The Feed Report: Issue #3

Every platform went crazy at Cannes. Here's what survived contact with reality.

Jun 25, 2026

If Coachella is the brand activation superbowl, Cannes was the platform AI rollout superbowl on steroids. When in France, I guess!

By Thursday’s count: six AI product launches, two measurement updates, one creator infrastructure play, a congressional committee vote, and a 500M user milestone from a platform that still doesn’t have a desktop app.

The week’s thread isn’t all AI, exactly. (I hope we never get to the point where that’s all I have to write about because quite f*cking frankly the bubble can’t pop quick enough.) Let me clear my throat though, the thread is two instinctual evolutions running in parallel.

Platforms are building more automation into everything: agentic ads (buzzword overuse alert going off!), AI creative models (eye roll), and finally, machine-driven optimization at the asset level (trigger ten million LinkedIn(fluncer) posts about optimizing ads within AI platforms without touching anything).

But, ironically enough, Meta is simultaneously giving users more manual control over their own feeds. Instagram extending topic control to the main Feed. Threads launched Your Algo. Despite the obvious contraction, these are platforms are responding to the same problem: when AI-driven recommendations started surfacing content nobody asked for, engagement metrics held steady but trust didn’t. The “Your Algo” rollouts are the correction. The agentic ad tools are the revenue play. Both are live at the same time because they have to be.

Congress moved while all of this was happening. The KIDS Act cleared committee with bipartisan support. It’s not law, and the Senate version isn’t reconciled yet. But a bill that passes committee with that kind of margin is closer to law than a bill sitting in committee. The compliance clock starts before the signature.

This issue is a marathon, not a 5k (although it is 5k words). Grab your go-go juice. Let’s go.

Burn Notice

Awareness-level headlines worth knowing.

Meta expands Live Video Ads globally + virtual card checkout

Meta opened Live Video Ads to all advertisers globally. Ads run inside livestreams with real-time interactive elements.

Virtual card checkout is the more interesting piece: product discovery to purchase without the user leaving the stream.

If your brand runs live shopping or live events on Facebook, this is available now. My call: test with some extra dollars before budgeting around it.

TikTok Shop bans AI-generated voices, prerecorded audio, and static images from promotional livestreams

US academy policy has been updated and is now being enforced.

Any use of AI narration, slideshows, looped footage and screen recordings are prohibited from TikTok Shop livestreams. All verbal communication must be real-time, spoken or signed. Still-frame content cannot exceed 50% of your screen.

Violations will run through the Creator Health Rating system and consist of: commission restrictions, content removal, and account bans are at stake here.

For recorded shoppable videos, here’s the requirements: minimum 3 seconds of dynamic content, creator’s face present on-screen, physical product present.

If any active TikTok Shop workflow depends on automated narration or static setups, audit it today.

Instagram rolls out per-slide carousel captions

20 photo dump, 20 captions. You can toggle multiple captions in the composer dropdown. I got access to this early and while requires more copy work, it’s a good way to add context to carousels without over-doing on-image text.

Only caveat I kind of need fixed? On web, this just turns into one long caption. Hopefully this is ironed out by launch.

Right now, this is rolling out globally and expected to be fully global by June 27th.

Carousels have routinely outperformed Reels in engagement since 2023. Per-slide captions increase slide-level context and will likely improve dwell time (and reach, we can only hope).

Edits app gets desktop version + AI production assistant

The Edits app was announced at a Meta creator event in LA. Desktop version confirmed, no ship date. This simply cannot come soon enough for those that hate CapCut (me).

Edit’s AI assistant will use Instagram performance data to surface content ideas and trending audio. Currently, Beta tab and expanded audience insights are live now. Meta reports Edits-created content has a 10% higher save rate and 2% higher reshare rate vs. non-Edits content.

Worth watching. Keep your eye on this.

Microsoft Ads adds LinkedIn job seniority targeting

Ten seniority tiers (CXO to Volunteer) have been added across Search and Audience campaigns in both targeting and observation modes.

LinkedIn-quality professional targeting on Bing are leveling CPC’s at $2-6 vs. $8-15+ on LinkedIn native.

