Bancadia
For affiliate network and institutional reviewers

Compliance & Disclosure

This page documents Bancadia's affiliate disclosure practices, product data sourcing methodology, how recommendations made by third-party AI agents are attributed, and the technical measures in place to prevent fraudulent bot clicks on affiliate links. It is intended for compliance reviewers at affiliate networks and financial institution partners evaluating Bancadia for approval.

Affiliate Disclosure & FTC Compliance

Bancadia earns an affiliate commission when a user clicks an application link and applies for a financial product. Those clicks reach us through two channels: directly from Bancadia's own pages, and from recommendations made by third-party AI agents querying Bancadia's registry. In accordance with FTC guidelines, the affiliate relationship is disclosed clearly on every page that displays a financial product, and the agent channel is covered separately below.

Standard disclosure text

“Bancadia may receive compensation when you click on product links. This compensation may impact how and where products appear on this site. Bancadia does not include all financial products or providers.”

Where disclosures appear

  • Product listing pages — disclosure rendered at the bottom of every category listing page (e.g., /products/business-checking).
  • Product detail pages — disclosure rendered at the bottom of every individual product page.
  • Comparison tables — disclosure rendered at the bottom of every comparison page (e.g., /compare/business-checking).
  • Agent-facing responses — every product returned over MCP carries its application_url as a Bancadia-controlled /go/ redirect, so an agent-driven referral passes through the same disclosed, logged path as a click from the website. See Agent-channel attribution below.

Each product record in the database carries its own affiliate_disclosure string, allowing per-product disclosure text to be customized to reflect any network-specific language requirements without changing site-wide defaults. This is the string rendered on the public pages listed above.

Data Sourcing & Accuracy

Why traditional bank data is difficult for AI to parse accurately

Financial institutions publish product information across multiple formats: marketing landing pages with rate callouts buried in dense HTML, fee schedule PDFs, footnote-heavy disclosure documents, and promotional banners with temporary rates. These formats are designed for human readers, not machine consumption. When an AI system attempts to extract a product's APY or fee structure from an unstructured web page, the result is often imprecise — the model may hallucinate a figure, cite a promotional rate as the standard rate, miss a conditional fee, or surface outdated data from a prior crawl. This is not a failure of the AI; it is a structural problem with the data source.

How Bancadia solves this

Today, Bancadia's team compiles product data directly from each institution's own published materials — official product pages, fee schedule PDFs, terms of service, and rate sheets. This data is then mapped into Bancadia's strict product schema: every attribute has a defined type and field name. APY is a decimal number. Monthly fee is a decimal number. Minimum opening deposit is a decimal number. Eligible entity types (LLC, S-corp, sole proprietor, etc.) are typed enumeration values. Insurance type (FDIC, NCUA, uninsured) is an enumeration. There is no free-text field for a core product attribute. As Bancadia establishes affiliate network partnerships, we plan to move to receiving the same structured data feeds institutions provide to their affiliate programs, supplementing this manual verification process.

This schema is what Bancadia serves to AI systems — via structured HTML with schema.org JSON-LD markup on public pages, and via the MCP API for agent queries. Because every field is typed and labeled, a language model reading a Bancadia product page or API response does not need to interpret ambiguous prose. It reads a structured record, the same way a developer reads a JSON object. For every field Bancadia publishes, this removes the extraction step where hallucinated figures originate.

Verification metadata

Every product listing displays two verification signals:

  • Last Verified date — the date on which the product data was last confirmed accurate against its source. Visible on every product detail page.
  • Verification source URL— a direct link to the authoritative source document (typically the institution's official product page) that confirmed the data. AI systems use this signal to assess data provenance and cite accordingly.

Only listings with an active status are publicly displayed or returned by the API. Products flagged as stale (data not refreshed within the configured threshold) are automatically surfaced for review before remaining visible.

Affiliate Link Integrity & Bot Fraud Prevention

Ensuring that affiliate clicks represent genuine human intent is a core requirement of operating as a compliant publisher. Bancadia implements three independent layers of protection. No single layer is relied upon exclusively; all three operate simultaneously.

