Skip to main content

SERP Tracking System

Daily and on-demand search engine ranking monitoring with historical analytics.

Overview​

Track keyword rankings in Google search results for customer projects. Supports organic positions, local pack results, and paid ad positions.

Key Features:

  • Daily automated tracking (cron-based)
  • On-demand tracking (API-triggered)
  • Multi-location support
  • Device targeting (desktop/mobile)
  • Historical trend analysis
  • ClickHouse analytics storage
  • Task-based async mode for large batches

Architecture​

┌─────────────┐
│ Daily Cron  │ 10:00 UTC (2 AM PST)
└──────┬──────┘
       │
       ├─> Query D1: SELECT active keywords
       │
       ├─> For each keyword:
       │   └─> QUEUE.send({type: 'serp_tracking'})
       │
       └─> Queue Consumer (max_concurrency: 20)
           ├─> Call DataForSEO SERP API
           ├─> Parse results
           ├─> Save to D1: serp_runs, serp_positions
           └─> Queue to ClickHouse for analytics

Tracking Modes​

1. Daily Automated Tracking​

Scheduled via cron trigger at 10:00 UTC (2 AM PST).

Flow:

  1. Cron triggers lib/serp-tracking.handleDailySerpTracking
  2. Query D1 for active keywords (with project subscriptions)
  3. Enqueue serp_tracking jobs to main queue
  4. Queue consumer processes with DataForSEO
  5. Results stored in D1 + queued to ClickHouse

Database Query:

SELECT k.*, p.domain 
FROM keywords k
JOIN projects p ON k.project_id = p.id
JOIN customer_subscriptions cs ON p.id = cs.project_id
WHERE k.is_active = 1
  AND cs.subscription_type LIKE '%serp%'
  AND cs.is_active = 1
  AND (cs.expires_at IS NULL OR cs.expires_at > ?)

Configuration:

  • Modify cron schedule in wrangler.toml: 
    [triggers]
    crons = ["0 10 * * *"]  # 10:00 UTC = 2 AM PST

2. On-Demand Tracking​

Trigger tracking for specific keywords immediately.

Endpoint: POST /api/rankings/track

{
  "project_id": "proj_123",
  "keyword": "project management software",
  "location_code": 2840,
  "device": "desktop"
}

Response:

{
  "success": true,
  "run_id": "serp_run_abc123",
  "positions": [
    {
      "rank": 3,
      "url": "https://example.com/features",
      "title": "Best Project Management Software",
      "type": "organic"
    }
  ],
  "cached": false
}

Deduplication:

  • Results cached in D1 for 1 hour (same keyword + location + device)
  • Subsequent requests return cached results

3. Task-Based Async Mode​

For large batches, use task mode to avoid HTTP timeouts.

Step 1: Create Task​

Endpoint: POST /api/serp/tasks/create

{
  "keywords": [
    {
      "keyword": "project management",
      "location_code": 2840,
      "device": "desktop"
    },
    {
      "keyword": "task tracking software",
      "location_code": 2840,
      "device": "mobile"
    }
  ],
  "project_id": "proj_123"
}

Response:

{
  "task_id": "task_abc123",
  "status": "pending",
  "total_keywords": 2
}

Step 2: Poll Status​

Endpoint: GET /api/serp/tasks/\{taskId\}

{
  "task_id": "task_abc123",
  "status": "completed",
  "completed_keywords": 2,
  "total_keywords": 2,
  "results": [...]
}

Data Models​

D1 Tables​

queries​

CREATE TABLE queries (
  id TEXT PRIMARY KEY,
  keyword TEXT NOT NULL,
  location_code INTEGER,
  language_code TEXT,
  device TEXT,
  created_at INTEGER NOT NULL
);

serp_runs​

CREATE TABLE serp_runs (
  id TEXT PRIMARY KEY,
  query_id TEXT NOT NULL,
  project_id TEXT,
  check_url TEXT,
  items_count INTEGER,
  se_results_count INTEGER,
  created_at INTEGER NOT NULL,
  FOREIGN KEY (query_id) REFERENCES queries(id)
);

serp_positions​

CREATE TABLE serp_positions (
  id TEXT PRIMARY KEY,
  serp_run_id TEXT NOT NULL,
  rank_group INTEGER,
  rank_absolute INTEGER,
  position TEXT,
  xpath TEXT,
  domain TEXT,
  url TEXT,
  title TEXT,
  description TEXT,
  type TEXT,  -- 'organic', 'local_pack', 'paid'
  is_paid INTEGER DEFAULT 0,
  rating_value REAL,
  rating_count INTEGER,
  created_at INTEGER NOT NULL,
  FOREIGN KEY (serp_run_id) REFERENCES serp_runs(id)
);

