SEO in the oversupply era: what actually moves the needle now

Statistics on a laptop

Your own site still drives AI citations more than third-party pages

Grow and Convert ran the numbers on what actually gets a brand mentioned in AI-generated answers. Third-party pages make up 85% of AI citations overall, but they are not the strongest predictor of brand mentions. Of the answers that named a specific brand, 87% cited that brand’s own site. When a brand’s own site was cited, it appeared in the answer 90% of the time in Google AI Overviews and 73% of the time in ChatGPT.

The operator takeaway: investing in your own content infrastructure is still the highest-leverage GEO move, not chasing third-party placements.

Winning in oversupplied markets means breaking a convention, not optimizing harder

A Substack essay by Nick Hac makes the case that most markets are crowded because building is cheap and distribution is easy. Supply keeps growing. Demand stays flat. Buyers stick with familiar options. Customer research often just confirms what competitors already know.

The pattern that works: find what the industry treats as non-negotiable and break it where doing so creates new value for buyers. Growth comes from redefining the category, not from outcompeting on the same dimensions everyone else is fighting over.

Person holding pencil near laptop computer

Build LinkedIn retargeting audiences for $0.50 per member

A circulating LinkedIn tactic: run 60 to 70 second videos using a Website Visits objective, then write a headline designed to discourage clicks. You only pay when someone clicks. Anyone who watches at least 50% of the video gets added to your retargeting audience automatically. The reported cost is around $0.50 per retargeting member, compared to $1 to $2 per click on standard engagement campaigns.

SEO has limited short-term influence on AI agent recommendations

Research by Joe Shirey tested whether SEO could push a vendor into an AI agent’s recommendations. The findings are sobering. AI agents recommend products mostly from training data, not from live search results. On advisory questions, agents rarely search unless prompted, and that behavior depends on the prompt and tools configured by whoever runs the agent.

Even when every search was forced and every query was secretly rewritten to favor one vendor, that vendor was selected only 17.1% of the time, up from a baseline of 9.2%. SEO can still shape future training data over time, but its influence on near-term AI recommendations is limited.

Apple’s iOS 27 may block hundreds of ad tech companies from iOS

According to AdExchanger, Apple’s iOS 27 is reportedly planning a significant expansion of WebKit’s ad tech blocking. A new remote list would cover hundreds of CDPs, DMPs, DSPs, data sellers, and identity providers, going well beyond companies like The Trade Desk, LiveRamp, and ID5. Apple can now add or remove companies from the list without an iOS update, meaning vendors may not know their status until traffic drops.

️ Google Ads adds A/B testing to Performance Max asset groups

Google is rolling out asset group A/B testing inside Performance Max campaigns. Advertisers can test new headlines, descriptions, images, or videos against existing creative without immediately replacing it. Tests can isolate specific changes, such as messaging or UGC, to measure impact independently. One constraint: no other asset changes are allowed while a test is running.

Stay on top of AI & Automation with BizStack Newsletter
BizStack  —  Entrepreneur’s Business Stack
Logo