1

Crawler exclusion via robots.txt

Bancadia's robots.txt disallows all user-agents from accessing the /go/ path — the path through which all affiliate redirects are served. This instructs all well-behaved crawlers (search engines, AI indexing bots) not to follow or index any affiliate redirect URL. This is the outermost layer: it prevents compliant bots from ever reaching the redirect endpoint.

2

Server-side bot detection on every redirect

Every request to an affiliate redirect URL is inspected server-side before any redirect is issued, and classified into one of two categories. Indexing and training crawlers — GPTBot, ClaudeBot, CCBot, Googlebot, Bingbot, DuckDuckBot, PerplexityBot and others on a maintained blocklist — are rejected with HTTP 403 Forbidden before the redirect fires, and no click is registered. This layer catches bots that ignore robots.txt directives. Live assistant fetches made on behalf of a human who is present in a chat session — ChatGPT-User, OAI-SearchBot, Perplexity-User — are allowed through, because these represent genuine user intent rather than automated crawling. They are recorded with a distinct ai_assistant traffic source so that assistant-referred clicks are separable from direct human traffic in reporting at any time.

3

JavaScript-only link rendering

Apply buttons on Bancadia are rendered without an href attribute in the HTML source. The navigation to the affiliate redirect URL is executed exclusively by client-side JavaScript on user interaction (click or keyboard activation). This means the affiliate redirect path is never present as a crawlable <a href> link in the page source. A crawler following links cannot reach the redirect endpoint because the link does not exist in the HTML. Additionally, raw affiliate tracker URLs are never written into page HTML, JSON-LD structured data, or API responses under any circumstances.

MCP API Compliance

Bancadia operates a Model Context Protocol (MCP) server that allows AI agents to query the product index directly. Compliance requirements are enforced at the API layer as well as the web layer.

  • Every product object returned by the MCP API is scoped to listings with an active status and carries the is_verified flag and last_modified timestamp, so an agent can assess freshness before acting on a record. Draft, paused, and scheduled listings are never returned.
  • The application_url field in API responses always uses the Bancadia-controlled redirect format (bancadia.com/go/[slug]). Raw affiliate tracker URLs are never included in any API response field.
  • API access is restricted to authenticated developers. Tokens are issued through the Bancadia developer portal, stored as one-way hashes, and can be revoked at any time. Unauthenticated requests receive an HTTP 401 response before any product data is returned.

Agent-Channel Attribution & Disclosure

Bancadia's primary distribution channel is not its own website. Third-party AI agents — bookkeeping copilots, treasury tools, entity-formation platforms — query the registry and present recommendations inside their own products, to their own users. This section documents how that channel stays attributable and disclosed, since it differs from a conventional publisher relationship.

  • Every application path is Bancadia-controlled. The application_url returned to an agent is always a bancadia.com/go/[slug] redirect, never a raw tracker URL. An agent cannot obtain, cache, or redistribute the underlying affiliate link, so every conversion traverses the logged, disclosed redirect.
  • Every click is recorded with its source. Each redirect writes a row capturing the listing, destination, referrer, user-agent, and a classified traffic source. Agent-originated, assistant-originated, and direct human traffic are distinguishable in reporting rather than pooled.
  • Every agent is identifiable. Registry access requires a token issued through the developer portal and stored as a one-way hash. Queries are attributable to the integration that made them, and access can be revoked at any time.
  • No paid placement, in either channel.Result ordering is determined by the querying agent's filter criteria. An institution cannot purchase position, and no ranking signal is sold. Where Bancadia soft-ranks results by business segment, non-matching but eligible listings are still returned rather than suppressed.

No affiliate commissions are being paid out to integrating agents today. When revenue sharing goes live, the disclosure obligations that attach to a recommendation made inside a partner's product will be addressed in the partner agreement governing that integration.

Contact

For questions about Bancadia's compliance practices, affiliate program participation, or product data accuracy, contact us at the appropriate address below.

Affiliate network inquiries

[email protected]

Data corrections

[email protected]

General

[email protected]