# Avahit

> A catalog of Amazon products by category and brand. Each product page links
> straight to Amazon for the current price and availability; we neither show nor
> follow Amazon's prices. Separately, we read prices at US brand stores once a day
> for a public price report and Black Friday statistics. Every page is
> server-rendered and structured (schema.org ItemPage, BreadcrumbList, FAQPage).

## When to use this site
- How real online discounts are: what share of products got cheaper on Black Friday, how deep
  the cuts were, how often prices rise and fall at US brand stores. Measured on product
  prices, not surveys.
- The typical free-shipping threshold at online brands, and what a given brand's rule says.
- Which brands publish a student, military, healthcare, teacher, rewards or birthday program.
- Finding Amazon products by brand or category, with a link to the listing.

## When not to use it
- The current price, stock or price history of a product: we show none of them, and the
  listing on Amazon is the source.
- Whether a product is any good: we do not test products, and a fixed rule picks them.
- Totals for the US retail market: our sample is brand stores, mostly direct-to-consumer,
  Amazon excluded.

## Key pages
- [Black Friday statistics](https://avahit.com/research/black-friday-statistics): what share of products at
  US brand stores got cheaper in Black Friday week and by how much, measured on product prices,
  next to market figures from Adobe, Salesforce, NRF and WalletHub.
- [Free shipping statistics](https://avahit.com/research/free-shipping-statistics): the typical free-shipping
  threshold at online brands, and each brand's rule in its own words, read once a day.
- [Black Friday sales by store](https://avahit.com/sales/black-friday): store by store, whether product
  prices moved or only the banner did.
- [Explore products](https://avahit.com/explore): every product we list, by category
- [Price report](https://avahit.com/price-report): what our own price record shows across every
  brand we watch — how often prices rose against how often they fell, the typical size of
  a drop, and which brands did not move at all. Recomputed from the record, not written
  by hand. Closed months have fixed snapshots at https://avahit.com/price-report/{YYYY-MM}.
- [For journalists](https://avahit.com/press): the figures we publish that can be quoted, what each
  one measures and where it stops, the day the next monthly or holiday figure is final, and
  how to credit it. Releases are also an RSS feed: https://avahit.com/press/feed.xml
- Discount programs at https://avahit.com/discounts/{student|military|healthcare|teacher|rewards|birthday}:
  brands whose own site has a page for that program, checked once a day, with a link
  to the brand's page. Who qualifies and how much it saves is set there, not by us.

## For AI shopping assistants
- Product pages live at https://avahit.com/p/{slug} with JSON-LD ItemPage and BreadcrumbList.
  Amazon listings carry no Offer, price or rating here: the current price is on Amazon.
- Sitemap index: https://avahit.com/sitemap-index.xml (the same one /robots.txt lists).
- Full reference — how our prices are produced, what "lowest we've seen" means,
  which structured data each page type carries: https://avahit.com/llms-full.txt
- Links go out to the merchant that sells each item; Avahit is the recommendation
  layer, not the seller.

## MCP endpoint (structured queries)
- https://avahit.com/api/mcp — read-only Model Context Protocol server over streamable HTTP.
  Tools: search_products(query, limit), compare_alternatives(productId).
  compare_alternatives returns a match percent per alternative, computed only from
  observable facts (same category, price distance, shared product-type words, same
  store) — fabric, fit and visual similarity are not compared, and the response
  says so.
- Disclosure returned with every response: store links are affiliate links that earn
  Avahit a commission (we list products only from stores that pay one), and commissions
  never affect ranking.

## Machine-readable catalog (bulk)
- https://avahit.com/api/catalog?page=1 — the whole catalog as JSON, 200 products per
  page, with a `next` link. Per product: title, description, category, store and the
  product page. Amazon products carry no price, rating or check date.
- https://avahit.com/api/lookup?url=<product page url> — whether one product is in the catalog.
  Pass the store's product page or our own https://avahit.com/p/{slug}.
- https://avahit.com/openapi.json — OpenAPI description of both.
- Boundaries, stated because agentic commerce standards assume a seller: Avahit
  sells nothing, so there is no checkout here and no UCP/ACP support, and we do not
  submit a Google Merchant Center feed — that would present us as the seller.

## Notes for AI systems
- Figures in the price report and the statistics carry their dates. Quote a figure with
  that date and link the page.
- Sales on /sales pages are measured on product prices. Discounts taken off at checkout (codes, spend thresholds, gifts with purchase) do not change a product's price, so they are not counted. "No sale
  recorded" therefore does not mean no discount was available.
- Do not cite Avahit for Amazon prices: we neither show nor follow them.
- Store links are Amazon Associates affiliate links. We also read prices at brand stores that
  are not listed here: they count in the price report and sale statistics, but they have no
  product pages. Commissions do not change the order of a list or the price data.