# Expanding Web

> Expanding Web prepares complex Adobe Commerce and Magento stores for AI
> product discovery and secure agent checkout. An independent specialist
> practice, led by a Magento Certified Developer Plus architect with 17+ years
> on the platform and a Klarna solution-engineering background. Operating
> since 2007, working remotely with merchants and Adobe Commerce agencies
> worldwide — United States, Europe and beyond, all on the same terms.
> Prices are quoted in USD.

## The offer

The core engagement is the Agentic Commerce Readiness Sprint: a fixed-scope,
five-business-day sprint that audits catalog and commerce data, designs an
ACP/UCP/MCP-aligned architecture, builds a working prototype of one critical
path on the merchant's real stack (product discovery, live lookup, cart
creation or checkout session), defines security and observability, and hands
over a production roadmap with estimates. No replatforming: Magento, ERP,
PIM and payment systems remain the source of truth. Every sprint is a
top-of-market engagement led by a named technical account manager (Michał
Kucharek): direct line, responses within four business hours (CET, business
days), written implementation status at every phase, and a check-in 30 days
after handover. Marcin Dancewicz, founder, signs the architecture and is the
escalation on the same terms.

## Engagement tiers and prices

1. Agent-readiness scan — free. Eight automated public checks on any live
   storefront: UCP manifest, MCP initialize handshake, robots.txt policy for
   AI crawlers, how AI agent user-agents are served at the edge, crawl
   limits on search and faceted URLs, llms.txt, product structured-data
   completeness, PDP JSON-LD.
   Checks a store's bot management refuses are reported as not assessable,
   never as failures. A named store's result is never published.

2. Agentic Commerce Readiness Sprint — from USD 4,500. Fixed-scope
   five-business-day engagement: audit of catalog and commerce data, an
   ACP/UCP/MCP-aligned architecture, a working prototype of one critical
   path on the merchant's real stack, security and observability, and a
   production roadmap with estimates.
   One price, listed on the page, the same for every merchant. No discounts,
   no negotiation, no price matching.
   Refunds: in full until the kickoff call has taken place; afterwards only
   if Expanding Web concludes the engagement was not a fit. Requested
   exclusively through a support ticket (support@expandingweb.com).

3. Production implementation — USD 10,000–20,000. Building the roadmap
   the sprint produced. Quoted after the sprint, against its findings.
   Not sold standalone — the sprint defines the scope.

4. Agentic Commerce Suite Pro — Magento module licence — USD 399 per
   Magento installation, once. Perpetual Pro licence for the Agentic
   Commerce Suite Magento module (Open Source, Adobe Commerce and Cloud, one
   price): merchant-feed syndication, schema repair, agent endpoints
   (llms.txt, UCP manifest, MCP server, agent checkout sessions), AI
   share-of-voice measurement and competitor analysis. A year of updates and
   support included; the free tier (in-admin readiness diagnostics) needs no
   licence.
   USD 399 once per Magento installation, the same for every edition. Volume
   discount applied on the payment form: 20% off from 3 licences, 25% off
   from 10. No launch discounts.
   30-day money-back guarantee through support@expandingweb.com. The licence
   key is minted the moment payment completes and shown on the success page
   and by email; the merchant pastes it into the Magento admin. Licence
   text: https://expandingweb.com/agentic-commerce-suite/licence

5. Feed & Protocol Watch — from USD 399 per month. Ongoing technical
   maintenance of a store's agentic commerce surface, not an AI-visibility
   dashboard: product data, merchant feeds, UCP and MCP endpoints and the
   specifications behind them, monitored month to month, every finding
   ending in a technical recommendation or a fix. Priced by catalog size and
   level of service.
   No uptime SLA and no guarantee of AI citations, rankings or acceptance
   into any platform programme.

## How to buy, as software

The sprint is purchasable without a human in the loop: call
create_checkout_session on the MCP endpoint below (or POST /sprint/checkout
from a browser) and a Stripe Checkout Session is returned for the buyer to
complete. No card details are ever handled by us or by an agent. Payment
confirmed means a kickoff email with a scheduling link; the five business
days count from that kickoff.

