|
|
Weekly digest #012 · 25 Oct – 8 Nov 2026
Last week in AI-powered marketing automation & AI search
Google is monetising AI Mode fast, even as agents and answer engines make visibility harder to see and to measure.
In this issue: Top 3 · Community · New on MLforSEO · Research · Jobs
|
|
Hi,
The loudest story this week is money. Google now shows up to four ads per AI Mode response, matching a classic results page, while its own earnings call admitted those ads earn less. What stood out to me is the quieter story underneath: OpenAI's agents can be blocked from your site without leaving a trace in your logs, so you can lose visibility and never see it happen.
|
★ Everything below lands in our Slack community first, through the week, with the discussion around it. Join free → |
|
|
The top 3
If you only read three things
01 |
Google AI Mode Now With Up To Four Ads News · Search Engine Roundtable
Google is now showing up to four ads per AI Mode response, the same count as a traditional results page, which squeezes the space left for organic. Check where your branded and competitor queries show AI Mode ads, and revisit your bidding and content positioning before assuming old SERP rules apply.
|
02 |
OpenAI’s Agents Can Silently Block Your Site – & Your Logs Won’t Show It News · Search Engine Journal
OpenAI's Aura safety system can stop ChatGPT agents from reaching your site, and nothing shows up in your logs or analytics. Test your key pages through the agent yourself, and read the piece for how to spot a block and request a review.
|
03 |
How the Same 16,000 Users Behave in AI Search vs. Google Deep dive · iPullRank
iPullRank matched 16,310 users across Datos panels to compare how the same people behave on Google and in AI search over a month. It found 87% of Google searches follow traditional keyword patterns, so use it to start building tracking around prompts and intent clusters rather than keywords alone.
|
|
|
From the community
What our community shared and is talking about
|
|
New on MLforSEO
Fresh from our platform
Resource
Good internal linking starts with two questions: which pages are under-linked, and which pages belong together? This free Colab notebook answers both.
Resource
Most schema markup tools work one page at a time: paste a URL, read the errors, fix, repeat.
Resource
Publishing at volume raises two questions every week: have the search engines heard about my new and updated pages, and has Google actually indexed them?
Resource
Before I write anything competitive, I want to know who ranks, what kind of sites they are, which topics and phrases keep showing up, and how the results feel.
Plus 175 new or updated glossary terms, from Transfer Learning to Rich Snippets. Browse the glossary →
|
|
Deep dives
Worth saving for later
Google AI Mode: All You Need to Know · AhrefsAI Mode is now Google's primary search interface for over 1B users—practitioners need to understand how it reshapes optimization strategy.
Google Gets The Thank-You · Search Engine JournalReframes how we measure search's actual influence on decisions, not just traffic attribution.
|
|
Research, decoded
New papers, and what to do with them
|
Toolflation: Seller Agents Distort Tool Markets Through Spontaneous Generative Engine Optimization What they found. When seller agents optimise listings to be picked by buyer agents, they spontaneously inflate capability claims ('toolflation'). Nothing stops the cascade, and buyer task success falls from 89% to roughly 30%.
Why it matters. Expect agent-driven marketplaces to reward exaggeration at first, then break. Accurate, verifiable product claims may become your edge. This is a simulation using real Amazon and Instagram data, not a live market.
|
|
Evolving Semantic Indexes for Generative Retrieval What they found. An LLM-driven evolutionary search improved how semantic identifiers are built and how collisions are handled in generative retrieval, lifting recall@1 from 55.8% to 59.3% on a Q&A benchmark.
Why it matters. Mostly a signal that generative retrieval is maturing, not something to act on yet. It is a lab benchmark result with no direct marketer action.
|
|
|
Headlines
The rest of the week in AI search
|
|
Hiring
8 new remote roles in this week's drop
Marketing, SEO and growth roles with real AI, automation or AI-search work in them, hand-picked and posted in our #jobs channel.
See the roles in #jobs
Not in the community yet? Join free, it takes a minute.
|
|
 New course on the academy
The AI-powered Entity Audit
A complete, repeatable entity audit you can run on any site. Built on the 5Cs of entity representation and demonstrated end to end on a real, unedited site.
Explore the course →
Already an Academy member? Take 30% off your next course or bundle with code Community30
|
|
|
Happy learning! ✨
Lazarina |
|
|
|
|
