Strategies

Strategies determine how providers are ranked and how many are included in the fallback chain.

Best Success Rate

Providers ranked by success rate (highest first). The top 2 providers form the fallback chain.

Example: Bright Data has 97% success on amazon.com, Oxylabs has 85%. The workflow routes to Bright Data first, falls back to Oxylabs on failure.

Per-domain: Each domain gets its own ranking. amazon.com might route BD → OXY while walmart.com routes OXY → BD.

Lowest Cost

Providers ranked by cost (cheapest first), filtered to those with at least 70% success rate. The top 2 providers form the chain.

Cost estimation: The system uses probability-weighted expected cost, not just the primary provider’s rate:

Expected cost = Provider[0].cost
  + (1 - Provider[0].successRate) * Provider[1].cost
  + (1 - Provider[0].successRate) * (1 - Provider[1].successRate) * Provider[2].cost

This gives you the real expected cost including fallback probability.

Maximum Reliability

All providers with sufficient data are included in the chain, ranked by success rate. This maximizes the chance that at least one provider succeeds.

Example: 3 providers with 90%, 85%, 80% success rates. Combined probability of all failing: 0.1 * 0.15 * 0.2 = 0.3%. Effective success rate: 99.7%.

Chain length

Strategy	Chain length
Success	Top 2 providers
Cost	Top 2 providers (above 70% success)
Reliability	All providers with data

Minimum data requirements

• At least 10 scraping attempts per provider per domain (routing scores)
• At least 2 providers with sufficient data to compare
• Benchmarks provide initial data, live scraping adds ongoing data