Model workload before price
We start with monthly workflow runs, executed processing steps and team size. Provider-specific adapters convert the same normalized whole-number workload into the unit each platform meters.
Keep provider exceptions visible
Zapier has non-billable built-in tools and variable task rates. Make can consume extra credits for polling checks, multiple bundles/items and some dynamic-rate operations. n8n counts complete workflow executions rather than multiplying usage by every step. When the baseline model cannot represent an exception, the calculator says so next to that provider's result.
Never extrapolate unpublished pricing
If an official page does not expose an exact price for the modeled usage tier and billing cadence, the site asks the user to verify the live tier. We do not multiply a lower-tier unit price and present that as a provider quote.
Keep plan constraints in the pricing catalog
Published capacities, prices, billing cadence, user limits and automatic-fit constraints live in the pricing data catalog. The engine reads those fields instead of repeating provider-specific numbers in selection code or explanatory copy.
Do not silently switch deployment models
A higher-capacity plan is not automatically a valid upgrade if it changes how the product is deployed. Reference-only plans remain visible on pricing pages but are excluded from automatic plan selection.
Track source and verification state
Every provider record has an official source URL and verification date. Automated checks distinguish an unavailable source from a changed page structure or a suspected pricing-claim mismatch, while stale verification dates force a manual review.