Define metrics and comparison periods

Starting conditions and inputs

Identify the organization scope, comparison period, and source records to evaluate.

Values for this task

entry_url, organization_id, record_id, source_records_url, statistics_url

Command sequence

open "$entry_url" && snapshot -i

# Outside browser: The evaluation owner defines numerator, denominator, comparison period and organization scope.

open "$source_records_url" && snapshot -i

get text body

open "$statistics_url" && snapshot -i

get text body

Ref targets and expected results

#	Command / ref	Target and verification
1	Outside browser	The evaluation owner defines numerator, denominator, comparison period and organization scope.
2	`open`	Source records for the evaluation
3	`get text`	Read period, organization and baseline source values.
4	`open`	Dashboard or chart view under the same conditions
5	`get text`	Compare source totals with the card's `numerator/denominator` summary.

Check the result

Source records match the criteria established for this task.

Resuming after interruption

Reread the same source IDs and resume at the unfinished check.

Comparing results before and after a change requires consistent denominators and time windows. Calling two different formulas by the same metric name leads to false conclusions.

This reference defines the six core value metrics on the dashboard. It also explains multi-branch deduplication and shows how to verify figures against source tables.

Each default card on the dashboard measures one value outcome. Monthly bars split the denominator into Converted (numerator) and Unconverted (denominator minus numerator). The card subtitle reports the overall percentage and (numerator/denominator) counts across the full window.

Value outcome	Dashboard chart title	Denominator (population)	Numerator (`Converted` rule)	Cohort bucket and exclusions
Increase inquiry satisfaction and conversion	`Conversation to reservation share`	Conversations started in the month (`conversations`)	Conversations whose customer created a non-canceled reservation at or after the conversation start	Bucketed by conversation `created_at`; excludes `canceled` reservations
Reduce repetitive staff work	`AI autonomous response share`	Conversations started in the month (`conversations`)	Conversations handled by AI without human status or human staff messages	Bucketed by conversation `created_at`
Increase return visits and retention	`Returning visit share`	Non-canceled reservations in the month (`reservations`)	Reservations that are the second or later booking for the same unified customer	Bucketed by reservation `created_at`; excludes `canceled` reservations
Increase new customer acquisition	`Engaged customer inflow`	Customers created in the month (`customers`)	New customers linked to at least one conversation, valid reservation, or valid purchase	Bucketed by customer `created_at`
Improve marketing and sales efficiency	`Reservation to purchase share`	Non-canceled reservations in the month (`reservations`)	Reservations whose customer completed a valid purchase at or after the reservation creation time	Bucketed by reservation `created_at`; excludes canceled or failed purchases
Keep connected customer records	`Connected customer record share`	Active customers in the month (`customers`)	Customers who have at least two record types among conversations, valid reservations, and valid purchases	Bucketed by customer `created_at`

Keep the counting unit explicit in internal reports: conversations, reservations, or customers. Do not use one metric name for two different denominators.

Choose the population and period

Fix the organization scope, comparison window, time zone, and identity rule before comparing two periods.

Organization scope (subtree vs organization): Selecting a parent headquarters organization aggregates all child branches (subtree). Customers across branches are unified by normalized phone number, or by lowercase email when no phone exists. Select an individual branch or set the chart scope to organization when evaluating a single location.
Cohort month: Each bar groups records by the month the denominator record was created. Recent months can include customers who have not yet booked or purchased.
Test records: Exclude synthetic test customers and test bookings when comparing operational periods.

Check source records

Dashboard charts calculate results directly from Customers, Conversations, Messages, Reservations, and Purchases. When a number looks unexpected, check the source tables in the following order.

Symptom	Source tables to inspect	What to check
Low conversation-to-reservation share	Conversations, Reservations	Confirm that reservations link to the same customer or phone number used in the inquiry.
Low returning visit share	Customers, Reservations	Check customer identity rules so repeat visitors are not saved as unconnected records without phone numbers.
Zero reservation-to-purchase share	Reservations, Purchases	Check that completed purchases link to the customer and are not marked `canceled`, `voided`, or `failed`.
Sudden drop in AI autonomous share	Conversations	Check whether staff took over conversations in Inbox during the period.

Interpret differences

A change in conversion percentage alone does not prove that one feature or campaign caused the difference.

Read the denominator alongside the rate: A rate increase from 30% to 50% with a denominator drop from 100 to 6 means fewer total conversions. Always check the (numerator/denominator) counts in the card header.
Account for external changes: Check whether ad spend, business hours, pricing, or seasonal demand changed during the same period.
Choose the next verification step: When unconverted counts rise, pick one area to verify first. Inspect workflow rules, AI prompt reviews, or booking availability.