OpenAI + SerpAPI: scam risk scores from one URL

You find a new website, and that little voice kicks in. The domain looks “fine,” the prices look “too good,” and now you’re juggling ten tabs trying to prove a negative.

Procurement managers feel this when vetting suppliers. Marketing leads run into it when a partner offers “exclusive inventory.” And founders do it too, usually late at night. This scam risk scoring automation turns one submitted URL into a clear 1 to 10 risk score and a structured summary you can actually act on.

Below you’ll see how the workflow evaluates a site from multiple angles, what you get back, and what you need to run it reliably inside n8n.

How This Automation Works

The full n8n workflow, from trigger to final output:

n8n Workflow Template: OpenAI + SerpAPI: scam risk scores from one URL

Click to explore

flowchart LR

    subgraph sg0["On form submission Flow"]
        direction LR
        n0["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/merge.svg' width='40' height='40' /></div><br/>Merge1"]
        n1@{ icon: "mdi:cog", form: "rounded", label: "Aggregate1", pos: "b", h: 48 }
        n2@{ icon: "mdi:brain", form: "rounded", label: "OpenAI Chat Model", pos: "b", h: 48 }
        n3@{ icon: "mdi:wrench", form: "rounded", label: "SerpAPI", pos: "b", h: 48 }
        n4@{ icon: "mdi:wrench", form: "rounded", label: "SerpAPI3", pos: "b", h: 48 }
        n5@{ icon: "mdi:wrench", form: "rounded", label: "SerpAPI4", pos: "b", h: 48 }
        n6@{ icon: "mdi:brain", form: "rounded", label: "OpenAI Chat Model2", pos: "b", h: 48 }
        n7@{ icon: "mdi:wrench", form: "rounded", label: "SerpAPI2", pos: "b", h: 48 }
        n8@{ icon: "mdi:brain", form: "rounded", label: "OpenAI Chat Model3", pos: "b", h: 48 }
        n9@{ icon: "mdi:brain", form: "rounded", label: "OpenAI Chat Model4", pos: "b", h: 48 }
        n10@{ icon: "mdi:brain", form: "rounded", label: "OpenAI Chat Model5", pos: "b", h: 48 }
        n11["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/form.svg' width='40' height='40' /></div><br/>On form submission"]
        n12@{ icon: "mdi:robot", form: "rounded", label: "Domain & Technical Details", pos: "b", h: 48 }
        n13@{ icon: "mdi:robot", form: "rounded", label: "Search Engine Signals", pos: "b", h: 48 }
        n14@{ icon: "mdi:robot", form: "rounded", label: "Product & Pricing Patterns", pos: "b", h: 48 }
        n15@{ icon: "mdi:robot", form: "rounded", label: "Content Analysis", pos: "b", h: 48 }
        n16@{ icon: "mdi:robot", form: "rounded", label: "Evaluator", pos: "b", h: 48 }
        n0 --> n1
        n3 -.-> n12
        n7 -.-> n13
        n4 -.-> n14
        n5 -.-> n15
        n1 --> n16
        n15 --> n0
        n2 -.-> n16
        n11 --> n12
        n11 --> n13
        n11 --> n14
        n11 --> n15
        n6 -.-> n12
        n8 -.-> n13
        n9 -.-> n14
        n10 -.-> n15
        n13 --> n0
        n12 --> n0
        n14 --> n0
    end

    %% Styling
    classDef trigger fill:#e8f5e9,stroke:#388e3c,stroke-width:2px
    classDef ai fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
    classDef aiModel fill:#e8eaf6,stroke:#3f51b5,stroke-width:2px
    classDef decision fill:#fff8e1,stroke:#f9a825,stroke-width:2px
    classDef database fill:#fce4ec,stroke:#c2185b,stroke-width:2px
    classDef api fill:#fff3e0,stroke:#e65100,stroke-width:2px
    classDef code fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
    classDef disabled stroke-dasharray: 5 5,opacity: 0.5
    class n11 trigger
    class n12,n13,n14,n15,n16 ai
    class n2,n6,n8,n9,n10 aiModel
    class n3,n4,n5,n7 ai
    classDef customIcon fill:none,stroke:none
    class n0,n11 customIcon

The Problem: Website trust checks are slow and inconsistent

Manual scam checks are messy because you’re doing investigative work with “whatever you remember to look at.” One day you check domain age and SSL, the next day you skim Reddit threads, and half the time you forget to verify basic business info on the site. It’s not just time. It’s mental load, context switching, and the nagging worry that you missed the one signal that mattered. And when this happens during vendor onboarding or a rushed purchase, people make decisions based on vibes instead of evidence. Honestly, that’s how bad sites slip through.

The friction compounds. Here’s where it breaks down in real teams.

Domain and technical checks get skipped because they feel “too technical,” even though they’re often the fastest red flags.
Search results are reviewed casually, so scam reports and forum warnings are easy to miss.
Pricing and product patterns are judged subjectively, which means inconsistent decisions across the team.
Findings live in DMs or browser bookmarks, so there’s no repeatable audit trail later.

