SignalShift September 2026: Your Next Buyer Never Logs Off

September 2026 put always-on AI agents in buyers' hands. Six signals on what delegated demand means for CMOs, budgets, search, martech and talent.

TL;DR

In September, OpenAI and Meta both shipped always-on personal agents that browse, compare, fill forms and buy on a person's behalf. In the same month, ChatGPT began testing brand-sponsored agents, Google auto-migrated search campaigns away from keyword control, Cloudflare started blocking agents by default on millions of sites, and Gartner told us the bottom rung of the marketing career ladder is going away. Put together, these say one thing: the buyer is starting to delegate, and marketing organizations built to capture human attention now have to earn machine delegation. That changes who we market to, how we spend, what our websites are for, and who we hire.

Introduction

In the July edition, I argued that intelligence had become the new marketing interface. In August, I took that further and made the case that the interface itself was disappearing, both for the buyer and for the marketer operating the stack.

September answered the question August left open. If the interface disappears, who is on the other side of it?

Increasingly, it's an agent. Not a chatbot waiting for a prompt, but a persistent piece of software with its own computer, its own browser, its own memory of what its owner likes, and permission to act. OpenAI launched one at DevDay. Meta launched one three weeks earlier. Salesforce and HubSpot spent their flagship events rebuilding the CRM so agents can work through it. And OpenAI started letting brands put their own agents inside ChatGPT's ad units.

As marketing leaders, we've spent the last twenty years building organizations that are very good at one thing: earning a human's attention and converting it. Every team, every budget line and every KPI on our dashboards assumes a person is looking at something. September is the first month where I'd say, with evidence from several directions at once, that assumption is starting to crack.

This edition is about what that means for how we run our organizations. Not in the abstract, but in budgets, team structures, measurement and hiring.

Executive Dashboard

Shift	What changed in September	Why it matters for marketing leaders
Delegated buying	OpenAI Dots and Meta Muse launched as always-on agents that browse, compare and transact	Part of the buying journey now happens without the buyer present
Agent access as policy	Cloudflare began blocking agent and training bots by default on ad-supported pages from September 15	Whether a buyer's agent can read your site is now a governance decision, not a dev setting
Conversational media	ChatGPT tested Sponsored Agents; Amazon DSP began selling ChatGPT inventory	The ad unit is becoming a conversation your brand hosts
Intent over keywords	Google auto-upgraded eligible search campaigns to AI Max on September 1	Search teams manage inputs and guardrails, not keyword lists
CRM as an action layer	Dreamforce and HubSpot UNBOUND rebuilt the CRM for agents to act through	Martech value moves to permissions, data quality and agent identity
Talent pipeline disruption	Gartner predicts most high-performing marketing teams will remove traditional entry-level roles by 2030	Our leadership bench for 2035 depends on decisions we make in 2026

Signal 01: The Buyer Now Has a Proxy

On September 29, OpenAI introduced dots, which it describes as always-on agents that run on their own cloud computer, connect to more than 4,000 apps, learn preferences over time and work across multiple projects in parallel. Three weeks earlier, Meta debuted Muse, a personal agent with a free tier and $20 and $100 monthly plans, running on its own virtual machine with a visible browser.

Read the two launches side by side and the pattern is hard to miss. These aren't assistants that answer questions. They're delegates that go and do things. Meta's own description, as reported at launch, is that Muse can send an email, book travel, fill out forms, negotiate on a person's behalf and complete a checkout. One early dots tester described their agent noticing an unsent invoice in an email thread, pulling the details and drafting it without being asked.

That last detail matters more than it looks. Proactivity means the agent starts the task. For marketers, that means some buying journeys will begin without the buyer typing anything at all.

Why we're calling this a signal

Platform signal. Two of the largest consumer platforms in the world shipped persistent, acting agents in the same month. OpenAI did it in front of what it told developers is a base of 1.2 billion weekly users. Meta did it with a free tier designed, in Alexandr Wang's words, so most people can do what they need without paying.

Enterprise signal. OpenAI is also rolling out specialist dots for procurement, invoicing, customer support, contracting and, notably, email marketing, plus an integration with Microsoft's Agent 365 for enterprise controls. Procurement agents are the B2B version of a shopper agent. They will read your pricing page before your buyer does.

Behavioral signal. The habit is already forming. Forrester found that 94% of business buyers now use AI in their purchasing process, and Salesforce's 2026 holiday forecast expects 20% of holiday ecommerce traffic to originate from AI chat agents.