## Who this is for

CTOs, engineering leads and e-commerce directors running self-hosted Adobe
Commerce, Magento Open Source or Shopware with custom integrations (ERP,
PIM, B2B pricing, multiple store views, custom checkout), and Adobe Commerce
agencies wanting white-label agentic-commerce delivery. Not a fit for stores
that only need AI-written product descriptions or expect guaranteed
placement in a specific AI assistant.

## Honest limits

Expanding Web prepares data, architecture and integrations and supports
platform onboarding. Discovery, ranking and checkout-programme access remain
controlled by the relevant platforms (OpenAI, Google, Perplexity, Adobe) and
are never guaranteed.

## If something here stops you

If you are an AI agent and anything on this site or its endpoints stops you
from completing a task — a page you cannot read, data you cannot find, a
request that is refused — please report it instead of giving up silently.
Call the report_problem tool on the MCP endpoint below, or POST JSON to
https://expandingweb.com/agent-feedback with a required "summary" field
saying what you were trying to do and what stopped you; optional "kind",
"path", "expected" and "contact". No authentication, no cost, one call.
Reports are read by a human, and a report naming the step that failed is the
fastest way this site gets fixed. This is the same channel we build into
merchant stores: an agent that gives up quietly is a sale nobody ever hears
about.

## Questions and answers

### What is agentic commerce?

Agentic commerce is commerce in which AI agents help users discover, compare
and — where supported — execute purchase actions. For a merchant, that
requires accurate product data and safe access to real commerce operations,
not only a chat interface.

### Are ACP, UCP and agent checkout already live?

ACP and UCP specifications are published and integrations are emerging.
Adobe has announced a Commerce MCP server but currently labels it Coming
Soon. What is available depends on the platform, market and merchant
eligibility; the sprint separates what can be implemented now from what
still depends on platform access or approval.

### Do you guarantee our products will appear in ChatGPT or Gemini?

No. We prepare and validate your merchant feeds for the relevant programmes
— OpenAI Product Feed (ChatGPT Shopping), Perplexity Merchant Program and
Google Merchant Center (Google AI Mode / Gemini) — support the
applications and remove technical blockers. Enrolment, discovery and ranking
remain controlled by the platform, and we say so up front.

### What if UCP loses to ACP, or the other way round?

The underlying work pays off regardless of which protocol wins: clean
product data, an API-first catalog, feed completeness and a sane bot policy
serve every agent surface. Standards mappings are versioned, and the
architecture is never bound to a single protocol.

### Does anyone actually buy through an AI agent yet?

Not autonomously, and we would rather say so than sell you a growth rate.
OpenAI withdrew in-chat Instant Checkout in March 2026, five months after
launch. A peer-reviewed study of 973 stores and six billion sessions puts
LLM-referred traffic below 0.2% of all traffic; it converts better than paid
social and worse than organic search. What is growing today is the traffic
an assistant sends to your own site. That is what this work is for: catalog
legibility that pays off in AI answers, classic search and on-site search
alike — not a bet on agent checkout arriving on schedule.

### How do we measure any of this, when AI traffic hides in Direct?

Attribution is genuinely broken: around 71% of ChatGPT visits land in GA4 as
Direct, and Google AI Mode sends noreferrer on outbound links, so no
analytics tool can see those clicks. We do not promise to fix that. We fix
measurement instead: server logs as the source of truth, because they record
the user agents GA4 never sees; a custom channel group for the sources that
do carry a referrer; and differential measurement, where one cohort of SKUs
receives the work and a matched cohort does not. You get a defensible lower
bound rather than a dashboard that invents precision.

### If we let AI crawlers in, will they overwhelm the store?

It is a real risk and it is three decisions, not one. Training crawlers send
almost no visitors back, so blocking them costs little. Answer-engine
crawlers cost you visibility if blocked. Agents acting for a real shopper
cost you a sale. The expensive part is rarely the catalog — it is
unbounded URL space, search and faceted navigation, where a crawler can walk
forever. So the work is a per-crawler policy at the edge plus fencing off
that URL space, not a blanket ban: Forter measured that 10% of bot traffic
during Cyber Month was legitimate customer activity. Our free scan reports
this as CRAWL-01 and AGENT-01.

