Every morning at 7:45 AM, a Telegram message arrives with three sections: keywords that moved more than two positions overnight, a competitor that appeared in the top 3 for a term we own, and two pages where organic CTR dropped below their 30-day average. The system that sends it costs roughly $12/month in API calls and replaces a rank tracking subscription that used to cost $149.
This guide walks through building that system with OpenClaw. It covers SERP checking via browser automation, historical rank storage, competitor position monitoring, Google Search Console integration, content performance alerts, and the daily briefing that ties everything together. You need a working OpenClaw installation (follow our setup guide if you do not have one) and a Google Search Console property with verified access.
How the Monitoring System Fits Together
Before building individual skills, it helps to see how the pieces connect. The SEO monitoring system is four skills and one heartbeat section, wired together through workspace files.
The daily cycle looks like this:
- 3 AM cron: The rank-check skill opens a headless browser, queries Google for each tracked keyword, and writes the results to
workspace/seo/rankings/[date].json. - 3:30 AM cron: The competitor-check skill runs the same queries but extracts where competitor domains appear in the results.
- 4 AM cron: The gsc-pull skill calls the Google Search Console API, pulls yesterday’s performance data, and saves it to
workspace/seo/gsc/[date].json. - 4:15 AM cron: The seo-alerts skill reads all three data files, compares them against thresholds and historical baselines, and writes any triggered alerts to
workspace/seo/alerts/[date].json. - 7:45 AM heartbeat: The morning briefing section reads the alerts file and sends a summary to Telegram.
Each skill runs in an isolated session to avoid context pollution. The workspace files are the glue.
SERP Checking via Browser Automation
Most guides tell you to install a pre-built rank tracking skill and move on. That works until you need to customize what gets tracked, handle rate limiting, or debug why your positions look wrong. Building the skill from scratch takes 20 minutes and gives you full control.
The Rank Check Skill
Create skills/seo-rank-check.md in your OpenClaw workspace:
## SEO Rank Check Skill
When given a list of keywords (from workspace/seo/tracked-keywords.md):
1. For each keyword:
- Open a headless browser session using Puppeteer
- Navigate to Google with the query
- Set the location to the target market (default: United States)
- Set the device type (default: desktop)
- Extract the top 20 organic results: position, URL, title, snippet
- Check if any result has a featured snippet or AI Overview
- Close the browser session
- Wait 8-15 seconds before the next query (randomized delay)
2. For each keyword, find our domain in the results and record:
- Current position (1-20, or "not found")
- URL that ranks
- Whether we appear in featured snippet or AI Overview
3. Save all results to workspace/seo/rankings/[YYYY-MM-DD].json
4. Compare today's positions against yesterday's file:
- Flag any keyword that moved more than 2 positions in either direction
- Flag any keyword where we dropped out of the top 20
- Flag any keyword where we entered the top 20
Format the comparison as a changes array in the output file.
Why Browser Automation Instead of an API
Paid rank tracking tools use their own infrastructure to check positions. OpenClaw uses a headless browser on your machine, which means the results reflect what a real user sees from your geographic location. The trade-off: you cannot check thousands of keywords because Google will rate-limit you. For most sites tracking 30-100 priority keywords, this works fine.
The randomized delay between queries matters. Without it, Google serves CAPTCHAs after 10-15 rapid queries. Randomized delays of 8-15 seconds typically allow checking 50 keywords in about 12 minutes without hitting blocks. If you track more than 100 keywords, split them across two cron runs spaced an hour apart.
The Tracked Keywords File
Create workspace/seo/tracked-keywords.md:
## Tracked Keywords
| Keyword | Target URL | Priority |
|---------|-----------|----------|
| openclaw setup guide | /guides/openclaw-setup-guide | high |
| ai agent for seo | /guides/ai-agents-seo | high |
| openclaw vs n8n | /guides/openclaw-vs-n8n | medium |
| how to automate seo reporting | /guides/openclaw-seo-monitoring | medium |
## Settings
- Target location: United States
- Device: desktop
- Check frequency: daily
- Domain to track: sfailabs.com
This file is the single source of truth. Add or remove keywords here, and the rank check skill picks up the changes on the next run.
Scheduling the Rank Check
Wire it to a nightly cron:
openclaw cron add --name "seo-rank-check" \
--cron "0 3 * * *" \
--session isolated \
--message "Run the seo-rank-check skill using the keywords in workspace/seo/tracked-keywords.md. Save results and flag any position changes greater than 2."
Storing and Tracking Rank History
Raw position data becomes useful when you can see trends. The rank check skill saves daily JSON files, but you also need a way to compare across weeks and months.
