Chisos teardown: fourteen sizes gone, and Add to Cart still wants a 7

I opened the Chisos No. 2 in Magic Black, picked EE, and fourteen of the fifteen sizes were crossed out. Size 7 still had one pair. Add to Cart was waiting on a size most wide-footed shoppers do not wear.

Chisos sells hand-lasted cowboy boots online, with a showroom at 3507 S 1st St in Austin. The fit page says the boots match a Brannock device, D is standard, EE is extra wide on select styles, and size exchanges are free in the lower 48. Their help center says they make small batches, not thousands of pairs a month. I went through chisos.com on November 2, 2026 and took screenshots as I went.

What Chisos does well

The homepage sells the boot, not a lecture. The hero I captured quotes a customer on comfort, then "Hand-lasted. Resoleable. Comfortable out of the box," with a Shop Boots button. The red bar at the top already says Free Exchanges and Texan Owned. Call or text 512-866-7222 sits in the header. Below the fold are real reviews and a 2,247 five star reviews button (screenshot taken Nov 2, 2026).

The product page already has the hard parts of a boot PDP. On the No. 2 at $595, the buy column shows color swatches, D and EE, a size grid from 7 to 15, Free Size Exchanges, and a Size Guide for a Great Fit link, with Add to Cart in red (screenshot taken Nov 2, 2026). Judge.me on that page reports 850 reviews. For a cowboy boot you cannot try on at home unless you live in Austin, that is the right pile of facts.

The fit guide is a real fit guide. How to find your boot size starts with Brannock, then D versus EE, then special cases: wide foot with no EE, high instep, different-sized feet, men's to ladies' conversion. There is a video with chapters. The card on that page says talk to us by phone, SMS, or email. Most boot sites bury this in a PDF.

When a whole color is gone, they already have a waitlist button. On the No. 7 in Whiskey, every size was crossed out, the button read Sold Out, and a red Notify me when available sat under it (screenshot taken Nov 2, 2026). AMP Back in Stock is loaded on the storefront. The out of stock help article even explains the waitlist order: oldest signup first, a window to buy, then the next group.

What Chisos does badly

The size you actually need is not selectable. On the No. 2 in Magic Black EE, only size 7 was live. It showed Low stock, 1 items left. The rest of the grid was disabled (screenshot taken Nov 2, 2026). I could not click 11 to ask for a restock. The help article says click the size you need and hit the notification button. The theme does not let you do that while another size is in stock. Add to Cart stays pointed at the leftover 7.

I fetched /products.json?limit=250 the same day. The catalog was 38 products and 595 variants. 152 variants were unavailable. On boot styles alone, 120 of 408 variants were out. EE exists on one product: the No. 2. Magic Black EE 11 is out. Brushed Brown EE 11 is in. Magic Black D 11 is in. The page can see that. The shopper cannot.

Sold-out colors still talk about a width they do not sell. The No. 7 in Whiskey had an empty width box, the line "Most men wear D width. Choose EE for wide," and a full sold-out grid (screenshot taken Nov 2, 2026). That style only has D. Other No. 7 colors still had stock: Magic Black 5 sizes, Brushed Brown 10, Roughout 10, from the same feed. The notify button is for Whiskey, not "this size in Brushed Brown."

Search does not know what wide means. A search for wide returned 40 results. The first row was Belt, Cowhide, Belt, Ostrich, then the No. 1 and the No. 2 (screenshot taken Nov 2, 2026). The No. 1 is D only. The only EE boot in the feed is the No. 2. Someone typing "wide" is doing the store's merchandising job for it.

The contact page is still the Shopify placeholder. Contact us opens with "Use this text to share information about your brand with your customers" (screenshot taken Nov 2, 2026). The header already has a phone number. The footer already has howdy@chisos.com. The page a worried buyer actually opens does not.

Fit help is written. It is not on the picker. Help Scout lists 25 Boot Sizing articles, including D versus EE: EE is roughly two thirds of a size larger, and 11.5 D is close to 11 EE once broken in. Chatbase is loaded as a Shopify app block. None of that sits next to the crossed-out 11.

An eComFuel episode describes inventory shortages, cutting ads, and introducing pre-orders. I am not repeating its figures here. The live site is enough: the waitlist copy is in Help Scout, the notify button is on sold-out colors, and the size you need is still a dead cell.

3 quick wins

1. Let people select a sold-out size, then give them a next step. What I saw: Magic Black EE on the No. 2, fourteen sizes crossed out, Add to Cart on the leftover 7 (screenshot above, Nov 2, 2026). The fix: make those cells selectable. If 11 EE is out, open a short panel from the same inventory the page already has: notify me for Magic Black 11 EE (AMP Back in Stock is already loaded), buy Brushed Brown 11 EE (in stock in /products.json that day), or try Magic Black 11 D. Why it matters: a crossed-out size is currently a bounce. Their own help article already promised a notification button on the size you need.

2. Put the fit rules on the width you are looking at. What I saw: the No. 7 in Whiskey still says "Choose EE for wide" on a D-only product, with an empty width box (screenshot above, Nov 2, 2026). The fix: if EE is not on this style, say so, and surface the half-size-up line they already wrote. If EE is on this style and that cell is out, point to D or to the color that still has EE. Why it matters: width is the expensive question on a $595 boot with free exchanges. Answering it after the box arrives is the exchange queue.

