATA STORIES
Shoppers are trading down, not dropping out
New Ipsos data on US households shows consumers easing out of crisis mode - but still counting every dollar.
63% of households are spending less on experiences like travel and dining out, up from 57% a year ago. That was the only cutback that grew. Everything else eased: skipping trips entirely fell to 33% from 40%, delaying a home purchase fell to 14% from 21%, and skipping medical treatment fell to 24% from 30%. Meanwhile 69% still cook at home more, and 20% still struggle with rent or mortgage payments.
My read: shoppers are not giving up the fun, they are doing it on a smaller budget. Shorter getaways, cheaper dinners, staycations. The actionable move for Q4: offer budget-friendly versions of what you already sell - smaller bundles, off-peak deals, "lite" tiers. Those might be your holiday wins.
QUICK UPDATES
Anthropic is banning fake sources for AI answers. From November 12, its usage policy prohibits feeding search engines or AI systems content that lies about who wrote it - including site networks posing as unrelated sources. If you run any GEO or AI-visibility work, review your setup now.
Your September Google ranking drops might not be real. SEOmonitor says Google served incomplete results to automated searches starting September 26, which regular searchers never saw. The spam update ran nearly 14 days, September 24 to October 8. Check Search Console clicks and impressions for that window before blaming the update.
Google ads under AI Mode answers are starting to look like normal search results. Tests show titles switching from black to blue - the color people associate with links - plus favicons and site names replacing the grey box. If this rolls out, check how your favicon and site name appear.
Shopify stores: AI shopping agents can now see your discounts. The Catalog API returns your compare-at price when it is higher than the current price. If you use that field for MSRP, check your Catalog Mapping settings - and note that stores hiding compare-at prices from EEA buyers will not share them with those buyers either.