For B2B advertisers already running Microsoft Ads: consider adding a seniority targeting as a test layer this month.

Meta’s Nielsen DMA → Comscore Markets deadline hit for automotive

Enforcement went live June 22. Automotive model ad campaigns using dma_codes stopped delivering.

Migration to Comscore Market IDs required via Catalog Manager or API. If you run automotive on Meta and saw a delivery drop around June 22, this is why.

Meta reportedly building a prediction markets app called Arena

I felt compelled to include this because of a gut feeling. Allegedly, this will be a smartphone app with a video game-style points system; real money possible later.

Meta has run a version of this before (Forecast, 2020-2022). Kalshi and Polymarket has a combined volume topped $130B this year.

Apparently ad revenue isn’t enough and they want a piece of the prediction market pie. Why not encourage and enable gambling addictions now that they’ve been been through the ringer for election interference, data-fraud and whatever else they’re probably paying out that we don’t know about that is ethically questionable. For those that have read Careless People, this is just adding so much more fuel to my Meta resentment.

The Safe is where I break down the week’s most important stories with full context, actionable next steps, and the read for your accounts and your clients. Paid subscribers only.

If you’re on the free tier, upgrade here for just over $5 a month on the annual plan.

The Safe

Paid-exclusive. Ten stories. The week’s most important news broken down for the person running the accounts.

LinkedIn shipped four features at Cannes. One of them changes how you think about who posts.

LinkedIn dropped four features this week, timed alongside its headline B2B partner status at the Cannes. (Is it just me or is Cannes much bigger than I realized? Anyways.) Taken individually, each feature is incremental. Taken together, it feels a little more coordinated. Let’s break it down.

Collaborative Posts went live June 22. Now, multiple users and company pages can co-author a single post. Both contributors appear at the top; reach combines across networks. It’s LinkedIn’s version of the Instagram Collab feature, and on a platform where employee voice already outperforms company page reach by a wide margin, I like this.

Employee advocacy math has changed. The practitioner answer to “exec post or brand page?” has been “exec, because reach” for years. Collaborative Posts blurs that line. You can publish from the brand account, tag an exec as co-author, and surface to both audiences in one post. This is structural change for any B2B team running thought leadership through individual accounts.

The first thing to do is pilot it: pair an exec with the company page on a real launch announcement, measure reach against your single-author baseline, and find out whether combined reach actually adds up the way LinkedIn’s pitch implies it does.

Connected Apps on Profiles launched June 17. Users can now connect 12+ platforms to their LinkedIn profile: HubSpot, Fiverr, Beehiiv, Descript, Duolingo, Lovable, Replit, and others. The connected app generates a short description of actual usage. Not self-reported skills. Platform-verifiable activity.

LinkedIn’s old endorsement system was useless. Anyone could endorse anyone for anything, and most profiles had a wall of “Microsoft Excel” from coworkers who’d never seen their work.

Connected Apps replaces credential claims with platform evidence. Your HubSpot profile shows 47 workflows built. Your Replit profile shows 12 public projects. That’s a different quality of signal than “experienced in CRM.” That could be an entire hiring decision just from that social proof. For agency hiring and B2B sales, this is a real lever.

Practitioners: connect the tools you actually use, not the tools you want to claim.

The bundle is a coordinated B2B-creator push. LinkedIn is the Headline Partner for the B2B track at Cannes this year. Shipping creator-side and brand-side features in the same week it’s running B2B sessions at the Carlton isn’t a coincidence.

LinkedIn is formalizing its position as the B2B creator platform (and clearly spending big money doing it). The shift to professional network with a content feed to platform with creator infrastructure that B2B brands build strategy around feels extra real now. TikTok and Instagram can’t contest that position. They’re consumer-first by design.

Brand Kits (June 21) let you configure colors, fonts, and brand voice for LinkedIn’s AI-generated content. If your team uses LinkedIn AI for brand posts, configure Brand Kits before publishing anything else. Ungoverned AI content on a brand page is not a small risk. This is the guardrail that should have been in place before you started using the feature.

