Know why subscribers cancel, from what they type

Fields

• Reason picked from the dropdown
• What the subscriber typed in the "anything else?" box
• Plan, as a short line

Questions

1. What main reason does the customer give for wanting to cancel? Choice

   Read the free text first; use the picked reason only when the text gives no reason of its own. Pick the reason the customer gives most weight. Do not guess a reason from the plan.

   • Price The boxes cost too much for what they get, or the price went up, with no mention of being unable to pay.
   • Cannot afford it The customer says they cannot pay right now, for example after losing income, rather than that the price is poor value.
   • Found a better deal elsewhere The customer names or describes another meal kit, shop or offer as the reason.
   • Not using the boxes Meals go uncooked, they order takeaway instead, or they no longer need the boxes, with no end date in sight.
   • A break for a while Travel, a busy season, a holiday or another period with an end, after which they may want the boxes again.
   • Food quality Ingredients arrived spoiled, bruised, short or wrong, or the meals did not taste good.
   • Delivery problems Boxes came late, did not arrive, arrived warm or were left in the wrong place.
   • Menu does not suit them Not enough variety, or no meals for a diet, allergy or taste in the household.
   • Billing problem A double, wrong or unexpected charge, or trouble with a payment.
   • Poor support Unanswered emails, unhelpful replies or other poor handling by our support team, as the main reason.
   • Moving home The customer is moving, or moving to an address we may not deliver to.
   • No reason given Neither the free text nor the picked reason gives a reason.

2. Does the free text give a reason for cancelling that the picked dropdown reason does not cover? Yes / No

   Compare the free text with the picked reason. Count a second reason, or a different main reason, that the picked reason leaves out. Yes: The text gives at least one reason the picked reason does not cover. No: The text repeats or explains the picked reason, gives no reason, or is empty.

3. Does the customer say they may come back, or that their reason lasts only for a time? Yes / No

   Count a stated end to the reason ("for three months", "until the baby arrives", "after the summer") and plain statements such as "I will be back". Liking the food is not enough on its own. Yes: The text says the customer may return, or that the reason has an end. No: The text gives no sign of a return or an end to the reason.

4. Does the text describe a specific problem our support team could put right? Yes / No

   Count a concrete problem with a box, a delivery, a charge or the account, such as a double charge, a missing ingredient or a box left at the wrong door. General dislike of the price or the menu is not a problem support can put right. Yes: The text names at least one such problem. No: The text names no problem of that kind.

5. Does the customer name or clearly describe a competing provider or its offer? Yes / No

   Count another meal kit, a grocery delivery service or an offer from one, named or clearly described. Cooking from the supermarket in general does not count. Yes: The text names or clearly describes a competing provider or its offer. No: No competing provider or offer is named or described.

6. How frustrated is the customer in what they wrote? Scale

   Rate the words about us, the boxes or the service. Capitals and punctuation count only together with the words.

   • None Calm, matter of fact or friendly, with no complaint.
   • Mild Mentions a problem or a let-down plainly, without blame or anger.
   • Clear Blames us directly, or complains about how they were treated.
   • Severe Anger, insults, shouting in capitals, or calling the service the worst they have had.

7. Does the customer ask for money back, compensation or a goodwill payment? Yes / No

   Count a direct or an implied request, such as "I want the second charge back" or "who pays for the box that never came?". Do not count a request only for an explanation or an apology. Yes: The customer asks for a refund, a charge back, compensation or a credit. No: The customer asks for no money.

8. Does the customer say they never signed up, or did not know they would be charged? Yes / No

   Count "I never signed up", "I thought it was a one-off box" and "I did not know it would renew". A double or wrong charge on a subscription the customer knew about is a billing problem, not this. Yes: The customer says they did not agree to the subscription or did not know it would charge them. No: The customer says nothing of the kind.

9. Does the customer say firmly that they want to cancel now, without offers? Yes / No

   Count "just cancel", "do not try to keep me", "no offers please" and words to the same effect. Yes: The customer says they want no offer or no delay. No: The customer says nothing of the kind.

10. Does the text describe a hazard or an injury from the product? Yes / No

    Count illness after a meal, an allergic reaction, a foreign object in the food, packaging that cut someone, or food that arrived unsafe to eat. A plain dislike of the taste does not count. These go to the food safety team. Yes: The text describes at least one hazard or injury. No: The text describes no hazard or injury.

11. Is there enough in the text to tell why the customer is cancelling? Yes / No

    Answer No when the free text is empty, a few characters, unrelated to cancelling, or too garbled to read. Yes: A careful reader could tell from the text why the customer is cancelling. No: The text is too short, empty or garbled to tell.

Lens columns

cancel_reason, cancel_reason_probability, reason_not_in_dropdown, reason_not_in_dropdown_probability, may_return, may_return_probability, fixable_problem, fixable_problem_probability, competitor_named, competitor_named_probability, frustration, frustration_average, redress_requested, redress_requested_probability, charged_without_consent, charged_without_consent_probability, no_offers_wanted, no_offers_wanted_probability, safety_mention, safety_mention_probability, enough_information, enough_information_probability