Part A - Clean: find and treat the quality problems. Record every decision on the Cleaning Log: what you did, how many rows, why, and whether it could change your later results.
Part B - Describe: delivery times, ratings and reorders - the typical value and the spread, overall and by kitchen and zone. Say which measures you chose and why.
Part C - Probability: how often orders are late (over 45 minutes), overall and by kitchen; whether lateness changes the chance that a customer reorders; one normal-distribution question (for example, the time limit that 90% of orders meet) and one binomial question (for example, the chance that 3 or more of a busy evening's 20 orders are late - and what that would cost).
Part D - Relationships and one test: which variables move with delivery time; then one hypothesis test that matters for the decision (a t-test or a chi-square test - you choose, and say why), reported with the 5-Step Interpretation Template.
Part E - Regression: model delivery time with the ToolPak. Interpret each coefficient in minutes, read the fit and significance, check the residuals, make one prediction, and state two limits of the model.
Part F - Chart and recommendation: one chart that carries your argument, and your answer to the founders - make the promise or not; where; at what time limit; what to fix first; and the main risk.