Animated posts (June 18) are native in the composer now. Minor feature. Worth knowing so you don’t spend 20 minutes troubleshooting why GIF uploads stopped failing (lol).

What we don’t know yet: Whether Collaborative Posts will be eligible for Sponsored Content amplification, Brand Kits API access for agencies managing multiple client pages, and the full Collaborative Posts rollout timeline beyond the Cannes pilot.

Actions:

Pilot Collaborative Posts the moment it’s available on your account. Baseline your current exec-post reach first so you have a comparison point.
Configure Brand Kits under Page Settings before using LinkedIn AI for any branded content.
Connect apps in your profile Skills section.
Update your LinkedIn SOP. None of this is in your current playbook.

Pinterest shipped four AI products in one day. The paid media story is getting undercovered.

On June 17, ahead of Cannes, Pinterest launched Business Assistant, an MCP server, an upgraded Performance+ Creative AI model, and a standalone shopping app called Ask Pinterest. Press coverage led with Ask Pinterest because it’s the most legible consumer story. The practitioner stories are the other three.

Performance+ Creative AI model is globally available now. This upgrade shifts optimization from the ad level to the asset level. Instead of selecting the best performing ad, the model evaluates a broader creative set and selects the best variant per impression.

Pinterest’s internal testing shows a 7.5% click-volume lift vs. the prior model.

For context: Performance+ is now driving approximately 30% of Pinterest’s lower-funnel revenue at 631M MAU and $1.008B in Q1 revenue (18% YoY).

If you’ve been writing off Pinterest as an intent-signal platform without real performance infrastructure, that read is out of date. The 2026 Pinterest brief is structurally different from 2023 Pinterest. The 30% lower-funnel revenue figure is the one that changes the Q4 platform mix conversation.

The practical implication of this asset-level optimization is that creative volume matters more than it did before. If you’re feeding three variants into an algorithm that selects per impression across a broader asset pool, you’re not using the feature. Audit your Pinterest creative volume before assuming Performance+ will do anything meaningful for your accounts.

Business Assistant is an AI co-pilot embedded in Ads Manager and on mobile. It surfaces trend data as graphs with actual Pins (not text walls), recommends Pins to promote and sends proactive mobile notifications. This is currently in closed US beta.

This is the SMB unlock Pinterest has needed for a while. Most small-budget brands skip Pinterest because the interface is intimidating without a dedicated media buyer. An embedded AI assistant that is able to tell you what to promote and shows you where the trends are changes who can actually run Pinterest ads.

Apply for the beta. Don’t budget around it yet.

Pinterest MCP is the one the trade press isn’t covering. Pinterest launched a Model Context Protocol server that lets external AI agents read campaign, analytics, and keyword data directly from your account. Alpha partners: PMG, Pacvue, Dentsu, Havas, Innovid by Mediaocean, and Omnicom’s Jump450. Those aren’t random names. They’re holding companies actively running agentic campaign workflows. Pinterest isn’t asking agencies to add another tab to their dashboard. It’s making itself available inside the AI stacks agencies already use.

As of this week, six platforms have first-party MCP servers: Pinterest, Microsoft Advertising, Google, LinkedIn, Snapchat, and Reddit. Meta doesn’t. TikTok doesn’t. If you’re running AI-assisted campaign analysis using Claude, ChatGPT, or Copilot, you can now pull Pinterest data natively. Request alpha access.

Ask Pinterest is experimental. This is a standalone conversational shopping app, web-based, US limited. Most AI shopping app experiments don’t survive their first year. Pinterest has a unique asset in the Taste Graph, but the competitive environment this week (Reddit shopping tools, Amazon Alexa+ Agentic Ads) is crowded. Honest read: interesting, not actionable.

Actions:

Audit creative volume on Pinterest accounts now. Asset-level optimization needs variants to function.
Apply for Business Assistant beta via Pinterest Ads Manager. If your team runs an AI stack for campaign analysis, request Pinterest MCP alpha access.
Add Pinterest to your Q4 platform mix evaluation if you’ve been ignoring it. The platform you wrote off in 2023 is not the platform running 30% of lower-funnel revenue in 2026.