Workspace File Structure
workspace/seo/
tracked-keywords.md # Keywords to monitor
rankings/
2026-03-28.json # Daily position snapshots
2026-03-29.json
2026-03-30.json
2026-03-31.json
gsc/
2026-03-28.json # Daily GSC performance data
...
competitors/
2026-03-31.json # Competitor position snapshots
alerts/
2026-03-31.json # Triggered alerts for the day
Trend Detection
The alerts skill reads the last 7 days of ranking files to detect trends rather than reacting to single-day fluctuations. A keyword dropping from position 4 to position 6 for one day is noise. The same keyword declining from 4 to 5 to 6 to 7 over four consecutive days is a trend worth investigating.
The trend detection logic in the alerts skill:
## Trend Rules
- **Declining trend**: Position worsened in 3 of the last 5 days → flag as "declining"
- **Improving trend**: Position improved in 3 of the last 5 days → flag as "improving"
- **Lost ranking**: Was in top 20, now not found for 2 consecutive days → flag as "lost"
- **New ranking**: Was not in top 20, now appears for 2 consecutive days → flag as "new entry"
This prevents the alert system from crying wolf on normal SERP volatility. Google results fluctuate daily, and reacting to every single-day movement is a fast path to alert fatigue.
Competitor Position Monitoring
Knowing your own rankings is half the picture. Knowing when a competitor moves into a position you hold, or takes a featured snippet you own, is what lets you react before traffic drops.
The Competitor Check Skill
Create skills/seo-competitor-check.md:
## SEO Competitor Check Skill
When given the tracked keywords list and a competitor domains list:
1. Read workspace/seo/competitors.md for the list of competitor domains
2. Read today's ranking data from workspace/seo/rankings/[today].json
(the rank check skill already captured the full top 20 for each keyword)
3. For each keyword, extract where each competitor domain appears:
- Position (1-20 or "not found")
- URL that ranks
- Whether they hold a featured snippet or AI Overview
4. Compare against yesterday's competitor data:
- Flag any competitor that moved into the top 3 for a keyword we track
- Flag any competitor that overtook our position
- Flag any competitor that gained a featured snippet we do not have
5. Save to workspace/seo/competitors/[YYYY-MM-DD].json
The Competitors File
Create workspace/seo/competitors.md:
## Competitor Domains
| Domain | Label |
|--------|-------|
| competitor-one.com | Competitor A |
| competitor-two.io | Competitor B |
| another-rival.com | Competitor C |
## Monitoring Rules
- Track up to 5 competitor domains per keyword
- Alert when any competitor enters the top 3
- Alert when a competitor overtakes our position
The competitor check runs after the rank check because it reuses the same SERP data already captured. No additional Google queries needed. This is a data extraction step, not a separate scraping run.
openclaw cron add --name "seo-competitor-check" \
--cron "30 3 * * *" \
--session isolated \
--message "Run the seo-competitor-check skill. Read today's ranking data and extract competitor positions. Compare against yesterday and flag movements."
Google Search Console Integration
Browser-based rank checking tells you where you appear in search results. Google Search Console tells you what happens after that: how many impressions you get, how many clicks, and what your CTR looks like. Combining both data sources gives you the full picture.
Connecting GSC to OpenClaw
OpenClaw accesses the Google Search Console API through a service account. The setup:
- Create a service account in Google Cloud Console
- Download the JSON key file
- Add the service account email as a user in your Search Console property
- Store the key file path in your
.env:
GSC_SERVICE_ACCOUNT_KEY=/path/to/service-account-key.json
GSC_PROPERTY=https://sfailabs.com
The GSC Pull Skill
Create skills/seo-gsc-pull.md:
## GSC Data Pull Skill
When triggered:
1. Authenticate with Google Search Console API using the service account key
2. Pull the Search Analytics report for yesterday:
- Dimensions: query, page, date
- Metrics: clicks, impressions, CTR, position
- Filter to the tracked keywords from workspace/seo/tracked-keywords.md
3. Also pull the full site-level data (no keyword filter) for:
- Total clicks and impressions
- Pages with the largest impression changes vs. 7-day average
- New queries that appeared for the first time
4. Save to workspace/seo/gsc/[YYYY-MM-DD].json
5. Calculate these derived metrics:
- CTR change vs. 7-day rolling average for each tracked keyword
- Impression change vs. 7-day rolling average
- Pages where position improved but CTR declined (title/snippet issue)
- Pages where impressions dropped more than 20% vs. 7-day average
What GSC Data Reveals That Rank Tracking Misses
A keyword can hold position 4 and still lose traffic if the CTR drops. This happens when a competitor above you adds a rich snippet, when Google inserts an AI Overview that pushes organic results below the fold, or when your meta description stops matching search intent.
The GSC pull skill catches these patterns because it tracks impressions and clicks independently from position. A page ranking at position 3 with 500 impressions and a 2% CTR is underperforming. The same page at position 3 with a 6% CTR is healthy. Without GSC data, both look identical in a rank tracker.