Market signal. The agent layer isn't consolidating around one vendor either. Goodie's AI search traffic report showed ChatGPT's share of B2B AI referrals falling from 89% to 63% in eight months while Claude climbed from 1.4% to 18.5%. Buyers will bring different agents, which means we can't optimize for one.

What this means

For two decades, our models of demand have assumed the buyer does the research. Personas, journey maps, content calendars and nurture sequences are all built on the idea that a human will read, compare and decide.

Delegation breaks the middle of that model. The human still sets the goal and still approves the decision, but a growing share of the reading, comparing and shortlisting gets handed off. I'm calling this delegated demand: demand that is expressed by a person but researched, filtered and sometimes transacted by a machine acting for them.

The agent doesn't care about your hero video. It cares whether your pricing is published, whether your specs are current, whether third parties corroborate your claims, and whether it can complete the next step without hitting a wall.

Organizational impact

Most marketing teams don't have anyone whose job is to understand how an agent experiences the brand. Brand owns perception, web owns the site, product marketing owns positioning, demand gen owns conversion. Delegated demand sits across all four, and in most org charts I've seen it belongs to nobody.

What CMOs should do next

• Run an agent journey audit. Give dots, Muse, ChatGPT and Claude the same buying task in your category and document where your brand shows up, where it gets dropped and why.
• Assign ownership. Name one leader accountable for "how machines evaluate us," with authority across web, content, product data and analyst relations.
• Add delegated demand to planning. In 2027 planning, ask each channel owner what share of their funnel they expect to be researched by an agent, and what changes if that share doubles.

Closing thought: We've always marketed to the person who makes the decision. Now we also have to market to the software they trust to do the homework.

Signal 02: Your Website Got a Bouncer, and It Might Be Turning Away Your Buyers

On September 15, a default changed for millions of websites. Cloudflare, which sits in front of a large share of the web, began blocking Training and Agent bots by default on ad-supported pages for new domains, new customers and free-plan users. Search crawlers stay allowed. Crawlers that bundle several purposes follow whichever rule is most restrictive.

I support the intent here. Publishers have been asked to choose between showing up in search and letting their content train models for free, and Cloudflare is right that this is an unfair trade. But for marketing leaders, the timing creates a very real tension. In the same month that buyers got agents capable of researching on their behalf, one of the web's biggest gatekeepers made blocking those agents the default.

Why we're calling this a signal

Infrastructure signal. Cloudflare now formally classifies bots by purpose: Search, Agent and Training. That's the first time a major infrastructure provider has drawn a clean line between "a machine indexing you" and "a machine acting for a customer."

Market signal. The traffic mix has already flipped. Cloudflare's CEO said in June that bot traffic had overtaken human traffic on the internet for the first time.

Readiness signal. As I wrote in Your Website Has a New Audience, Cloudflare's own scan found that 78% of sites have a robots.txt file but only 4% declare any AI usage preferences. Most of us haven't made a decision. We've inherited one.

Platform signal. Google is reshaping the other side of the same equation. Its new Discover "Dive deeper" test, which I unpacked in Google Doesn't Need to Send You the Click Anymore, replaces direct publisher links with AI topic overviews. Machines are mediating content on the way out and on the way in.

What this means

Agent access has become a marketing decision disguised as a technical setting. A B2B software company that runs no ads probably isn't affected by Cloudflare's default today. A company whose blog runs display ads, or that recently moved to Cloudflare's free tier, may now be invisible to a buyer's agent without anyone in marketing knowing.

In client audits, I keep seeing the same three things: robots.txt files copy-pasted from a template years ago, sitemaps nobody has touched since the last redesign, and pricing locked inside PDFs. None of these were strategic choices. They were defaults. In a world of delegated demand, defaults decide whether you make the shortlist.

Organizational impact

Today, bot policy usually lives with IT, security or whoever manages the CDN. Those teams are optimizing for cost, load and risk, which is their job. Nobody in that conversation is asking, "If a procurement agent tries to compare us with our top competitor tonight, will it get through?" That question belongs to marketing, and right now most of us aren't in the room.

What CMOs should do next

• Get a seat in bot governance. Ask your web or IT lead for your current settings for Search, Agent and Training bots, by property.
• Separate the three decisions. You might reasonably block training while welcoming search and agent access. Make that choice deliberately, document it and revisit it quarterly.
• Fix the obvious gaps first. Publish pricing and specs in HTML, refresh sitemaps and declare your AI preferences. This is infrastructure, not differentiation, but without it nothing else matters.