Threads crossed 500M users. The reach question just became a platform planning question.

Threads hit 500M monthly active users on June 16, adding 100M since August 2025. The milestone came alongside two feature launches: Your Algo and the graduation of Communities out of beta.

Threads opened globally to ads just in February of this year. For the first year, the practitioner response was “watch this.” At 500M MAU, that response is overdue for revision. 500M MAU isn’t a signal of future potential. It’s a reach number. Every additional user is monetized inventory. If Threads isn’t in your paid media plan for Q3/Q4, the conversation is no longer “should we eventually test this.” It’s “why haven’t we.”

Your Algo lets users tell Threads to show more or less of specific topics for 1, 3, or 7 days. This has been rolling out in US, Canada, UK, Australia, and New Zealand.

The organic reach variable to track: when users can configure their topic feed at the platform level, distribution gets less predictable for accounts that don’t cleanly map to a named interest.

Ask this for every brand you manage on Threads: if a user filtered your content category down for a week, does your account survive that? If the answer is no, that’s a content positioning signal, not a platform problem.

Communities graduated out of beta with custom icons, a Communities Hub, and expanded champion recognition. The Instagram Groups parallel is obvious. The LinkedIn Groups parallel is more instructive. LinkedIn Groups had enormous early community value, then organic reach dried up and engagement collapsed. Threads Communities is early enough that the outcome isn’t predetermined. The read for B2B brands: monitor which communities form in your category now. First-mover advantage in early-stage community features is real and doesn’t last long.

Actions:

If Threads isn’t in your paid media plan for Q3/Q4, add a test line item. 500M MAU is the inflection point for that conversation.
For organic: audit your content topics against how a user might configure their topic preferences. Could someone name what you post about and filter for it? If not, that’s the adjustment to make.
For B2B: identify relevant Communities forming in your space now, before the space fills up.

Get 20% off forever

Instagram gave users control over their main feed algorithm. That affects your distribution.

Instagram extended “Your Algorithm” topic control to the main Feed this month. Just outside the strict window for this issue, but the main feed expansion is the story, and most practitioners haven’t processed what it means yet.

Timeline:

Reels in October 2025.
Explore in April 2026.
Main Feed in June 2026. Three surfaces in eight months.

That’s not a feature rollout cadence. That’s a stiff chiropractic adjustment for the platform.

Under Settings → Content Preferences, users can now see which topics Instagram’s AI has connected to their account and manually add or remove them. This is the user-level override of the recommendation algorithm on the main feed, not just secondary discovery surfaces.

The reach implication for niche brands is real and being underreported. Accounts with clear, named topic categories (fitness, food, travel) are probably fine. The algorithm’s categorization likely maps to what those accounts already produce.

Accounts with niche, mixed-topic, or categorically ambiguous content are more exposed than they were six months ago. If your account doesn’t cleanly match a topic a user would recognize and configure, your distribution has a new variable you don’t control.

Mosseri, Mosseri, Mosseri. He acknowledged publicly that AI-driven recommendations had eroded users’ sense of control. That’s the platform chief confirming the algorithm broke something and this is the fix.

For practitioners: algorithmic distribution is now partly user-governed. Content that earns saves, deliberate follows, and clear topic alignment is more resilient than content that relies on pure algorithmic push. You can’t fully optimize your way into every feed anymore.

The pattern tells you where this is heading. Reels, Explore, main Feed. If Instagram extends user-level topic control to every surface, the optimization question changes from “how do I get the algorithm to show this” to “how do I become the account a user actively wants in their feed?” Those questions have different answers. The second one is harder to fake and more durable over time.

Actions:

Review your accounts through the lens of topic clarity. Could a user name what you post about and find you by searching that topic? If not, that’s a positioning question that is now also a reach question.
Follow Adam Mosseri if you don’t already. His public statements are the closest thing to early warning practitioners get on Instagram.
Do not treat this as a consumer feature with no brand implications. User-level algorithmic control on the main feed affects every account that depends on algorithmic distribution.

The KIDS Act cleared committee. The US is late to this, and it still isn’t done.