The Solution: Multi-agent scam scoring from a single URL

This workflow starts with a simple form submission: you paste in a URL you want to evaluate. From there, four specialized AI agents run in parallel, each looking at a different “slice” of trust. One agent focuses on domain and technical signals (things like domain age and SSL). Another checks search engine signals, including public scam reports and community discussions. A third looks for suspicious product and pricing patterns, and the fourth evaluates the on-site content quality and legitimacy markers such as business details and claims. Finally, a fifth agent consolidates the findings, assigns a scam likelihood score from 1 to 10, and returns a structured summary plus a clear disclaimer for context.

The workflow begins when your URL hits the n8n form trigger. SerpAPI is used as the research layer for the agents, while OpenAI models do the reasoning and summarization. Once the agents’ outputs are merged and aggregated, the Risk Scoring Agent produces the final “decision-ready” report.

What You Get: Automation vs. Results

What This Workflow Automates

Results You’ll Get

Collects a URL via an n8n form trigger and standardizes the input.
Runs four independent research-and-analysis passes using OpenAI agents connected to SerpAPI.
Merges and aggregates all agent findings into one consistent payload.
Generates a final 1 to 10 scam risk score with a structured narrative summary.

A clear risk score (1 to 10) you can paste into a vendor decision.
Fewer “open ten tabs” moments, so reviews stay fast even when you’re busy.
More consistent judgments across your team because the same dimensions are checked every time.
A readable breakdown of why the score is high or low, not just a number.
Cleaner documentation for approvals, audits, and “why did we choose them?” follow-ups.

Example: What This Looks Like

Say you vet 10 new sites a week for suppliers, affiliates, or one-off tools. A manual check is usually 20 minutes per site: a couple searches, a quick scan for scam reports, and a gut-check on pricing and content. That’s about 3 hours weekly. With this workflow, you submit the URL in under a minute, then wait a few minutes for the agents to finish and return a single structured summary. You still make the final call, but the research grind is gone.

What You’ll Need

n8n instance (try n8n Cloud free)
Self-hosting option if you prefer (Hostinger works well)
OpenAI for the multi-agent analysis and scoring
SerpAPI to pull search signals and mentions
OpenAI API key (get it from platform.openai.com/api-keys)

Skill level: Intermediate. You’ll connect API keys, test a few sample URLs, and adjust prompts if you want a specific output format.

Don’t want to set this up yourself? Talk to an automation expert (free 15-minute consultation).

How It Works

A URL gets submitted via a form. The workflow is triggered the moment someone enters a website link, which means this can be used by non-technical teammates without touching n8n.

Four parallel reviews run at the same time. Each OpenAI agent uses SerpAPI to gather public signals, then summarizes what it found from its angle (technical domain clues, search reputation, pricing authenticity, and content quality).

The workflow consolidates the evidence. n8n merges the outputs into one aggregated bundle so the final scoring agent isn’t guessing or re-researching; it’s evaluating consistent inputs.

A scoring agent produces the final report. You get a 1–10 scam likelihood score and a structured explanation that’s easy to paste into a decision log, email, or spreadsheet.

You can easily modify the scoring rubric to fit your risk tolerance based on your needs. See the full implementation guide below for customization options.

Step-by-Step Implementation Guide

Step 1: Configure the Form Trigger

Set up the intake form that captures the website URL to evaluate.

Add and open Form Entry Trigger.
Set Form Title to Website URL.
Under Form Fields, ensure a single field label is Website URL.
Set Form Description to https://www.scam.com/ as a usage example.

Step 2: Connect AI Review Agents and Tools (Parallel Branches)

Four review agents run simultaneously to analyze different risk signals from the submitted URL.

Connect Form Entry Trigger to the four agents so they run in parallel: Domain Tech Review, Search Signal Review, Pricing Pattern Review, and Content Quality Review.
In each of the four agent nodes, set Text to {{ $json['Website URL'] }}.
Ensure each agent keeps Prompt Type set to define and preserves its system message content for its specific analysis focus.
Attach language models to each agent: Chat Model Alpha → Domain Tech Review, Chat Model Beta → Search Signal Review, Chat Model Gamma → Pricing Pattern Review, Chat Model Delta → Content Quality Review.
Credential Required: Connect your openAiApi credentials on Chat Model Alpha, Chat Model Beta, Chat Model Gamma, and Chat Model Delta (these provide the language model for the four agents).
Attach the search tools to the agents: Search Tool A → Domain Tech Review, Search Tool B → Search Signal Review, Search Tool C → Pricing Pattern Review, Search Tool D → Content Quality Review.
Credential Required: Connect your serpApi credentials. Search Tool A, Search Tool B, Search Tool C, and Search Tool D are AI tools for the agents—ensure credentials are applied via the parent agent connections.

⚠️ Common Pitfall: The four agents must run in parallel from Form Entry Trigger. If one branch is not connected, Combine Streams will not receive all inputs and downstream scoring may be incomplete.