3. Draft sizing and exchange tickets before a human touches them. What I saw: Help Scout's out of stock article and a Boot Sizing category with 25 articles. The contact page is a blank form. Chatbase is in the theme. The fix: classify inbound mail and chat against those articles and the live variant feed. Size too tight, wants EE, wants a restock, wants an exchange label. A human still sends the reply. Why it matters: they already wrote the answers. The cost is a person reading the same three questions every day.

Managed AI agents

These are jobs, not a chatbot on every page. Each one reads a store they already have, does one thing, and stops for a person.

Fit and support. Reads the Help Scout sizing articles, the live variant list (size, width, available), the fit page, and Judge.me reviews on that product. Drafts an answer to "I wear a 10.5 EE in other boots" or "this is tight across the instep." A human in Help Scout approves the send. Chatbase can stay as the public door. It should not send a size recommendation on its own.

Merchandising. Reads /products.json or the Shopify inventory API on a schedule. Flags a color with zero available sizes (No. 7 Whiskey that day), a width with one leftover pair (Magic Black EE 7), and a search query like "wide" that is ranking belts. Drafts a badge, a hide, or a "this size is in Brushed Brown" note. A merchandiser publishes it.

Product content. Writes the restock and pre-order sentence for a sold-out size from inventory and the waitlist rules they already published. Pushes a draft into Klaviyo or Postscript, which are both on the storefront. A person approves the send. No new list vendor.

Ops, returns, inventory notes. Reads exchange tickets and the returns portal. Labels them: wrong length, needs EE, instep, scuffed sole, wants a refund. Drafts the label and the suggested size. A human still creates the Shopify exchange. The eComFuel conversation is about inventory and pre-orders as an operations problem. This is that problem on the ticket, not a new storefront.

Build options I would actually use, from docs I opened:

• Claude Agent SDK runs the agent loop in a process you operate, with tools, MCP, and permissions so some tools auto-run and some need approval. Claude Managed Agents is the hosted version: you define the agent, an environment, and a session, and Anthropic runs the harness in a cloud or self-hosted sandbox.
• OpenAI's Agents SDK is agents, tools, guardrails, and a human-in-the-loop hook inside your app. The Agents guide also describes the Agents API (OpenAI hosts the harness) and the Responses API if you want to own the loop yourself.
• Gemini Enterprise Agent Platform, which Google documents as the evolution of Vertex AI Agent Builder, has Agent Studio for a low-code canvas, the Agent Development Kit for code, RAG Engine for private docs (the Help Scout articles), and Agent Gateway for policy on tool calls.

I would start with one agent, fit replies, with a human approve step. Shopify already holds the variants. Help Scout already holds the copy.

Before and after

The left column is the live No. 2 buy column on November 2, 2026, Magic Black EE. The right column is a mockup. It is not a Chisos page. Size 11 is selectable even though it is out. The panel uses stock I pulled from /products.json that day and the half-size note from their D versus EE article.

Technology detected

I fetched chisos.com on November 2, 2026, including the homepage, the No. 2 and No. 7 product URLs, search, the fit and contact pages, Help Scout, /cart.js, and /products.json?limit=250. I listed only vendors I could see in response headers, loaded scripts, app blocks, cookies, or those payloads. A name in a comment was not enough unless a script also loaded. I did not talk to Chatbase, so I will not judge its answers.

Engineer's take

Impulse on Shopify is why this page can have a size grid, a width toggle, and a notify button without a separate app. AMP Back in Stock, Judge.me, Help Scout, Klaviyo, and Postscript are the right vendors if you are going to buy them.

The friction is that sold-out sizes are disabled cells. AMP cannot attach to 11 EE while 7 EE is in stock. Variant availability is in the product payload. The Help Scout article already describes click the size, then notify. Wiring those together is theme work, not a replatform.

What I would watch: this page already loads GTM, Klaviyo, Postscript, Chatbase, Help Scout, Judge.me, AMP, StarApps, Kutoku, and Blockify. That is a lot of JavaScript next to a size picker that still cannot be clicked. I would keep new embeds off the product page until a crossed-out size can open notify, another color, or the half-size note.

Architect's take

I would not replatform this. The structural move is to treat size, width, availability, and the waitlist as one record the product page can read. Shopify already holds the variants. Impulse already renders the grid. AMP already sends the alert. Help Scout already wrote D versus EE. Chatbase is already on the domain. The missing path is from "I wear an 11 EE" into that record, instead of Add to Cart on a leftover 7.

That is an AI layer on top of a store that already knows how to make a boot and already knows it will run out. The question is not whether Chisos has a waitlist. It is whether the assistant can see the same facts the theme already has, at the size you wanted, before you bounce.

If you run a commerce brand and the expensive moment is a size-and-width grid that keeps saying no, it is what my AI Opportunity Teardown is for. For $2,000 I spend a week on your site and data, and you get a written one-pager and a 15-minute Loom walking through where AI would actually move a restock, a fit reply, or an exchange, and where it would not.