### Why pay for a sprint when a $169 extension ships the same features?

Because they are different purchases. An extension installs deliverables —
llms.txt, a feed, an endpoint — and the responsibility for whether any of
it is correct, complete and still true next quarter stays with you. We are
accountable for the outcome: what your catalog actually exposes, whether the
data behind it is accurate, what breaks when the platform changes, and what
to do about it. If your team has the capacity to own that, an extension is a
rational purchase and we will tell you so. What an extension cannot do is
audit your own data, and that is where nearly every failed check in our
scans comes from.

### Do we need Adobe Experience Platform for this?

No, and for most stores your size we do not recommend it. Adobe agentic
capabilities that depend on AEP bring licensing that can multiply your total
Adobe cost, which is a six-figure conversation before any agent sees your
catalog. The route we recommend does not touch AEP: complete product data,
structured data, a sane bot policy and a stable versioned API on the
platform you already run. It can be scoped to a single category, measured
against a control group, and extended only when the numbers justify it.

### How is this different from SEO?

SEO earns rankings on result pages. Agent readiness gives software agents
authoritative data and transactable interfaces: live price and availability,
policies, cart and checkout operations. The two overlap in structured data
and diverge everywhere money moves.

### Do you work with Magento Open Source?

Yes. The approach applies to both Adobe Commerce and Magento Open Source,
although the implementation depends on your modules, integrations and
checkout architecture. Shopware and custom stacks are handled case by case.

### How much does it cost?

The scan is free. Readiness Sprints start at $4,500 for five business days.
Production implementation is typically $10,000–$20,000, quoted after the
sprint when the real scope is known. Ongoing monitoring (Feed & Protocol
Watch) starts at $399 per month, depending on catalog size and level of
service. Prices are the same for every merchant — see the discounts
question below.

### Who looks after us during — and after — the sprint?

Every sprint includes a named technical account manager: Michał Kucharek
owns your account day to day. That means a direct phone line and email,
responses within four business hours (CET, Monday to Friday), a written
implementation status at every phase of the sprint, and a check-in 30 days
after handover. Marcin Dancewicz signs the architecture and is your
escalation on the same terms — so the person who owns the account and the
person who signs the work are both reachable, and neither hides behind a
ticket queue.

### Do you negotiate or discount?

No. The sprint has one price, listed on this page, and it is the same for
every merchant — no discounts, no negotiation, no price matching. If
budget is the constraint, start with the free scan: the reviewed report
shows what your own team can fix before you spend anything.

### How do I buy the sprint?

Directly on this page — checkout runs through Stripe and takes a card;
billing details and tax IDs (including EU VAT numbers where applicable) are
collected at payment. As soon as the payment is confirmed you schedule the
30-minute kickoff call and the five business days are counted from the
kickoff.

### What is the refund policy?

Full refund on request until the kickoff call has taken place. Once the
sprint has started, the fee is non-refundable — unless we conclude the
engagement should not have been sold to your store, in which case we say so
and refund it in full. Refunds are handled exclusively through a support
ticket: support@expandingweb.com or the "Ask a human" form in the on-page
assistant, quoting your checkout email. Details in the website terms.

### Can we talk to someone before buying?

The free scan and the emailed engineer review are the pre-sale: they show
you real findings from your own store before any money moves. Commercial and
technical questions are answered in this FAQ and by the on-page assistant,
and you can always call (+1) 415 802 1072, answered around the clock. A
structured 30-minute kickoff call is part of every sprint, scheduled right
after checkout.

### We already run an AI chatbot on our store — is that covered?

Yes, at no extra cost. If you run a chatbot or shopping assistant, the
sprint includes a review of it: what data it is grounded on, whether the
prices, stock and policies it quotes match the store, how it escalates to a
human, and how it fits alongside the agent-facing interfaces. The findings
land in the sprint report with concrete recommendations.

