In partnership with
Happy Monday.
Here's a fun way to start the week: go and ask your DSP for the iOS 27 win logs. Then count the ads you paid for that never rendered.
That's not a hypothetical. Two agency buyers told AdExchanger their win logs for clients using The Trade Desk contain iOS 27 ads that clearly did not serve. Won the auction, paid, nobody saw it.
iOS 27 shipped in mid-September with a private blocklist baked into WebKit, the engine every browser on an iPhone has to use, Chrome included. If a domain is on the list, Safari stops talking to it.
The first names caught were The Trade Desk's ad-serving domain, LiveRamp, Permutive, Experian's Audigent, ID5 and the Unified ID 2.0 program. No announcement, no guidance. One engineering leader described it to AdExchanger as "Very subtle, easy-to-ignore signals, then all of a sudden."
The Trade Desk got a reprieve last week: an iOS 27.2 beta takes its domain off the list, after a back-and-forth on a public bug tracker. Everyone else is still blocked.
And Apple has written code, not shipped yet, that would let it add companies without a software update, which is the bit that should bother you. Rowena Lam at IAB Tech Lab put it in one line for Digiday: "What this will allow Apple to do is make this block list dynamic."
Forget The Trade Desk. LiveRamp, UID2 and the rest are what stitch your streaming placements to your mobile ads to your web traffic. They're the glue under the cross-channel dashboard your finance team learned to trust, and Safari on iPhone is where your most valuable readers are.
Digiday heard the execs picking it apart in the queues outside Advertising Week, off the record. One exec didn't expect CEOs throwing themselves out of windows yet, "But if I was a banker today I definitely would not invest in an ID company."
Jay Friedman at Cartograph's view is that mistakes tend to be random, and this one wasn't. Apple didn't respond to either outlet, which honestly tracks with every Apple privacy move I've lived through.
So, your Monday question. Which number on your attribution deck quietly stopped being true in September, and who in your building would know?
Today’s sponsor:
One workspace where buyers, sellers, and AI close deals together
Instead of running your deals across scattered email threads and forgotten attachments, Aligned brings everything into one shared deal workspace.
Buyers get a single place to evaluate, loop in their team, and say yes. Sellers get visibility into what is actually happening when they're not in the room and AI Deal Insights that help you act accordingly, before it's too late.
Best part? Your first deal room takes only minutes to stand up.
TALK ON THE MARKET
Tim Soulo, CMO of Ahrefs, on what AI content leaves you to compete on:
He then describes the spreadsheet-of-topics-plus-agent setup a lot of your peers are running, and where it ends: "then the algorithm catches up and they get a “Google slap.”"
The man sells the SEO tool. He's telling you the shortcut is the problem.
Want to see case studies of AI-powered campaigns?
NUMBERS MATTER
72% vs 45%. When buying a smartphone, 72% of consumers would let an AI agent do the research, but only 45% would let it pay, in KPMG Germany's Consumer Snapshot out last Thursday. And 64% prefer a general AI for objectivity; just 7% prefer a retailer's or manufacturer's.
People will hand the agent the homework and keep the wallet. Your branded assistant starts from 7%.
DAILY PROMPT
Our Cross-Channel Attribution Model prompt, for the week the IDs went missing:
You are a marketing analytics strategist helping a [COMPANY_TYPE] company build a unified cross-channel attribution model. Our challenge: we generate massive volumes of customer touchpoints across [NUMBER_OF_CHANNELS] channels (email, paid social, organic search, direct, affiliate, etc.), but we lack clarity on which interactions actually drive conversion and revenue. Read every attached export before answering, and run the analysis on the data itself with code execution rather than describing what an analysis would look like.
## Current Situation
- Monthly customer interactions: [MONTHLY_TOUCHPOINTS]
- Average customer journey length: [AVG_TOUCHPOINTS_PER_CUSTOMER]
- Current attribution model: [CURRENT_MODEL: first-touch, last-touch, linear, or none]
- Primary conversion metric: [CONVERSION_TYPE: purchase, signup, demo request, etc.]