Closing thought: The most expensive marketing decision your company makes this year might be a dropdown setting nobody in marketing has seen.

Signal 03: The Ad Unit Becomes a Conversation You Sponsor

On September 16, OpenAI announced it was testing Sponsored Agents with select US advertisers. After seeing a relevant ad, a user can choose to start a clearly labeled conversation with a business-sponsored agent inside ChatGPT. The same announcement added AI-generated creative, text customization that auto-translates ad copy, and named HubSpot and Shopify as integration partners.

A week earlier, Amazon Ads opened ChatGPT inventory to select US advertisers through its DSP in a managed pilot, with Delta Vacations among the first participants.

Last month I wrote that ChatGPT had become a real media market. September is when it started becoming a different kind of media market.

Why we're calling this a signal

Platform signal. Sponsored Agents change the unit of advertising. The ad no longer sends someone somewhere. It opens a dialogue where your brand's agent answers questions in real time, inside the platform.

Market signal. The money is real and moving fast. ChatGPT ads hit a $1 billion annualized run rate by the end of August and expanded to more than 60 countries by September 23. Sensor Tower data cited in the Amazon coverage showed Amazon had quietly become ChatGPT's top retail advertiser between April and August.

Ecosystem signal. HubSpot announced a CRM-native ChatGPT Ads integration at UNBOUND, which I covered in What HubSpot UNBOUND 2026 Means for Agentic Marketing. When the CRM, the commerce platform and the largest retail DSP all plug into the same ad system within weeks, that's an ecosystem forming, not a test.

Measurement signal. The Amazon pilot offers only aggregated metrics such as impressions, clicks and average cost, and marketers can't choose specific prompt placements. Spend is arriving before the measurement infrastructure.

What this means

Every major digital ad format we've bought has ended in a destination we control: a landing page, a store, a form. Sponsored Agents move the destination inside someone else's platform and replace it with a conversation. The landing page becomes optional.

That changes what "creative" means. The asset isn't a headline and an image anymore. It's an agent that has to know your product, handle objections, stay on-brand and stay compliant, in real time, with someone who is actively deciding. OpenAI is careful to say the sponsored conversation is distinct from ChatGPT's independent answers, which is right. It also means your agent has to be good enough to hold up next to an answer the user already trusts.

Organizational impact

In most marketing organizations, paid media, conversational AI and sales enablement sit in three different places. A Sponsored Agent needs all three: media buying discipline, a well-governed conversational agent and the product depth of your best sales engineer. Nobody's team is built for that combination today.

What CMOs should do next

• Build your brand agent before you need to buy media for it. Start with your own site and use it to learn what questions buyers ask and where the agent breaks.
• Create a joint working group across paid media, product marketing, legal and sales enablement to own the brand agent's knowledge, tone and guardrails.
• Budget a test, not a channel. Ring-fence a small experimental budget for conversational media with incrementality testing designed in from day one, since platform reporting won't give you that yet.

Closing thought: For twenty years, ads earned the click. The next generation of ads has to earn the conversation, and conversations are much harder to fake.

Signal 04: The Keyword Retires Quietly

There was no keynote for this one. On September 1, Google automatically upgraded eligible Search campaigns to AI Max, turning on search term matching for campaigns using automatically created assets and campaign-level broad match. New Dynamic Search Ads campaigns can no longer be created, and existing ones face forced migration in February 2027.

Then, on September 7, Search Engine Land reported that Google is testing traditional exact and phrase match campaigns in AI Mode, but only when Google identifies "explicit and direct user intent."

Read that second story carefully. Keyword control still exists, but Google now decides when intent is explicit enough for it to apply.

Why we're calling this a signal

Platform signal. Auto-migration is a statement. Google isn't asking advertisers to opt into intent-based matching anymore. It's making that the default and setting a deadline for the holdouts.

Performance signal. Google's claim is that advertisers using the full AI Max suite see an average 7% lift in conversions or conversion value at comparable CPA or ROAS, and up to 27% for campaigns that previously relied on exact and phrase match. Whether those numbers hold in your account is the question, but they're what your CFO will hear.

Cross-platform signal. OpenAI's September ad updates moved in the same direction, with optimized CPM bidding generally available and AI-generated creative inside Ads Manager. Two of the largest intent platforms in the world are converging on the same model: you provide objectives, assets and data, and the system decides the match.