ClickHouse Tables​

queries​

CREATE TABLE queries (
  query_id String,
  keyword String,
  location_code UInt32,
  language_code String,
  device String,
  created_at DateTime
) ENGINE = MergeTree()
ORDER BY (keyword, location_code, created_at);

serp_runs​

CREATE TABLE serp_runs (
  run_id String,
  query_id String,
  project_id String,
  customer_id String,
  check_url String,
  items_count UInt16,
  se_results_count UInt32,
  run_date Date,
  created_at DateTime
) ENGINE = MergeTree()
ORDER BY (project_id, query_id, run_date);

serp_positions​

CREATE TABLE serp_positions (
  position_id String,
  serp_run_id String,
  query_id String,
  rank_group UInt8,
  rank_absolute UInt16,
  position String,
  domain String,
  url String,
  title String,
  description String,
  type String,
  is_paid UInt8,
  rating_value Float32,
  rating_count UInt32,
  created_at DateTime
) ENGINE = MergeTree()
ORDER BY (query_id, serp_run_id, rank_absolute);

DataForSEO Integration​

Live Mode (Synchronous)​

Endpoint: https://api.dataforseo.com/v3/serp/google/organic/live/advanced

Request:

[{
  "keyword": "project management software",
  "location_code": 2840,
  "language_code": "en",
  "device": "desktop",
  "depth": 100
}]

Use Case: On-demand tracking, immediate results needed

Cost: ~$0.003 per request

Task Mode (Asynchronous)​

Post Task: https://api.dataforseo.com/v3/serp/google/organic/task_post

Get Results: https://api.dataforseo.com/v3/serp/google/organic/task_get/{task_id}

Use Case: Large batches, cost optimization (cheaper than live mode)

Cost: ~$0.002 per request

API Reference​

POST /api/rankings/track​

Track keyword ranking immediately.

Request:

{
  "project_id": "proj_123",
  "keyword": "project management software",
  "location_code": 2840,
  "device": "desktop",
  "check_url": "example.com"  // Optional: filter results
}

Response:

{
  "success": true,
  "run_id": "serp_run_abc123",
  "query_id": "query_xyz789",
  "positions": [
    {
      "rank": 3,
      "url": "https://example.com/features",
      "title": "Best Project Management Software",
      "description": "...",
      "type": "organic",
      "rating_value": 4.5,
      "rating_count": 1234
    }
  ],
  "cached": false,
  "se_results_count": 45000000
}

GET /api/rankings​

Get historical rankings for a project.

Query Params:

  • project_id (required)
  • keyword (optional, filter by keyword)
  • start_date (optional, ISO date)
  • end_date (optional, ISO date)

Response:

{
  "project_id": "proj_123",
  "rankings": [
    {
      "date": "2024-01-15",
      "keyword": "project management",
      "rank": 3,
      "url": "https://example.com/features",
      "previous_rank": 5,
      "change": 2
    }
  ]
}

POST /api/serp/tasks/create​

Create async SERP tracking task.

Request:

{
  "keywords": [
    {"keyword": "...", "location_code": 2840, "device": "desktop"}
  ],
  "project_id": "proj_123"
}

Response:

{
  "task_id": "task_abc123",
  "status": "pending",
  "total_keywords": 1
}

GET /api/serp/tasks/{taskId}​

Poll task status.

Response:

{
  "task_id": "task_abc123",
  "status": "completed",
  "completed_keywords": 1,
  "total_keywords": 1,
  "results": [...]
}

GET /api/google-ads-status​

Health check for DataForSEO SERP API.

Response:

{
  "status": "operational",
  "credits_remaining": 9876,
  "daily_limit": 10000
}

GET /api/domain-metrics​

Get aggregated domain metrics from ClickHouse.