- Data infrastructure: [INFRASTRUCTURE: CDP, GA4, CRM, data warehouse, or fragmented]
- Attached data: [ATTACHED_FILES: touchpoint exports, conversion table, spend by channel]
## Your Task
Design a practical, implementable cross-channel attribution workflow that:
1. **Consolidates fragmented touchpoint data** from our [NUMBER_OF_CHANNELS] channels into a single customer journey view
2. **Assigns credit intelligently** using a [ATTRIBUTION_MODEL: data-driven, time-decay, position-based, or hybrid] approach that reflects our actual customer behavior
3. **Identifies high-impact channel combinations** that drive disproportionate conversion rates
4. **Provides actionable budget reallocation recommendations** based on true channel contribution
5. **Accounts for channel interactions** (e.g., paid social awareness, then organic search, then email conversion)
6. **Names what AI-assistant referrals and other untracked discovery are doing to the model**, and how to treat them
## Constraints & Context
- We have [DATA_AVAILABILITY: full customer IDs, partial matching, or IP-based tracking]
- Our conversion window is [CONVERSION_WINDOW: 7, 30, or 90 days]
- Privacy compliance requirements: [COMPLIANCE: GDPR, CCPA, or both]
- Current tech stack: [TOOLS: list your actual platforms]
- Team capacity: [TEAM_SIZE: 1-2 analysts, 3-5, or 5+ data engineers]
## Deliverables
Provide:
1. A step-by-step implementation roadmap (phased approach, 90 days)
2. Specific data requirements and transformation logic
3. Attribution model recommendation with mathematical rationale, tested against the attached data where possible
4. KPIs to track model accuracy and business impact
5. Common pitfalls and how to avoid them
6. Sample dashboard structure showing channel contribution
## Important Context
We're not looking for theoretical perfection. We need a pragmatic model that works with our current data quality and team resources. Prioritize speed to insight over statistical purity. Assume we'll iterate based on initial findings, and state clearly which conclusions the attached data supports and which are assumptions.Run it with "partial matching" in the data line, because that's what you have on iPhone now.
CMO CORNER
A vague strategy creates busy leaders. Precise systems make you optional.
What Apple broke is the vague system, the one where nobody could say which ID did what. If your measurement only works when every vendor cooperates, it was never yours.
We provide consulting services for CMOs on AI strategy for marketing teams through our Teamless offering.
TOOL UPGRADE
The data nobody can block is the relationship you own. Three ways to hold more of it:
Postiz — best for teams that want social scheduling an AI agent can drive, through its MCP server and public API. Try Postiz here →
Kittl — best for teams that need social graphics and marketing visuals at quality without hiring a designer. Try Kittl here →
InVideo — best for demand-gen teams turning scripts into production-ready video when the plan suddenly wants streaming TV. Try InVideo here →
→ We have 300+ curated, filtered, and categorized AI tools in our directory. Browse to find the best ones that help your marketing team.
LEARN THIS
The week you need to read a WebKit bug thread: Tech Literacy for CMOs. Built so you can follow a vendor's technical excuse far enough to know whether it's one.
→ We publish workshops, courses, guides, manuals, and more on AI marketing for free. Browse our learning resources for free.
BRAND SLOP
Instagram users have been seeing an ad for a pet incinerator. Orange box, and in front of it, a cartoonish AI-generated cat and dog, posing.
Futurism traced it to an AliExpress listing and found Alibaba's crematorium ads in Meta's ad library carrying Meta's own dynamic-product disclaimer: the advertiser supplies images and text, and Meta's system assembles the combination for the audience. Meta didn't respond.
A veterinary clinic needs that machine. The ad went to people who don't. When the platform assembles your creative, its judgment becomes your brand's. Have you read what yours is allowed to combine?
Reach marketing leaders through us
You’re reading this, and so are nearly 20,000 of your peers. Some of them are your next customers.
If you sell to marketing teams, a single placement is €500 and a full week is €2,000.
If you want to place an ad, simply reply to this email or check our media kit.