Skip to main content
Fine-tune the balance between response speed and match quality. Learn how to configure message analysis modes, intent filtering, and fallback behavior for your specific use case.
Variable for playing around with keyword quality, link accuracy, and response times.
ValueNotesExpected LatencyUse Case
fastQuickest, but less likely to find a working affiliate link~150msWhen response times matter and you’d prefer to fail fast
standardStrong keyword quality and decent link matching~1.4sBalanced default
bestStrong keyword and strong matching~2.5sIf you’re okay with a longer response time and want the keyword/link combo that will drive most conversions
Try these optimizations:
  1. Use quality: "fast" - Most response times are in the 100-200ms range, but often does not find matching affiliate links in time. Good if you’d prefer to fail fast.
  2. Limit message length - Really long messages can cause slight delays in analysis
  3. Cache responses - You could build your own database for cached results by keywords.
Fill rate varies based on your traffic quality:
  • Product-focused chatbots (shopping assistants, product Q&A): 60-80%
  • General-purpose chatbots (customer support, general AI): 20-40%
  • Mixed traffic: 30-50%
If your fill rate is lower than expected, check that messages contain clear product references and come from US, English-speaking users.
Yes, results are deterministic for identical inputs. The same message with the same parameters will return the same affiliate link, making caching effective.However, results may change over time due to:
  • Product availability changes
  • Price fluctuations
  • New affiliate partnerships
For most use cases, caching responses for 1-24 hours is safe and reduces API calls.