Step 3: Merge and Aggregate Review Outputs

Combine the four review streams into a single aggregated payload for final scoring.

Configure Combine Streams with Number of Inputs set to 5.
Connect the four review agents into Combine Streams so each outputs to a separate input index.
Connect Combine Streams to Aggregate Outputs.
In Aggregate Outputs, set Fields to Aggregate with five entries, each using Field to Aggregate = output and Output Field Name values Output 1, Output 2, Output 3, Output 4, Output 5.

Step 4: Configure the Final Risk Scoring Agent

Use the aggregated content to produce a final risk score and structured summary.

Open Risk Scoring Agent and set Text to {{ $json['Output 1'][0] }}{{ $json['Output 1'][1] }}{{ $json['Output 1'][2] }}{{ $json['Output 1'][3] }}.
Ensure Prompt Type is define and Has Output Parser is enabled.
Connect Aggregate Outputs to Risk Scoring Agent.
Attach Main Chat Model as the language model for Risk Scoring Agent.
Credential Required: Connect your openAiApi credentials on Main Chat Model and set Model to gpt-4o-mini.

Step 5: Test and Activate Your Workflow

Validate the end-to-end execution before turning the workflow on.

Click Execute Workflow and submit a sample URL via Form Entry Trigger.
Verify that Form Entry Trigger outputs to Domain Tech Review, Search Signal Review, Pricing Pattern Review, and Content Quality Review in parallel.
Confirm Aggregate Outputs produces fields named Output 1 through Output 5.
Check Risk Scoring Agent for a final response that includes a disclaimer, a 1–10 risk score, and sectioned summaries.
When satisfied, toggle the workflow to Active for production use.

🔒

Unlock Full Step-by-Step Guide

Get the complete implementation guide + downloadable template

Common Gotchas

OpenAI credentials can expire or be pasted into the wrong node. If things break, check the OpenAI Chat Model credentials used by each agent first.
If you’re using Wait nodes or external rendering, processing times vary. Bump up the wait duration if downstream nodes fail on empty responses.
SerpAPI can hit monthly limits faster than you expect because one run may use several searches. When results look “thin,” check your SerpAPI dashboard and confirm the API key still has quota.

Frequently Asked Questions

How long does it take to set up this scam risk scoring automation?

About 30 minutes once you have both API keys.

Do I need coding skills to automate scam risk scoring?

No. You connect OpenAI and SerpAPI in n8n, then test with a few URLs.

Is n8n free to use for this scam risk scoring workflow?

Yes. n8n has a free self-hosted option and a free trial on n8n Cloud. Cloud plans start at $20/month for higher volume. You’ll also need to factor in OpenAI API costs (this workflow is about $0.01 per run) and your SerpAPI usage.

Where can I host n8n to run this automation?

Two options: n8n Cloud (managed, easiest setup) or self-hosting on a VPS. For self-hosting, Hostinger VPS is affordable and handles n8n well. Self-hosting gives you unlimited executions but requires basic server management.

Can I customize this scam risk scoring workflow for stricter vendor approvals?

Yes, and you’ll probably want to. You can adjust the Risk Scoring Agent’s rubric so a missing business address or a very new domain pushes the score higher. Common tweaks include changing the 1–10 definitions, adding a “block automatically at 8+” rule with an If node, and saving results to Google Sheets or Airtable for an approval trail.

Why is my SerpAPI connection failing in this workflow?

Usually it’s a missing or exhausted SerpAPI key. Open the SerpAPI account dashboard to confirm you still have searches left, then re-paste the key into the SerpAPI tool nodes used by the agents. If the key is fine, check whether your queries are being blocked by account limits or you’re sending unusual locations/languages that require extra settings. One more thing: if the workflow runs many sites in a short time, rate limits can show up as intermittent failures.

How many URLs can this scam risk scoring automation handle?

It depends mostly on your n8n plan and your SerpAPI quota. On n8n Cloud, your monthly execution limit is based on your plan, and each URL check is one execution. If you self-host, there’s no execution cap, but your server still needs enough CPU and memory for bursts. Practically, most teams start with a few dozen checks a week and scale from there once they see their SerpAPI usage per run.

Is this scam risk scoring automation better than using Zapier or Make?

Often, yes, because this workflow relies on multi-step logic and multiple AI agents working in parallel. n8n handles branching and merges cleanly, and self-hosting can make high-volume screening more affordable. Zapier and Make can still work if you only want a simple “URL in, summary out” flow, but they get awkward once you add parallel research, aggregation, and scoring. If you need this to feed a review queue, notify Gmail, and log to Airtable, n8n tends to feel less restrictive. Talk to an automation expert if you want help picking the right stack.

After you set this up once, every new URL check becomes a repeatable decision instead of a mini investigation. The workflow does the digging. You keep control of the call.

OpenAI + SerpAPI: scam risk scores from one URL

How This Automation Works

n8n Workflow Template: OpenAI + SerpAPI: scam risk scores from one URL

The Problem: Website trust checks are slow and inconsistent