Fasterize SEO Recommendations

The Fasterize SEO Recommendations node applies advanced SEO recipes (CSS targeting, variables, datasets, CSR) to a Fasterize-accelerated site through the SEO Recommendation API.

What does the Fasterize SEO Recommendations node do?

The Fasterize SEO Recommendations node pushes advanced SEO recipes to the Fasterize SEO Recommendation API (/v2/configs/{id}/apps/seo-recommendation). Unlike Bulk SEO, which applies one rule per URL, this node sends recipe-based transformations with CSS selector targeting, variable substitution, dataset lookups, and optional client-side rendering (CSR). Each recommendation declares the URLs it applies to and the recipes that mutate the HTML at the edge.

Common use cases:

• Injecting structured data (schema.org, OpenGraph) by selector across page templates without redeploying the origin.
• Rewriting headings, anchors, or canonical tags driven by a per-URL dataset (titles, descriptions, internal links).
• Running content-aware SEO experiments that depend on CSR to mutate dynamic HTML.
• Rolling out template-level fixes (breadcrumbs, hreflang, robots) to a list of URLs in a single deploy.

Quick setup

Configuration parameters

Required fields

What does the node output?

The node returns the raw response data from the Fasterize SEO Recommendation service. On success the response describes the recipes that were registered for the targeted URLs; on failure it returns a structured error dict that you can branch on.

Error shape when the payload is missing or empty:

{ "error": "No recommendations provided" }

Error shape when the JSON is malformed:

{ "error": "Invalid JSON format for recommendations" }

Error shape when no URL field is found:

{ "error": "Recommendations must include 'url' or 'urls' field or 'target.request_url.in' structure" }

Error shape when the service call itself fails:

{ "error": "Failed to execute Fasterize SEO Recommendation" }

Usage examples

Example 1: Inject schema.org markup with a recipe

You want every product page in a list to receive a Product schema.org block, generated by an upstream LLM and injected through Fasterize without touching the CMS.

Configuration:

• Fasterize integration: Production
• Recommendations (JSON): wired from the LLM output

Input payload (simplified format):

[
  {
    "name": "product-schema-org",
    "description": "Inject Product JSON-LD on product pages",
    "enabled": true,
    "csr": false,
    "datasets": [],
    "variables": {},
    "recipes": [
      {
        "type": "html-target",
        "selector": "head",
        "position": "beforeend",
        "value": "<script type=\"application/ld+json\">{...}</script>"
      }
    ],
    "urls": [
      "https://example.com/product/widget-a",
      "https://example.com/product/widget-b"
    ]
  }
]

The node wraps the payload as needed and POSTs it to /v2/configs/{id}/apps/seo-recommendation.

Example 2: Forward a full Fasterize API export

You exported the live recommendations from another Fasterize environment and want to replay them in a target environment.

Configuration:

• Recommendations (JSON):

{
  "installed": true,
  "settings": {
    "parameters": {
      "recommendations": [
        {
          "name": "blog-canonical-fix",
          "description": "Force canonical on blog category pages",
          "enabled": true,
          "csr": false,
          "datasets": [],
          "variables": {},
          "recipes": [
            { "type": "canonical", "value": "{{absolute_url}}" }
          ],
          "target": {
            "request_url": {
              "in": [
                "https://example.com/blog/category/news",
                "https://example.com/blog/category/guides"
              ]
            }
          }
        }
      ]
    }
  }
}

The node detects the settings.parameters.recommendations shape, extracts each recipe, and flattens target.request_url.in into the urls array before sending.

Common issues

The node returns: No recommendations provided

Cause: The recommendations input was empty, missing, or resolved to {}.

Solution: Verify the upstream connection feeding Recommendations. If the producer is an LLM, place a Find and Replace node before this one to strip Markdown fences (backticks) so the parsed JSON is not empty.

The node returns: Invalid JSON format for recommendations

Cause: The string passed to the node could not be parsed as JSON — typically a stray prefix, suffix, or unescaped character left by an LLM.

Solution: Pipe the LLM output through a Find and Replace or HTML Cleaner first, or wrap the generation step in a JSON-mode prompt so the model returns strictly valid JSON.

The node returns a missing url or urls field error

Cause: At least one recommendation in the payload has no targeting information — neither url, urls, nor target.request_url.in is present.

Solution: Ensure every recommendation declares the URLs it applies to. When converting from a Fasterize export, keep the target.request_url.in array intact; the node will flatten it for you.

Recipes are accepted but the page is unchanged in the browser

Cause: Fasterize caches HTML at the edge, so cached responses may still hide the new recipes until the TTL expires.

Solution: Trigger a Fasterize cache invalidation on the affected URLs after the call, or wait for the configured TTL before validating in the browser. For DOM mutations that depend on the live page, set csr: true on the recommendation so the recipe runs on the client.

Best practices and pitfalls

How does it fit into a workflow?

Fasterize SEO Recommendations is typically the last step in a recipe-generation pipeline that mixes structured data, dataset lookups, and LLM-drafted content.

graph LR
    Source[URL list] --> LLM[LLM drafts recipes]
    LLM --> FR[Find and Replace cleans JSON]
    FR --> Reco[Fasterize SEO Recommendations]
    Reco --> Notify[Email Sender notifies the team]

Related nodes