Root-Cause a Sudden 10% Single-Day Fall in OpenTable Booking Conversion

OpenTable's reservation conversion rate fell by 10% from one day to the next. How would you work out what caused it? ```hint Split the ratio first A conversion rate can fall because fewer people booked or because more people visited. Find out which part moved, and whether the drop is real, before looking for a product cause. ``` ### Clarifying Questions - How is conversion defined here: completed reservations divided by sessions, by unique visitors, or by diners who viewed a restaurant's availability? - Is the drop 10% relative (say from 20% to 18%) or ten percentage points? - Has the rate stayed low since that day, or did it recover the next day? - Is the drop visible on every platform and in every market, or did it first appear in one place? ### What a Strong Answer Covers - Pinning down the metric definition and confirming the drop is real rather than a tracking or pipeline artifact - Separating numerator from denominator changes and locating the drop in the booking funnel - Segmenting by platform and app version, market, traffic source, new versus returning diners, and restaurant availability - Testing one-day causes in a deliberate order: releases and experiments, outages and partner integrations, traffic and marketing shifts, supply changes, external events - Ending with a confirmed cause, a decision (roll back, fix or accept) and a view on whether the drop actually hurts restaurants and diners ### Follow-up Questions - The cause turns out to be a test that asks for a card at booking, and it cuts no-shows. Do you keep it? - A promotion doubled traffic: conversion fell but total bookings rose. Is that a problem? - What alerting would you add so the next drop is caught and localized within hours?