The House Energy and Commerce Committee approved H.R. 7757 on June 22 with bipartisan support. The bill packages a rewritten Kids Online Safety Act with 13 additional child safety measures. It’s not law. But a bill that passes committee with that margin is closer to law than a bill sitting in committee, and the compliance window is shorter than most practitioners think.

The compliance clock is already running. The bill’s effective date is 18 months after enactment. For brands running teen or family-targeted campaigns, the audit starts now. Not when it passes the Senate. Not when it’s signed. The moment to map which accounts and audiences are in scope is before everyone else is scrambling to do the same thing.

Federal preemption is the structural change. Right now, national campaigns require navigating California, Texas, Arkansas, Utah, and a growing list of state-level child online safety frameworks separately. A federal standard collapses that patchwork. One standard, not fourteen.

The catch: the floor rises everywhere. States with looser requirements get pulled to the federal baseline. That affects targeting parameters, ad creative standards, and platform availability for any brand whose audience includes users under 18.

The Senate fault line. The House version dropped the “duty of care” provision that the Senate KOSA version requires. Senate sponsors said a bill without it is dead-on-arrival. What actually passes will be determined by whether the chambers can reconcile on that provision.

I have to include this quote: “KOSA without a duty of care isn’t KOSA—it’s a blank check to Mark Zuckerberg to exploit children. The House’s toothless & tepid capitulation is dead in the Senate & a betrayal of families suffering from Big Tech’s greed,” Sen. Richard Blumenthal, a co-author of the Senate version, posted this on socials this week.

The practitioner read: don’t act on the House version alone. The version that becomes law will look different. Watch the Senate reconciliation before restructuring anything.

The US is behind, and the gap is wider than most practitioners know. Australia passed a law in November 2024 banning users under 16 from social media entirely, with the enforcement burden placed on platforms, not parents. The UK’s Online Safety Act is rolling out now with specific provisions governing how platforms design experiences for children at scale. The EU’s Digital Services Act has child protection requirements for large platforms already in force. The common thread: the legal burden has shifted to the platforms, and the threshold for what counts as adequate protection keeps rising.

The US version, even if it passes with the duty of care provision intact, is more conservative than what’s already enforced in Australia and the UK. But the direction is the same. The US tends to follow on platform regulation, not lead. It’s following. If your brand markets to families, teens, or anyone under 18, look at what compliance looks like for brands operating in Australia right now. That’s the closest working preview of where US requirements are headed.

Actions:

Identify every account and campaign where the target audience includes users under 18. This is your scope list for compliance planning.
Audit current targeting and creative for those accounts against existing state laws (California, Texas, Arkansas, Utah). Federal preemption will eventually replace this work, but you need to know your baseline now.
Watch Senate reconciliation on the duty of care provision. That’s the variable that determines how significant this legislation actually is.
Look at what Australian brands are doing on compliant social media. It’s a working preview of the US in 18-24 months.

TikTok’s agentic play: Symphony Agent writes your briefs, Custom Creator Networks replaces your influencer software.

TikTok dropped two products at Cannes that are easy to underestimate because they sound like adjacent-to-ads features. They’re not.

Custom Creator Networks is the bigger story for anyone running creator marketing. TikTok is building proprietary creator management pools inside TikTok One.

Brands can vet and save creators, centralize briefs and contracts, and track performance in a single dashboard. The tools this competes with: Grin, AspireIQ, CreatorIQ, and the spreadsheet-plus-email-chain workflow most teams still run. Third-party creator marketing platforms charge subscription fees for functionality TikTok is now providing natively. That’s either a cost saving or a lock-in play, depending on how you look at it.

If you’re running creator marketing on TikTok through a third-party platform, compare what you’re paying against what TikTok One now provides. And think about what happens to your creator database if it migrates into TikTok One. Creator relationship data is an asset. Know where it’s housed before you move it.

Symphony Agent lets advertisers input a brief and receive creative concepts, scripts, and strategic recommendations. Not a copy generator.

The play is compressing the gap between brief and first creative pass for teams without a full in-house creative workflow.