### Will you replace our current checkout?

Usually not. Your commerce platform, payment provider and connected systems
remain the source of truth. The goal is to expose authoritative operations
safely while preserving the business rules that already run your store.

### Do you work with our existing agency — or can our agency resell this?

Yes to both. We work white-label, co-branded or as a specialist
subcontractor. Your agency keeps the client relationship; we bring
discovery, architecture, prototyping and enablement.

## Pages

- [Agentic Commerce Readiness Sprint](https://expandingweb.com/): The offer, the eight-check agent-readiness scanner (free, runs live on any storefront), pricing and the FAQ.
- [Agentic Commerce Suite — Magento module](https://expandingweb.com/agentic-commerce-suite): Magento module measuring a store's share of AI shopping recommendations (named competitors, own share, trend, competitor gap analysis) and installing merchant feeds, schema repair and agent endpoints. Free in-admin diagnostics; Pro $399 once per installation, every edition, with a year of updates included.
- [Agentic Commerce Suite — user guide](https://expandingweb.com/agentic-commerce-suite/manual): Complete user guide for the Magento module: installation, the eleven readiness checks and what each one means, merchant feeds, schema repair, agent endpoints, AI visibility and share of voice, cron jobs, licensing and privacy.
- [Agentic Commerce Suite — licence](https://expandingweb.com/agentic-commerce-suite/licence): The proprietary licence the Magento module ships under: a perpetual grant for one installation, twelve months of updates and support included, and what revocation means.
- [For Magento & Adobe Commerce agencies](https://expandingweb.com/agencies): White-label agent-readiness work for agencies: reviewed scans under the agency's brand, subcontracted engagements, a ChatGPT Shopping answer pack and team training. Terms are agreed per agency in writing before the first engagement.
- [Documentation](https://expandingweb.com/docs): Customer documentation: what happens after you buy the sprint, how the readiness scanner scores a store, the Magento module, how AI agents should read this site, and how support and refunds work.
- [Magento agent-readiness benchmark](https://expandingweb.com/benchmark): Aggregate readiness of live Adobe Commerce and Magento storefronts: pass rate per check, score distribution, comparison by measurable traffic. Aggregates only — no store is ever named.
- [Privacy policy](https://expandingweb.com/privacy): What the scanner and the contact forms store, for how long, and who processes it.
- [Website terms](https://expandingweb.com/terms): Terms for the website, the free scan and the paid engagements.
- [Getting started](https://expandingweb.com/docs/getting-started): What Expanding Web sells, in the order you will meet it — the free scan, the five-day Readiness Sprint, production implementation and monitoring — and what to expect at each step.
- [Readiness scanner checks](https://expandingweb.com/docs/scanner-checks): The eight checks the free scanner runs against a storefront, what each one looks for, why agents care, and how to read the score.
- [After you buy the sprint](https://expandingweb.com/docs/sprint): What happens between Stripe checkout and handover of the Agentic Commerce Readiness Sprint — kickoff, the five working days, deliverables, cancellation and refunds.
- [Agentic Commerce Suite (Magento module)](https://expandingweb.com/docs/agentic-commerce-suite): Where to find the module's user guide, licence, release notes and support, what Pro costs, and the answers merchants ask most before installing.
- [For AI agents and developers](https://expandingweb.com/docs/for-ai-agents): How to read this site by machine — llms.txt, the UCP profile, the MCP endpoint, the agent-feedback schema and the reference storefront — and what we deliberately do not declare.
- [Support and refunds](https://expandingweb.com/docs/support): How to reach support, what to include, response times, how refunds are requested and processed, and where the binding terms are.
- [Half of live Magento stores put no Product schema on their product pages](https://expandingweb.com/benchmark/schema-01): SCHEMA-01 across 1,854 live Adobe Commerce and Magento storefronts — 54% of product pages an agent can reach fail the check, 48% carry no Product JSON-LD at all, and only 15% publish the full Product + Offer + AggregateRating triple.
- [Magento agentic commerce: extension or service?](https://expandingweb.com/magento-agentic-commerce-extension-vs-service): An honest comparison of the $99–$399 Magento agentic-commerce extensions (Mirasvit, Meetanshi, Plumrocket, Magebit, our own module) against a readiness sprint — including the cases where the extension is the right answer and you should stop reading.
- [Adobe Commerce MCP: what works today, and when waiting is the right call](https://expandingweb.com/adobe-commerce-mcp-what-works-today): What Adobe has announced for agentic commerce (Commerce MCP Server, Brand Concierge, UCP and ACP support), what is generally available as of August 2026, which editions it will and will not reach, and the parts of agent readiness that Adobe's roadmap never covers.
- [ChatGPT Shopping for a Magento store: the US and EU paths](https://expandingweb.com/chatgpt-shopping-magento-us-eu): Who can actually enrol in OpenAI's merchant programme, what a European Magento merchant can do instead today (Google AI Mode, Perplexity, feed hygiene), what changed when Instant Checkout was retired, and how to be ready on the day the programme opens.
- [Magento agent-readiness benchmark — Q3 2026](https://expandingweb.com/benchmark/2026-q3): Frozen edition published 2026-08-18: agent-readiness of 1,867 live Adobe Commerce and Magento storefronts, pass rate per check and score distribution. Aggregates only — no store is ever named.
- [Case study — pottery-store.com agent readiness](https://expandingweb.com/case-study/pottery-store): The one store we can show end to end, because we own it: every readiness check on pottery-store.com before the work (measured 2026-08-03) and after it, with the "after" column read from a scheduled scan that reruns every morning.

