AI Marketing · Published October 2026

What I'm Reading: October 2026

Six reads that say the boring parts matter more now.

Part of the AI Marketing guide

Every good thing I read in October said the same thing: AI makes the fundamentals more important, not less.

October was loud. New models every week. New agents every day. Somebody on X is always one prompt away from replacing your whole team.

The reads that stuck with me were the quiet ones. Google’s own timelines. An analytics veteran calling out the “AI CMO.” A media buyer’s spend data. An engineer at a trading firm arguing for more discipline, not less. They’re the six I’d hand anyone building an AI marketing plan right now.


1. Google finally told us how long SEO takes

Source: Gary Illyes’ timeline data, via Search Engine Roundtable and Jake Ward’s one-page graphic

Google shared typical timelines for crawling, indexing and serving. A few that matter:

  • Discovering a new page: about 20 hours
  • Recrawling a page Google already knows: about 30 days
  • Indexing once crawled: about 1.5 hours
  • Updating a title or snippet: 1 to 2 days
  • Recovering from a core update: 3 to 6 months

Why it matters: These steps stack. Google can index your change in 90 minutes and still not see it for a month, because it hasn’t come back to crawl.

My take: Print this and tape it to the wall of every SEO meeting. A lot of the “SEO isn’t working” conversations I have are really “Google hasn’t recrawled yet” conversations. Check the last crawl date in Search Console before you change anything else.


2. Avinash on the “AI CMO”

Source: Avinash Kaushik, The Marketing Analytics Intersect #530 (subscribe here)

Avinash took aim at vendors selling “agentic” marketing and influencers declaring the human CMO obsolete. His read: the loudest sellers show limited understanding of marketing, limited understanding of AI, and confidence that comes from not knowing either.

Why it matters: Marketing isn’t just advertising. An agent can run ad ops. It can’t decide your positioning, your pricing, or which market to walk away from.

My take: I use agents every day and I’m with him. When someone pitches you an “AI CMO,” ask which strategic decision it made last quarter.


3. Q4 media costs are already off the forecast

Source: The Media Buyer by Northbeam (weekly platform spend email)

Northbeam’s October 23 issue reported Meta CPMs up about 25% in October, before November even started. Their read is that Amazon pulled roughly $2B of ad spend forward out of Black Friday.

Why it matters: If you built your November budget on last year’s CPM curve, it’s already wrong.

My take: Re-baseline your Q4 plan this week, not after the first bad Monday report. If demand keeps moving earlier, Black Friday stops being a week and turns into a six-week stretch. Your creative refresh schedule becomes the bottleneck, not your budget.


4. More automation won’t save your marketing

Source: The Boring Marketer newsletter

The point: three agent products launched in Oct (Muse, Dot, Duet) and all three led with the same feature. Permission. One asks a human when it isn’t confident. One makes you declare what it can do alone, what needs sign-off, and what it never touches.

Why it matters: The race moved from “what can the agent do” to “when should it stop and ask.”

My take: Write the permission list before you build the automation. For every workflow, mark it green (the agent ships it), yellow (the agent drafts, a human signs), or red (never). That flow is worth more than any tool you’ll buy this quarter.


5. Building reliable software in the age of agents

Source: Ron Minsky (Jane Street), Bug Bash 2026 talk (talk page)

It’s an engineering talk, and it’s the best marketing talk I watched this month. Minsky walks through how Jane Street brings AI into its development process. His argument: agents should make people more capable and better informed, not push them out. So type systems, tests and code review matter more now, not less.

Why it matters: Swap “code” for “campaigns” and it still holds. Agents need guardrails they can check themselves against. Humans need to keep understanding what shipped.

My take: Your brand guidelines, your QA checklist and your approval step are your type system and your tests. If your team skipped them because “the AI handles it,” that’s the first thing I’d put back.


6. Cognitive Mercantilism

Source: Venkatesh Rao, Contraptions

Rao argues most AI panic is an old idea in new clothes. Mercantilists believed wealth was fixed and had to be hoarded. Rao says a lot of AI fear treats intelligence the same way. It’s scarce, it belongs to credentialed experts, and everyone else is low-value cognitive labor. His counter is farming. Productivity freed most people from producing food, and new kinds of work followed. AI could free most people from producing intelligence just to get by.

One line I keep coming back to: “The returns on attention rise, and with them the incentive to pay attention.”

Why it matters: The scarce thing in marketing isn’t producing content. It’s earning attention.

My take: It’s the most optimistic thing I read all month, and the most useful. Stop measuring your team by how much they produce. Start measuring what your customers choose to pay attention to.


One thing to do this week

Pick one workflow you’ve handed to AI. Write down where it has to stop and ask a human.

That stop point is the part of your marketing no agent can own. It’s also the part you get paid for.

Next issue: the 10 AI releases that mattered to marketers from Oct to Nov.


Sources