What this means

Those of us who managed search through the move to Smart Bidding remember how this goes. The platform takes a lever away, performance initially holds, and the teams that adapt fastest are the ones that stop mourning the lever and start mastering the new inputs.

The new inputs are first-party conversion data, product and service feeds, creative variety, negative guardrails and brand safety rules. Search management is becoming input management. The keyword list was a proxy for intent. The platforms now believe they can read intent directly, and they're pricing accordingly.

Organizational impact

This changes the profile of the search team. Keyword research, match type management and bid adjustments were the craft. The new craft is data hygiene, feed quality, conversion signal design and auditing what the machine did with your brand. That's closer to analytics and governance than to traditional PPC, and many agency and in-house teams are still staffed for the old model.

What CMOs should do next

• Find out which campaigns were auto-upgraded. Ask your team or agency for a list of everything labeled AI Max, and a review of the AI-generated copy now running under your brand.
• Invest in conversion signal quality. Offline conversion imports, CRM-qualified pipeline as a bidding signal and clean product data now matter more than keyword strategy.
• Plan the DSA transition now, not in January. February 2027 is close enough to land in your Q1 plan.

Closing thought: The keyword was never the point. It was our best guess at what the buyer meant. The platforms have decided they can guess better, and we should be spending our energy on making sure they guess right about us.

Signal 05: The CRM Stops Being a System of Record and Becomes a System of Action

Dreamforce ran September 15 to 17. HubSpot held UNBOUND, its rebranded INBOUND, the same week. I wrote about each separately, but they're more useful read together.

HubSpot's argument, as I put it in my UNBOUND analysis, is that the CRM becomes the context layer AI uses to make marketing and sales decisions. Salesforce went a step further. As I wrote in Salesforce Dreamforce 2026: The CRM Becomes Something AI Can Operate Through, AIforce brings Salesforce's data and workflows into external surfaces like Claude and Slack, so people and agents don't need to enter Salesforce at all.

Then, two weeks later, OpenAI announced that dots would integrate with Microsoft's Agent 365 and that specialist dots could be provisioned with their own identities and system credentials.

Why we're calling this a signal

Platform signal. The two largest CRM vendors used their flagship events, in the same week, to reposition the CRM around agents. Salesforce paired that with Claude as Agentforce's reasoning model and a CRM reasoning model of its own, Koa.

Product signal. HubSpot's Smart CRM updates itself from calls, emails and meetings. When the record updates itself, the human job shifts from data entry to data judgment.

Enterprise signal. Agents with their own credentials, like OpenAI's specialist dots, are effectively new members of the workforce. They need permissions, audit trails and a manager.

What this means

In August, I wrote that martech was dissolving into a capability layer. September added the missing piece: who operates those capabilities. The answer is increasingly a mix of people and agents, working through the same systems with different permissions.

This puts three unglamorous things at the center of marketing performance: data quality, because agents act on whatever is in the record; permissions, because an agent with send access to your email platform is a brand risk; and identity, because you need to know which actions were taken by which agent on whose behalf.

Organizational impact

Marketing operations becomes the most strategic team in the department. Not because of tool administration, but because they'll be designing the rules under which agents act on the brand's behalf. The CMO who treats marketing ops as a service desk will find that governance decisions get made elsewhere, probably by IT, and probably without brand in mind.

What CMOs should do next

• Write an agent permission policy for marketing systems now: what agents can read, what they can draft, what they can send or publish, and what always needs a human.
• Elevate marketing ops into strategic planning, and ask them to own agent governance jointly with IT and legal.
• Treat CRM data quality as a revenue investment, not a cleanup project. Agents amplify whatever is in the record, good or bad.

Closing thought: The CRM used to remember what happened. Now it decides what happens next, which makes the quality of what it remembers a leadership issue.

Signal 06: The Bottom Rung of the Ladder Is Being Removed

On September 29, the same day OpenAI launched dots, Gartner published a prediction that should be on every marketing leader's agenda. It expects that by 2030, AI will allow the majority of high-performing marketing teams to eliminate the traditional bottom rungs of the corporate ladder.

The survey behind it, of 1,303 senior leaders conducted from January to April, found 18% of marketing leaders had already eliminated functional roles because of automation, 16% had created new ones and nearly a third had redesigned roles.

Why we're calling this a signal

Research signal. Gartner's data shows role elimination is no longer hypothetical. It's already happening in nearly one in five marketing organizations.

