Daily competitor signals compared with your company

Daily competitor signals compared with your company

An n8n workflow that looks at each competitor in a Google Sheet once a day. An AI agent gathers the past week's signals
through Explorium and finds the company's LinkedIn page with Tavily. Two Apify Actors add follower, staff and traffic
numbers, and a second agent compares all of it with your own company in a paragraph. Every run adds one row for each
competitor to the sheet.

What is inside

• workflow.json: the workflow as n8n exports it, fifteen nodes and no sticky notes. It holds no credential.
• README.md: these notes.

What we changed from the original copy

• The author's sheet and its three tab ids are gone. The three Google Sheets nodes ask for YOUR_SHEET_ID and find their
  tab by name: Competitor details, My company or Competitor analysis.
• The Authorization headers held Bearer <token>. They now read Bearer <YOUR_APIFY_TOKEN> in the two Apify nodes and
  Bearer <YOUR_TAVILY_KEY> in tavilySearch, so you can see which key goes where.
• The Explorium credential is gone from Explorium. Its address is unchanged.

Before you start

• An n8n instance, on n8n Cloud or your own server.
• An OpenAI API key that can use gpt-4.1-mini.
• An Explorium account with credit, and a token its MCP server accepts.
• A Tavily account and its API key.
• An Apify account and its API token, with credit for two Actors.
• A Google account, and a Google Sheet with three tabs (below).

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Make the sheet with the three tabs below, then copy its ID: the long part of its address between /d/ and /edit.
3. Add a Google Sheets OAuth2 credential. In Get row(s) in sheet, Get row(s) in sheet1 and Append row in sheet,
   select it and replace YOUR_SHEET_ID with the ID.
4. Add an OpenAI credential and select it in OpenAI Chat Model. The two agents and the parser all use it.
5. Open Explorium, add a Bearer Auth credential with your Explorium token and select it.
6. In tavilySearch, replace <YOUR_TAVILY_KEY> in the Authorization header with your key. The brackets go too, so the
   header reads Bearer, a space and the key.
7. In Linkedin Followers + Employee count and Website visits, replace <YOUR_APIFY_TOKEN> with your Apify token in
   the same way.
8. Open Schedule Trigger and pick a day and an hour. The file sets none, so it runs daily at midnight.
9. Put one competitor in Competitor details and press Execute workflow. Press Publish when the new row looks right.

Explorium is an MCP Client Tool of version 1, which only speaks SSE, pointed at https://mcp.explorium.ai/sse with
Bearer Auth. Explorium's own sign-in page, read on 2 November 2026, names https://mcp.explorium.ai/mcp over HTTP
Streamable, signed in with OAuth. Both addresses asked for sign-in when we called them without a token that day. If your
token does not get through, add a fresh MCP Client Tool node. Set its Endpoint to the /mcp address, its transport to
HTTP Streamable and its authentication to MCP OAuth2, then connect it to AI Agent as a tool and delete the old one.

The sheet

| Tab | Columns |
|---|---|
| Competitor details | Company name, Company domain, one competitor in each row |
| My company | Your own figures in the first row, under headers you choose, such as LinkedIn followers, staff and monthly visits. Only that row is read |
| Competitor analysis | Company, Website Traffic Growth, Content Output, Ad Activity, Product Launch Frequency, Partnerships Integrations Added, LinkedIn Followers, Employee Count, Press Mentions PREvents, Technology Stack Changes, Insights |

Spell the headers of the first and last tabs as above, since the nodes match them by name.

How a run goes

1. Schedule Trigger starts Get row(s) in sheet, which reads every row of Competitor details. Loop Over Items
   hands them on one at a time, and Edit Fields keeps the name and the domain.
2. AI Agent runs gpt-4.1-mini. Its prompt has it match the business in Explorium, call Explorium's enrichment tools,
   then search linkedin.com through tavilySearch for the company page. It writes the whole Tavily request body itself.
   Structured Output Parser holds its reply to seven fields: contentOutput, adActivity, productLaunches,
   partnershipsIntegrationsAdded, pressMentionsPREvents, technologyStackChanges and linkedin. The prompt asks for
   signals from the past week, and the agent retries when it fails.
3. Linkedin Followers + Employee count sends the LinkedIn address to Apify's LinkedIn Profile Scraper by anchor. The
   workflow reads follower_count and the second item of company_size from its answer.
4. Website visits sends the domain to Apify's Ahrefs Scraper by radeance, with only traffic switched on. The workflow
   reads organic from the fifth entry of website_traffic_history.
5. Get row(s) in sheet1 reads My company, and Aggregate gathers its rows into one list.
6. Insights, a second agent on the same model with no tools and no parser, gets the findings, the three numbers and
   your first My company row as JSON. It writes a paragraph of about 200 words.
7. Append row in sheet adds a row to Competitor analysis, then the loop takes the next competitor.

Both Actors answered when we looked them up on 2 November 2026. Each is called synchronously, and Apify ends such a call
after 300 seconds, according to its API reference.

Change the look

• How often: Schedule Trigger. The prompt asks for the past week, so a weekly run matches it.
• What the research looks for: the system message of AI Agent and the example in Structured Output Parser. Change
  both together, then the mapping in Append row in sheet.
• The comparison: the system message of Insights, which sets the length at about 200 words.
• Your baseline: the first row of My company. Use the same kinds of figures the competitors get, so the agent compares
  like with like.
• Model: OpenAI Chat Model, shared by both agents and the parser.

Good to know

• Each competitor costs two gpt-4.1-mini agent runs, plus a repair call when the parser rejects a reply. Explorium
  spends credits on its lookups, Tavily counts every search, and both Actors bill per result on Apify. Start with one
  competitor.
• OpenAI shuts down gpt-4.1-mini on 14 January 2027 and names gpt-6-luna as the replacement. Pick the new model
  in OpenAI Chat Model before that date.
• Explorium's sign-in page offers a free trial of 100 credits and says calls fail once the balance is spent.
• Website Traffic Growth gets one organic traffic figure, not a rate of growth. A young domain may have no fifth entry
  in its history, which leaves the column empty.
• Employee Count takes the second item of the Actor's company_size. Check in a first run what that item holds.
• Rows are appended and never updated, so the tab keeps a history and grows on every run. Clear old rows now and then.
• The prompts carry a few typing slips from the original copy, such as "tody". They do not change what the agents do, so
  they are left as they were.
• Competitor names and domains go to OpenAI, Explorium, Tavily and Apify. Your first My company row goes to OpenAI.
• The Tavily key and the Apify token sit in headers, so take them out before you share the workflow.