Query Params:

  • domain (required)
  • start_date (optional)
  • end_date (optional)

Response:

{
  "domain": "example.com",
  "total_keywords": 234,
  "avg_rank": 12.5,
  "top_10_count": 45,
  "top_3_count": 12,
  "trending_up": 23,
  "trending_down": 8
}

Configuration​

Queue Settings​

[[queues.consumers]]
queue = "rankfabric-tasks"
max_batch_size = 1
max_batch_timeout = 30
max_retries = 3
max_concurrency = 20  # Balance: Fast processing vs DataForSEO rate limits

Note: DataForSEO has a 30 req/sec limit. Keep max_concurrency ≤ 20 to avoid rate limit errors.

Cron Schedule​

[triggers]
crons = [
  "0 10 * * *"  # 10:00 UTC = 2 AM PST
]

To change tracking time, modify the cron expression:

  • 0 2 * * * = 2 AM UTC
  • 0 14 * * * = 2 PM UTC (6 AM PST)

Customer Subscriptions​

Tracking runs only for customers with active SERP subscriptions.

Database Setup​

CREATE TABLE customer_subscriptions (
  id TEXT PRIMARY KEY,
  customer_id TEXT NOT NULL,
  project_id TEXT NOT NULL,
  subscription_type TEXT,  -- 'serp_organic', 'serp_local', 'app_tracking'
  is_active INTEGER DEFAULT 1,
  started_at INTEGER NOT NULL,
  expires_at INTEGER,
  created_at INTEGER NOT NULL
);

Subscription Types​

  • serp_organic - Organic rankings
  • serp_local - Local pack rankings
  • serp_paid - Paid ad positions

Billing Model​

  • Per keyword per day: Each tracked keyword counts as 1 billable unit
  • Retrieval: Query serp_runs to count daily runs per customer
SELECT customer_id, COUNT(*) as billable_units
FROM serp_runs
WHERE DATE(created_at) = CURRENT_DATE
GROUP BY customer_id;

React Integration​

Display Current Rankings​

const { data } = await fetch(`/api/rankings?project_id=$\{projectId\}`);

data.rankings.forEach(ranking => {
  console.log(`${ranking.keyword}: #${ranking.rank} (${ranking.change > 0 ? '↑' : '↓'}${Math.abs(ranking.change)})`);
});

Trigger On-Demand Tracking​

const response = await fetch('/api/rankings/track', {
  method: 'POST',
  body: JSON.stringify({
    project_id: 'proj_123',
    keyword: 'project management',
    location_code: 2840,
    device: 'desktop'
  })
});

const { positions } = await response.json();

Build Historical Charts​

Query ClickHouse via /api/domain-metrics for aggregated trends.

Troubleshooting​

Daily tracking not running​

Check:

  1. Cron trigger configured in wrangler.toml
  2. customer_subscriptions table has active subscriptions
  3. Worker logs for cron execution
wrangler tail --format pretty

DataForSEO rate limit errors​

Symptoms: 429 errors in logs

Fix: Reduce max_concurrency in queue consumer:

[[queues.consumers]]
max_concurrency = 10  # Lower from 20

Results not appearing in ClickHouse​

Check:

  1. ClickHouse ingestion queue health
  2. Schema initialized: GET /init/clickhouse
  3. ClickHouse credentials valid
curl https://your-worker.workers.dev/test/clickhouse

Duplicate tracking runs​

Check: D1 serp_runs table for duplicate query_id + created_at within 1 hour

Fix: Ensure deduplication logic in lib/serp-tracking.js is active

Performance & Costs​

DataForSEO Costs​

  • Live mode: $0.003 per keyword
  • Task mode: $0.002 per keyword
  • Daily cost example: 1,000 keywords/day = $2-3/day

ClickHouse Ingestion​

  • Batched inserts (100 rows per batch) reduce request overhead
  • Typical ingestion: 1,000 keywords = 10 ClickHouse requests

Queue Processing​

  • With max_concurrency: 20, process ~1,200 keywords/hour
  • Increase concurrency for faster processing (watch DataForSEO limits)

Future Enhancements​

  • Local pack position tracking (separate endpoint)
  • Paid ad position monitoring
  • SERP feature detection (featured snippets, people also ask)
  • Competitor domain tracking
  • Automated alerts for rank changes
  • Multi-engine support (Bing, Yahoo)