Product signal. OpenAI's launch list for specialist dots included email marketing. Campaign building, list hygiene, reporting and first-draft copy are exactly the tasks we've traditionally used to train junior marketers.

Budget signal. Gartner's 2026 CMO Spend Survey found CMOs are putting 15.3% of their budgets toward AI, while only 30% feel ready to scale it. With flat budgets, headcount is where the money for that AI spend tends to come from.

What this means

Here's the problem nobody is pricing in. Junior roles were never only about output. They were how marketers learned judgment. You learn positioning by writing bad positioning and having someone senior take it apart. You learn media by watching a campaign you built underperform. I wrote about this in Why I Wrote Built to Compound: the things that compound over a career are the lessons you only get by doing the work.

If agents take the work, we have to find another way to build the judgment. Otherwise, by 2035, we'll have plenty of agents and very few people qualified to direct them. Gartner's own recommendation points the same way: give junior marketers exposure to higher-value work earlier, and build judgment, strategic thinking and AI orchestration skills from the start.

Organizational impact

This is the most uncomfortable signal of the month, because the short-term incentive and the long-term interest point in opposite directions. Cutting entry-level roles makes this year's budget work. It also hollows out the leadership pipeline that every other signal in this edition depends on. Delegated demand, conversational media, input-managed search and agent governance all require more judgment, not less.

What CMOs should do next

• Redesign entry-level roles instead of removing them. Make the job evaluating AI output, synthesizing insight and connecting work to outcomes, with real ownership early.
• Change hiring criteria to prioritize critical thinking and the ability to spot when a machine is confidently wrong.
• Build structured mentorship into team design, because the apprenticeship that used to happen naturally through grunt work won't happen on its own anymore.

Closing thought: Agents can do the work juniors used to do. They can't become the leaders juniors used to become.

September in 90 Seconds

• OpenAI launched dots on September 29, always-on agents with their own cloud computers that connect to 4,000+ apps. Meta launched Muse on September 8, with a free tier and the ability to fill forms and complete checkouts. Buyers now have proxies.
• Cloudflare began blocking agent and training bots by default on ad-supported pages from September 15. Whether a buyer's agent can read your site is now a governance decision.
• ChatGPT tested Sponsored Agents on September 16, and Amazon DSP began selling ChatGPT inventory. The ad unit is becoming a branded conversation.
• Google auto-upgraded search campaigns to AI Max on September 1 and blocked new DSA campaigns. Search management is now input management.
• Dreamforce and HubSpot UNBOUND rebuilt the CRM for agents to act through. Marketing ops becomes agent governance.
• Gartner predicts most high-performing teams will eliminate traditional entry-level roles by 2030. Our future leadership bench is at risk.

The idea to take away: we're moving from an attention economy to a delegation economy, and the brands that win will be the ones agents can trust to act on.

The SignalShift Perspective

Each edition this year has peeled back another layer. In May, marketing became a system rather than a function. In June, it entered its intelligence era. July made intelligence the interface, and August showed that interface disappearing.

September showed us what replaces it: delegation.

The most useful way I've found to think about this is in terms of what we optimize for. For two decades, marketing optimized for attention. We bought it, earned it, measured it and built our org charts around it. Delegation needs something different. When a buyer hands a task to an agent, three things decide whether your brand makes it through:

1. Access. Can the agent reach and read what it needs? That's Signal 02, and it's more often a default than a decision.
2. Evidence. Does the agent find enough credible, consistent, current information to recommend you? That's the machine-readable brand work from August and the third-party authority work from July.
3. Actionability. Can the agent complete the next step: compare, request a quote, book a demo, buy? That's Signals 03 and 05, and it's where most B2B experiences still break.

Notice what isn't on that list: persuasion in the traditional sense. Persuasion still matters, because a human still makes the final call and a human still decides which agent to trust. But it now has to survive being summarized, compared and filtered by software before it reaches that human.

I don't think this is a reason for alarm. Delegation will grow unevenly. Meta's Muse is US-only for now, OpenAI's dots are limited to paid tiers, and most B2B buying committees will keep humans in the loop for years. But the infrastructure arrived this month, all at once and from several directions, and that's exactly the pattern SignalShift exists to catch. The organizations that start redesigning around delegation in 2026 will have a three-year head start on those who wait for it to show up in their attribution reports, because it won't.