Schedule it after the competitor check:
openclaw cron add --name "seo-gsc-pull" \
--cron "0 4 * * *" \
--session isolated \
--message "Run the seo-gsc-pull skill. Pull yesterday's Search Console data, calculate derived metrics, and save to workspace."
Content Performance Alerts
The alert system is the layer that turns raw data into decisions. It reads the ranking data, competitor data, and GSC data, then applies threshold rules to determine what deserves your attention.
The SEO Alerts Skill
Create skills/seo-alerts.md:
## SEO Alerts Skill
When triggered:
1. Read today's files:
- workspace/seo/rankings/[today].json
- workspace/seo/competitors/[today].json
- workspace/seo/gsc/[today].json
2. Apply these alert rules:
### Ranking Alerts
- Position drop greater than 3 in a single day → URGENT
- Position declining for 3+ consecutive days → WARNING
- Dropped out of top 20 → URGENT
- New top-10 ranking achieved → INFO
### Competitor Alerts
- Competitor entered top 3 for our tracked keyword → WARNING
- Competitor overtook our position → WARNING
- Competitor gained featured snippet we lack → INFO
### GSC Performance Alerts
- CTR dropped more than 30% vs. 7-day average → WARNING
- Impressions dropped more than 25% vs. 7-day average → WARNING
- Page with position improvement but CTR decline → INFO (possible title/snippet issue)
- New query generating more than 50 impressions → INFO (content opportunity)
3. Save triggered alerts to workspace/seo/alerts/[YYYY-MM-DD].json
4. Each alert includes: type, severity, keyword, current value, previous value, recommended action
Alert Severity Levels
Three levels keep the system manageable:
- URGENT: Requires same-day investigation. Position losses of 3+ spots, dropped out of top 20.
- WARNING: Review within 48 hours. Competitor movements, CTR declines, multi-day trends.
- INFO: Awareness items for the weekly review. New rankings, opportunities, minor shifts.
The temptation is to make everything urgent. Resist it. Running aggressive thresholds can produce 15-20 alerts per day. After tuning the thresholds to the levels above, expect the daily count to drop to 3-5 actionable alerts. The signal-to-noise ratio matters more than comprehensiveness.
openclaw cron add --name "seo-alerts" \
--cron "15 4 * * *" \
--session isolated \
--message "Run the seo-alerts skill. Read today's ranking, competitor, and GSC data. Apply alert rules and save triggered alerts."
The Daily SEO Briefing
The briefing is the interface between the monitoring system and your attention. Without it, data sits in workspace files that nobody reads. With it, you get a 30-second scan each morning that tells you whether anything needs action.
Configuring the Briefing in heartbeat.md
Add this section to your OpenClaw heartbeat.md:
## SEO Morning Briefing
If the current time is between 07:30 and 08:00 in my timezone:
1. Read workspace/seo/alerts/[today].json
2. If no alerts exist, send a brief "All clear" message and stop
3. Group alerts by severity:
- URGENT alerts first (red flag emoji)
- WARNING alerts second
- INFO alerts last (only include if fewer than 5 total alerts)
4. For each URGENT and WARNING alert, include:
- The keyword and current position
- What changed and by how much
- One specific recommended action
5. Add a summary line:
- Total keywords tracked
- Keywords in top 3 / top 10 / top 20
- Net position change vs. yesterday
6. Send via Telegram to the SEO channel
What the Morning Briefing Looks Like
A typical Telegram message:
SEO Briefing — March 31, 2026
1 URGENT “openclaw setup guide” dropped from #3 to #7 overnight. New competitor page from edgedigital.net entered at #2. Action: Review their page, check if our content needs updating.
2 WARNINGS
- “ai agent seo” — competitor overtook us (#4 to their #3). Competitor: fennecseo.app
- /guides/openclaw-heartbeat-scheduling CTR fell 34% vs. 7-day avg (3.2% to 2.1%) while position held at #5. Possible snippet/title issue.
Summary: 42 keywords tracked. 6 in top 3, 18 in top 10. Net change: -3 positions across portfolio.
On quiet days, the message is three lines: “All clear. 42 keywords tracked. 6 in top 3, 18 in top 10. No changes greater than 2 positions.” Those are the good mornings.
