Every CookUnity meal collects ratings and written reviews, and they arrive within days of delivery. On the Jambalaya Pasta page alone (4.5 from 7,440 reviews), this week's reviews include "Was not close to being creamy this time around", "nothing creamy about it" and "I wouldn't order it again". Your retention data scientist post says the gap between a healthy and a churned customer "is often just a few weeks of activity", and the new lifecycle manager is asked to segment on churn signals. A low meal rating is the earliest, most specific churn signal you have, and it lands days before the next menu choice.
First action
Count last month's 1- and 2-star ratings and how many of those eaters ordered the following week.
How to measure it
Orders the week after a low rating
- Now
- –
- Target
- +15% vs holdout
- By
- 90 days
Steps
Baseline
Next-week order and churn rates after low ratings.
Senior Data Scientist, Retention & Product 3 days
Write the message
Meal name, one question, three picks.
Senior Lifecycle Marketing Manager 3 days
Build the trigger
Braze event on rating of 2 stars or less.
CRM Specialist - Lifecycle Marketing & Retention 1 week
Pick the meals
Recommendations from the eater's top ratings.
Senior Data Scientist, Retention & Product 1 week
Exclude the meal
Drop it from automatic selections.
Sr. Product Manager, Shopping Experience 1 week
Set the holdout
10% of low-raters get nothing new.
Senior Lifecycle Marketing Manager 1 day
Send to chefs
Weekly reason summary per meal.
Chef Partnerships team weekly
Read out
Retention, reasons and chef fixes.
Senior Lifecycle Marketing Manager monthly
Why it works
Within 24 hours of a 1- or 2-star rating, send one message (push or SMS, then email) that does three things: acknowledges the meal by name, asks one tap-question about what went wrong (taste, texture, portion, heating, arrived late) and recommends three meals for next week based on what that eater rated highest. Remove the disappointing meal from automatic picks. Send the reasons to the chef weekly, grouped by meal, so recipe drift (like "not creamy this time") is fixed at the source. Keep a 10% holdout to measure the effect on next-week ordering.
Customers rarely cancel after one bad meal; they cancel after nobody noticed. A fast, specific response turns the complaint into a better next box, and the reason data gives the churn model the "why" it is being built to classify.
Ready to use
CopyLow-rating message
Push/SMS: "Sorry the [Meal] missed the mark. What went wrong? Taste / Texture / Portion / Heating / Delivery. Here are 3 meals eaters like you rated 4.5+ for next week: [A], [B], [C]." Email follow-up: same three picks with photos, chef names and a link to edit next week's order.
Risks and how to handle them
Feels automated
Name the meal and the chef; let a person reply to answers.
Too many messages
One message per eater per week at most.
Evidence
- Your websiteMeal pages carry ratings and written reviews: Jambalaya Pasta & Creamy Creole Sauce shows "4.5 Based on 7440 reviews", with reviews dated up to October 21, 2026 that include "Was not close to being creamy this time around. Good flavor but dry.", "This was okay but I wouldn't order it again." and "there was nothing creamy about it". The sitemap lists 3,798 meal pages.Source
- HiringSenior Data Scientist, Retention & Product: "In a weekly subscription business, retention is the engine of growth - the gap between a healthy customer and a churned one is often just a few weeks of activity." The role powers "save flows, skip/pause deflection, lifecycle messaging", must "classify why customers churn so the response fits the reason rather than being generic", and optimize offers "without over-discounting".Source
- HiringSenior Lifecycle Marketing Manager (posted October 6, 2026): owns "deepening subscriber engagement, improving retention, and growing customer lifetime value", sets goals for "churn, order frequency, and LTV", segments by "lifecycle stage, ordering patterns, engagement, and churn signals", and runs "holdouts, incrementality" while assessing "promotion costs and longer-term retention alongside near-term conversion".Source
- HiringCRM Specialist and CRM Marketing Associate, Lifecycle Marketing & Retention (posted September 1 and 15, 2026): build "emails, SMS, push notifications (both ad hoc and automated)" and customer segments in Braze, in a CRM team with a Sr Director of CRM, a Lifecycle Marketing Manager and a Lifecycle Marketing Lead.Source
- Your websiteHomepage: title "Save 50% Off 1st Week"; "Over 90 million meals delivered to more than 1 million customers"; chef cards with average ratings and counts (for example 4.24 from 128,338 ratings). How it works: plans of 4 to 16 meals a week, menus posted two weeks ahead, "A few days before your next delivery, we'll text with our recommendations", and "you can reschedule, skip a week or pause your subscription anytime".Source