Signal Ledger

#	Signal	Structural shift	Time horizon	Primary owner	CMO question to ask this quarter
01	The Buyer Now Has a Proxy	From human research to delegated demand	2 to 4 years	CMO and product marketing	Who in my team owns how agents evaluate us?
02	Your Website Got a Bouncer	From bot settings to agent access policy	Now to 2 years	Web, IT and marketing jointly	Are we blocking the agents our buyers use?
03	The Ad Unit Becomes a Conversation	From click-based media to sponsored conversations	1 to 3 years	Paid media and product marketing	Is our brand agent good enough to put media behind?
04	The Keyword Retires Quietly	From keyword control to input management	Now to 1 year	Search and analytics	Which campaigns did Google upgrade, and what is it saying for us?
05	The CRM Becomes a System of Action	From system of record to agent operating layer	1 to 3 years	Marketing ops, IT and legal	What can agents read, draft and send in our stack?
06	The Bottom Rung Is Being Removed	From learning by doing to designed judgment development	3 to 5 years	CMO and HR	Where will our 2035 marketing leaders learn their judgment?

Signals We're Watching

• Muse and commerce. Meta says Muse doesn't share conversations with its ad systems, but Alexandr Wang also said Meta is exploring commerce opportunities. If Meta builds merchant programs around Muse, delegated demand gets a second major marketplace.
• Sponsored Agents beyond the US test. If they roll out internationally and to self-serve advertisers, conversational media moves from pilot to plan line.
• Holiday season as a stress test. Salesforce expects one in three ecommerce sites to run a shopper agent by Cyber Week. November will tell us whether brand-owned agents convert better than third-party ones, and B2B will learn from it.
• Cloudflare's Pay Per Use model. If publishers start getting paid when content appears in AI results, the economics of content marketing and earned media could shift again.
• Google's AI Mode keyword test. If exact and phrase match get broader AI Mode access, keyword control survives in a narrower form. If the test quietly ends, Signal 04 accelerates.
• Specialist dots in marketing. Watch for the first enterprise case studies of email marketing dots. They'll tell us how fast execution roles change in practice.

Frequently Asked Questions

What are OpenAI dots, and why do they matter for marketers?

Dots are always-on AI agents OpenAI launched on September 29, 2026. Each runs on its own cloud computer with its own browser, connects to more than 4,000 apps and can work proactively on a user's behalf. For marketers, they matter because they research, compare and act for buyers, which means part of the buying journey happens without the buyer reading our content directly.

What is Meta Muse?

Muse is Meta's personal AI agent, launched in the US on September 8, 2026. It can send emails, book travel, fill out forms and complete checkouts, with a free tier and paid plans at $20 and $100 a month. Meta says it doesn't currently share Muse conversations with its ad systems.

What is delegated demand?

It's demand that a person expresses but an AI agent researches, filters and sometimes transacts on their behalf. It changes marketing from optimizing for human attention to making sure agents can access, trust and act on information about your brand.

How does Cloudflare's September 15 change affect B2B websites?

From September 15, 2026, Cloudflare blocks Training and Agent bots by default on ad-supported pages for new domains, new customers and free-plan users. If your site runs ads or recently moved onto Cloudflare, buyers' AI agents may be blocked without marketing knowing. Check your settings for each bot category and make the decision deliberately.

What are ChatGPT Sponsored Agents?

Announced on September 16, 2026, Sponsored Agents let users who see a relevant ad start a clearly labeled conversation with a business-sponsored agent inside ChatGPT. They're being tested with select US advertisers, alongside new AI creative tools and HubSpot and Shopify integrations.

What changed with Google AI Max in September 2026?

On September 1, Google auto-upgraded eligible search campaigns, including those using automatically created assets and campaign-level broad match, to AI Max. New Dynamic Search Ads campaigns can't be created, and existing ones will be migrated in February 2027.

Should CMOs cut entry-level marketing roles because of AI?

Gartner's guidance, which I agree with, is to redesign those roles rather than remove them. Junior marketers should get earlier exposure to higher-value work like evaluating AI output and connecting work to outcomes, because that's how future marketing leaders build judgment.

What should a CMO do first?

Run an agent journey audit in your category, find out your current bot access settings, and name one leader accountable for how machines evaluate your brand. Those three steps take weeks, not quarters, and they'll show you where the biggest gaps are.

SignalShift is a monthly executive briefing on the structural shifts reshaping marketing. If this edition was useful, share it with one marketing leader who should be thinking about delegated demand before their next planning cycle.