What This System Costs
The honest breakdown for monitoring 50 keywords daily:
| Component | Model/Resource | Daily Cost | Monthly Cost |
|---|---|---|---|
| Rank check (50 queries) | Claude Sonnet 4.6 + Puppeteer | ~$0.15 | ~$4.50 |
| Competitor extraction | Claude Sonnet 4.6 | ~$0.03 | ~$0.90 |
| GSC data pull | Claude Sonnet 4.6 + GSC API | ~$0.04 | ~$1.20 |
| Alert processing | Claude Sonnet 4.6 | ~$0.02 | ~$0.60 |
| Daily briefing (heartbeat) | Claude Haiku | ~$0.02 | ~$0.60 |
| Total | ~$0.26 | ~$7.80 |
Scale to 100 keywords and the cost roughly doubles to $15/month. Compare that to SEMrush ($140/month for 500 keywords), Ahrefs ($129/month for 750 keywords), or even budget tools like SE Ranking ($65/month for 500 keywords).
The trade-off is accuracy and scale. Paid rank trackers check from multiple locations simultaneously, track thousands of keywords, and have purpose-built SERP APIs that avoid rate limiting. OpenClaw’s browser-based approach is accurate for your specific location but limited to the volume your machine can handle without triggering CAPTCHAs. For a team tracking 50-100 priority keywords, the economics are compelling. For an agency tracking 5,000 keywords across 50 clients, a paid tool still makes more sense.
Frequently Asked Questions
Can OpenClaw replace Ahrefs or SEMrush for rank tracking?
For position monitoring on 50-100 keywords, yes. OpenClaw checks actual Google results from your location using a headless browser and stores the data in your workspace. Where it falls short is backlink analysis, site audits at scale, and tracking thousands of keywords simultaneously. Those require the proprietary databases that paid tools maintain. Most teams use OpenClaw for daily monitoring of priority keywords and keep a paid tool for monthly deep-dive audits.
How accurate is browser-based rank tracking compared to paid tools?
Browser-based results match what a real user sees from your machine’s IP and location. Paid tools aggregate data from proxy networks across multiple locations, which can produce slightly different results. For single-location tracking, OpenClaw is often more accurate because it reflects your actual target market. The discrepancy grows for multi-location tracking, where paid tools have an infrastructure advantage.
How do I track rankings in different countries or cities?
Set the location parameter in your tracked-keywords file. OpenClaw’s browser automation can set the Google country domain (google.co.uk, google.de) and use geolocation headers. For city-level precision, you would need a proxy service like Bright Data or Oxylabs configured as a browser proxy. This adds $20-40/month depending on usage but gives you accurate local results.
What happens if Google blocks my rank checking?
Google serves CAPTCHAs after detecting automated queries. The randomized 8-15 second delay between checks prevents this for up to about 60 queries per run. If you hit blocks, increase the delay range to 15-25 seconds or split your keyword list across two cron runs separated by 2-3 hours. At 50 keywords with the default delay range, blocking is unlikely.
Can I track AI Overview appearances alongside regular rankings?
Yes. The rank check skill already extracts whether a keyword triggers an AI Overview and which URLs appear in it. Add a section to your alerts skill that flags when a competitor appears in an AI Overview for your tracked keyword, or when an AI Overview appears for a keyword where it previously did not exist. This is increasingly important as AI Overviews expand to more query types.
How much storage does the ranking data use?
Each daily JSON file for 50 keywords runs about 15-25 KB. A full year of daily checks produces roughly 7-9 MB of ranking data. GSC data files are similar in size. The storage is negligible. Keeping 90 days of daily data in the workspace and archiving older files to a workspace/seo/archive/ folder is a practical approach.
Key Takeaways
- An OpenClaw SEO monitoring system combines four skills (rank check, competitor check, GSC pull, alerts) and one heartbeat section (daily briefing). Each runs in an isolated session on a cron schedule, with workspace JSON files as the data layer.
- Browser-based rank tracking works well for 50-100 priority keywords with randomized delays of 8-15 seconds between queries. Above that volume, split across multiple cron runs or consider a paid tool for the long tail.
- Google Search Console data fills the gaps that rank tracking misses: CTR changes, impression drops, and the disconnect between position improvements and traffic declines. Always pair GSC data with SERP positions.
- Alert thresholds determine whether the system is useful or annoying. Start with conservative thresholds (3+ position drops for URGENT, 30%+ CTR decline for WARNING) and tighten them after two weeks of baseline data.
- The full system costs $8-15/month in API calls for 50-100 keywords, replacing $65-150/month in paid rank tracking subscriptions for the core monitoring use case.
Related Resources
- How to Set Up OpenClaw the Right Way: 10 Steps — prerequisite installation, model selection, and Telegram configuration
- OpenClaw Heartbeat Scheduling: Configure Proactive Agent Behavior — deep dive on intervals, cost control, and the briefing pattern used here
- OpenClaw Cron Jobs — detailed cron configuration for the scheduling used in this guide
- How to Use OpenClaw for Content Creation: Automated Writing Workflows — the content pipeline that benefits from SEO monitoring data
- OpenClaw Memory Configuration: Make Your Agent Remember Everything — persistent memory setup for long-running monitoring workflows
SFAI Labs