The honest caveat: anyone who has briefed an AI tool on a nuanced brand voice already knows what happens. It produces content that looks structurally right and reads tonally wrong. The vocabulary is present. The brand voice isn’t. Symphony Agent is faster at first draft. It is not better at final output. Use it for brief-stage ideation: three directions in less time than one. Then evaluate and build from there. The practitioners who treat it as a starting point will get value from it. The ones who activate-and-publish won’t.

Actions:

Evaluate Custom Creator Networks against your current creator marketing stack. If you’re already managing TikTok-specific creator relationships, the case for a separate subscription just got harder to make.
Use Symphony Agent for brief-stage ideation on campaigns where speed matters. Do not use it as a shortcut past brand voice review.
Confirm where your creator data lives before migrating anything into TikTok One.

Google Ads will auto-classify your Customer Match lists in August. You have eight weeks.

Starting in August, Google Ads will automatically classify Customer Match lists against its Sensitive Category policies. Lists that contain or are inferred to include users flagged for sensitive categories (health conditions, financial status, political beliefs, religious affiliation) will be restricted or barred from targeting.

Eight weeks is not a comfortable runway. Audit your active Customer Match lists now and map them against Google’s Sensitive Category definitions. You don’t find out at enforcement. You find out when a campaign stops delivering on the day you were expecting to launch something.

The broader read. Google has been systematically pushing advertisers toward first-party data that is clean, consented, and categorically unambiguous. Customer Match built from your own purchase history, direct email subscribers, or loyalty program data is the low-risk zone.

Customer Match built from third-party lists, enrichment services, or broad CRM exports is the risk zone. Google is narrowing the acceptable surface area for audience targeting, and the direction it’s narrowing toward is data you collected directly with explicit consent.

This is a structural posture shift playing out in increments. The advertisers building clean first-party data infrastructure now will have fewer compliance disruptions in 2027 than the ones treating each update as a one-time patch.

Actions:

Audit active Customer Match lists before August. Export list names and sources, evaluate against Google’s Sensitive Category definitions, and flag anything built from third-party data or enrichment services.
Brief clients with significant Customer Match usage before they find out from a delivery drop.
Start mapping first-party data collection touchpoints for every brand you manage: email sign-ups, loyalty programs, purchase history. Clean lists now are a structural advantage in an environment where the audience targeting surface is shrinking.

Amazon launched agentic ads inside Alexa+. Here’s what that actually means.

Amazon launched Alexa+ Agentic Ads at Cannes. The premise: an Alexa+ user asks to “find and book a restaurant for Saturday,” and the ad surfaces, selects, and executes the transaction without the user leaving the conversation.

The conversion funnel doesn’t get shorter. It disappears from the user’s perspective.

Amazon’s position here is structurally different from every other platform making agentic ad claims. No other platform has the combination of purchase intent signals, transaction history, Prime membership behavior, and voice commerce infrastructure at Amazon’s scale.

When Amazon makes an agentic ad play, it does so from a data position that Meta, Google, and TikTok cannot fully replicate. The intent data isn’t browsing signals. It’s purchase completion data.

What agentic ads actually mean for the buy side. Traditional ads ask users to click. Agentic ads ask an AI agent to execute on the user’s behalf. Here, you’re buying a position in the agent’s decision stack. How those decisions get made, what signals determine which advertiser the agent acts on, and whether attribution exists in a transaction the user never actively approved: these are open questions without industry answers yet.

Honest read: too early to act on. Alexa+ Agentic Ads are in early access. No CPM benchmarks. No attribution standards. No practitioner case studies. The budget case doesn’t exist in any standardized form yet. The right posture is understand the format and watch the early access data. Don’t allocate budget until the measurement story is clear.

Actions:

If you manage brands with Amazon presence (retail, restaurants, services), request early access to understand the format before your competitors do.
Watch Q3 for CPM and conversion benchmarks from early adopters. Start thinking about what “winning an agentic recommendation” looks like in your category. The optimization discipline for agentic ad placements will be real, and the practitioners who understand it earliest will have a meaningful head start.

YouTube updated its ad measurement stack. The attribution gap is narrowing.

