Multi-gateway analysis
Multi-gateway analysis is the handling of conversion data when a product collects sales through several separate points (multiple paywalls, multiple checkout flows, multiple subscription tiers) and each customer passes through exactly one of them.
Definition
A set of analytical operations available when the commercial outcome event partitions across multiple measurement points — multiple paywalls, multiple checkout flows, multiple subscription tiers — and each conversion fires exactly one of them.
Partitioned vs. redundant events
The structural condition that enables multi-gateway analysis is partitioning: every conversion goes through exactly one of the candidate events. Five paywalls, each capturing a possible conversion path, with each user passing through one — that's partitioned. The events are different measurement points for the same conceptual signal, and together they cover the full conversion population without overlap.
The contrasting case is redundancy: multiple events fire on the same conversion — a frontend "click buy" event and a backend "transaction complete" event both triggered by one purchase. Compositing redundant events would double-count conversions. The right move with redundant events is outcome-event substitution: pick the most reliable single event and run multiplier analysis against that.
So the diagnostic question before doing anything else: does each conversion fire exactly one of the candidate events, or multiple? Partitioned events composite; redundant events resolve to picking one.
Compositing into a single result cohort
When the conversion event partitions across gateways, the baseline move is to composite them into a single result cohort. "Completed-purchase-via-any-paywall" is a reliable commercial-outcome proxy gathered from multiple measurement points. Using only one paywall as the result cohort treats it as if it were the only conversion path, which it isn't — the analysis discards conversions captured through the other gateways and computes multipliers against an artificially narrow result population.
The composite event is what multiplier analysis runs against by default once partitioning is confirmed.
Gateway proportions as intent signal
The relative volume of conversions through each gateway is itself diagnostic. If 60% of conversions go through Paywall A and 10% through Paywall E, that proportion carries signal about which value-paths are doing the work. Gateways function as early evidence about which user-paths the offering is genuinely serving, not just as collection points. Reverse-engineering from gateway-conversion proportions can surface intent clusters or value-paths the team hadn't fully recognized.
Per-gateway multiplier analysis
Running multiplier analysis with each gateway as its own commercial outcome event — rather than compositing across all gateways — produces a multiplier profile per gateway. Comparing those profiles surfaces what differentiates users who took different paths to conversion. If Paywall A's high-multiplier behaviors differ from Paywall E's, the comparison reveals that the gateways are capturing structurally different user populations on different value-paths, distinct in ways that go beyond price point or feature set.
The two reads beyond compositing are complementary. Proportions tell which gateways carry most weight; per-gateway multipliers tell what behaviors lead users to each. Together, they support designing differentiated paths around the gateway structure — investing in the high-volume gateways, understanding what behaviors lead users to each, and routing users in ways that match the path their behaviors suggest. Per-gateway multiplier profiles can also feed into sorting-hat design as input for cluster discovery, surfacing intent categories grounded in observed conversion behavior rather than only in qualitative investigation.
Example
A productivity SaaS offering has five paywalls, each surfaced in a different part of the product: an upgrade prompt in the export flow, a prompt on the team-invite screen, a prompt when hitting the free-tier project cap, a prompt in the integrations directory, and a generic "upgrade" entry point in account settings. Each conversion fires exactly one of these — the user converts via the paywall they actually encountered.
Compositing produces a result cohort of all paid users regardless of paywall, which is what multiplier analysis runs against by default. Looking at proportions reveals that 55% of conversions come through the project-cap paywall and 5% through the integrations-directory paywall — directional signal that the project-cap gate sits on a much more common value-path than the integrations one. Running per-gateway multiplier analysis shows that the project-cap paywall's high-multiplier behaviors center on creating multiple projects early, while the team-invite paywall's high-multiplier behaviors center on early collaborator invitations — evidence that these gateways serve different user populations on different paths.