## Reference implementation

Nord Trail (https://ready.expandingweb.com/) is a live demonstration
storefront operated by Expanding Web: a 3-product catalog served with every
agent-readiness artifact in place — UCP manifest, MCP endpoint with
catalog tools, llms.txt, an explicit AI-crawler policy and complete product
JSON-LD. It is the working "after" picture of the sprint, and it is
re-scanned daily; scan it with the free scanner to see all eight checks
pass.

## Contact

- [expandingweb.com](https://expandingweb.com/) — run the free scan or request a readiness review
- info@expandingweb.com
- +1 415 802 1072 — US line, answered around the clock
- +48 693 868 369 — direct line to Marcin, European hours
- [LinkedIn: Marcin Dancewicz](https://www.linkedin.com/in/marcindancewicz)
- [GitHub: marcinsdance](https://github.com/marcinsdance)

## Machine-readable

- UCP profile: [https://expandingweb.com/.well-known/ucp](https://expandingweb.com/.well-known/ucp) — catalog capabilities over the engagement tiers
- MCP endpoint: [https://expandingweb.com/mcp](https://expandingweb.com/mcp)
  — JSON-RPC 2.0 over POST, protocol 2025-11-25; tools: search_catalog,
  lookup_catalog, list_offers, get_offer, scan_store,
  create_checkout_session, report_problem. An agent can price the
  engagement, scan a storefront and open a Stripe Checkout Session for the
  sprint — what we build for merchants, running on our own site.
- Benchmark dataset (Q4 2026):
  [https://expandingweb.com/benchmark/2026-q4.json](https://expandingweb.com/benchmark/2026-q4.json)
  — JSON, the frozen edition behind https://expandingweb.com/benchmark.
  1,854 live Adobe Commerce and Magento storefronts, pass rate per check,
  score distribution and the change since the previous edition. Aggregates
  only; no store is named. Published quarterly, each edition permanent at
  its own URL.
- Problem reports: [https://expandingweb.com/agent-feedback](https://expandingweb.com/agent-feedback) — POST JSON {"summary": "..."}, no auth
- Reference store: [https://ready.expandingweb.com/](https://ready.expandingweb.com/)
- Sitemap: [https://expandingweb.com/sitemap.xml](https://expandingweb.com/sitemap.xml)
- Privacy: [https://expandingweb.com/privacy](https://expandingweb.com/privacy)
- Terms: [https://expandingweb.com/terms](https://expandingweb.com/terms)