Google announced expanded YouTube ad measurement tooling on June 23, including Insights Finder, currently in beta. The existing stack (Brand Lift, Search Lift, Conversion Lift) now has additional tools accessible via beta application.

The measurement gap YouTube has always had. Meta’s attribution tools are imperfect, but they’re mature and well-understood by practitioners. YouTube’s measurement have historically lagged.

Brand Lift studies require minimum spend thresholds. Conversion Lift attribution relies on Google’s modeled signals rather than direct data. The full-funnel case for YouTube has been harder to prove to clients than it should be for a platform with the reach that it has (say it LOUDER).

A more complete measurement stack is the infrastructure that closes that gap. If it works as described, the data case for YouTube in full-funnel budget conversations gets materially stronger. That changes what you can say to a client who keeps pulling YouTube budget to put into Meta.

Apply for Insights Finder beta now. Beta access matters because early users influence how the tool develops. If YouTube measurement is a priority for your accounts, being in the evaluation period is worth more than waiting for GA and reading someone else’s case study.

The honest caveat. The research on what Insights Finder actually does is thin. The product page is light, and no practitioner reporting exists yet because the beta is new. Apply for access. Evaluate once you have it. Don’t restructure your YouTube measurement approach based on a product description. Find out what it actually does before telling a client it exists.

Actions:

Apply for Insights Finder beta via the YouTube Ads product details page.
While waiting: audit your current YouTube measurement setup against the full Brand Lift, Search Lift, and Conversion Lift suite.
If you’re not running all three on eligible campaigns, start there. When access comes through, build a test case with one campaign before applying it broadly.

Reddit launched its most significant ad product suite. One number is worth acting on now.

Reddit dropped four products at Cannes under the Community Intelligence umbrella: Free-form Ad Generator (Beta), Tailored Creative Assets (Beta), Redditor Highlights (GA), and Shopping Listing Ads (Alpha). All built on 25B+ posts and comments as training data.

The data asset is genuinely different from what other platforms have. Meta knows what you click. Google knows what you search. Reddit knows what communities actually say to each other about specific products and problems. Unfiltered, in depth, at scale.

Tailored Creative Assets uses that community content to inform ad creative. The premise: if you know what a subreddit actually says about your product category, you can make ads that don’t read like ads. Whether the execution delivers on that premise is the real question (more on that below).

The number worth acting on now: 130% VTR lift and 71% completion lift for 6-second video optimization. Reddit’s new 6-second engaged video view optimization goal produced those numbers in early testing. If you’re running video on Reddit, or have avoided it because completion rates were mediocre, test the 6-second format with this optimization goal before anything else. That’s not a “watch this” stat. That’s a “run a test this quarter” stat.

Reddit’s creative problem is real and is not solved by an AI tool. Brands fail on Reddit because ads that perform everywhere else don’t perform there.

Reddit users are specifically hostile to content that reads like marketing content in a space they didn’t opt into marketing content. Tailored Creative Assets uses community language to try to solve this natively.

The honest question is whether an AI model trained on community content produces ads that actually feel native to a subreddit, or whether it produces ads that use the right vocabulary but still read like an outsider guessing. No external validation exists yet.

Start with what’s at general availability. Of the four products, only Redditor Highlights is GA. Everything else is beta or alpha.

Don’t restructure your Reddit strategy around tools that may change significantly before general availability. Add the 6-second video test, apply for beta access on Free-form Ad Generator and Tailored Creative Assets, and evaluate from there.

Actions:

If you’re running video on Reddit: test the 6-second optimization goal this quarter. The early performance lift is significant enough to warrant a real test with budget behind it.
Apply for Free-form Ad Generator and Tailored Creative Assets beta access.
Use Redditor Highlights now. It’s live, and understanding your Reddit audience before building creative is not optional.
If Reddit isn’t in your current stack: don’t add it reactively. Evaluate whether the audience is there for your specific category first.

That’s Issue #3. Longer than usual because Cannes Lions turned into a platform product summit for the ages while Congress decided to move at the same time. Not the easiest week to run a newsletter. The sun has set.

Until Next Thursday,
Mckenna

Thanks for reading Burn Your Feed! This post is public so feel free to share it.