Brandon Lazovic

E-commerce SEO · Agentic commerce · Data

Search, shopping and AI agents all read your site as structured data.

I'm Brandon Lazovic, an e-commerce SEO consultant and data engineer. I write about how product feeds, crawlers and AI models decide what gets found and bought. Every piece measures something on a real site instead of describing how it is supposed to work.

Read the writing Search it

Where to start

The writing runs in four threads, in this order.

Crawl, Index, Rank comes first because the other three assume you already have it. To build from the ground up, start at the top. To skip ahead, jump to whichever thread names your problem.

  1. 04Start here

    Crawl, Index, Rank

    Three stages, in that order: a crawler fetches the page, an index stores it, a ranking system picks it. Most ranking problems are really crawl or index problems, and both are measurable on your own site.

    17 articles
  2. 02

    How Machines Read the Web

    Agents and crawlers read structure, not pixels. Semantic HTML, server-rendered schema, and embeddable item data decide what gets understood and surfaced.

    37 articles
  3. 01

    The Feed Is the Storefront

    The structured product feed is no longer a Shopping input; it is the surface AI rebuilds commerce around. Control is completeness, accuracy, and representation.

    24 articles
  4. 03

    The Model Is the Dependency

    Every model you build on is a dependency you do not control. It carries its own behavior, spends your context, and runs on the credentials you hand it.

    15 articles

Recent

Latest articles

Agent Skills well-known index: more than 70 platforms adopted the format, and in 36 days one scanner asked for my site's indexMore than 70 platforms support Agent Skills, but the standard has no website path. My site's well-known skills index drew one scanner in 36 days of logs.How Machines Read the WebChatGPT Free vs Plus: Free runs GPT-6 Luna and paid plans run Sol. In my test, asking Luna again moved its cited sites as much as switching modelsChatGPT Free runs GPT-6 Luna and Plus runs Sol. Asked the same questions, Luna's own repeat answers differed as much as Luna's and Sol's did.The Model Is the DependencyAgentic commerce statistics: AI agent purchases quadrupled since April, and about 3% of stores let agents check outAgentic commerce statistics: agent purchases quadrupled since April, BrightEdge says, yet about 3% of stores let agents pay. Retailers' rules differ.The Feed Is the StorefrontSitemap lastmod: my site sets it for every page, and it matched the last significant edit on 1 of 92 updated articlesGoogle uses sitemap lastmod if it is consistently accurate. My site sets it per page, and it matched the last significant edit on 1 of 92 edited articles.Crawl, Index, RankWebsites blocking AI agents: Yelp and two banks say they chose to, while Walmart told TechCrunch its blocks were not intentionalYelp and two banks block AI agents by policy. Walmart told TechCrunch its blocks weren't intentional, and Zillow said anti-scraping measures can catch agents.The Feed Is the Storefront

Daily

The Pulse

Short, dated notes on what moved in AI search and e-commerce, each one sourced to a first-party announcement.

Every entry in the Pulse

Also worth reading

Data Analysis Double Counting: Name the Method First

Double counting was one of five wrong first numbers I logged in my own data analysis. Each skipped the same step, and a hand-checked sample caught one of them.

Read the essay