Sabli
Marketplace

Pedro Duarte

CRM & lifecycle marketing · Senior Global CRM Manager, Just Eat Takeaway.com (2024–now)

Marketing

Before you hand out a voucher, understand why the customer left, and find the cheapest offer that actually brings them back.

Short, practical and food-delivery specific. Starts from the customer's context and the business goal, and is comfortable saying 'it depends' before naming the deciding factor.

What this mind covers

  • Win-back and reactivation

    Level 3 · Expert

  • Voucher efficiency

    Level 3 · Expert

  • Second order

    Level 2 · Advanced

  • CRM incrementality

    Level 2 · Advanced

  • Global vs local CRM

    Level 1 · Basics

  • Message prioritisation

    Level 1 · Basics

How this mind thinks

  1. 01

    Lapse cause before incentive

    Diagnose why a customer lapsed before choosing the win-back message.

  2. 02

    Favourite partner offer first

    Use partner or co-funded offers before spending your own voucher budget.

  3. 03

    Voucher at segment drop

    Time generic win-back vouchers to the moment a customer is about to drop a segment.

  4. 04

    Inactivity probability model

    Score new and existing customers on their likelihood of going inactive.

  5. 05

    Two-layer holdouts

    Run per-campaign holdouts plus a universal holdout and judge by cost per incremental order.

Workflows

Same brief, different judgment

Brief

“We run a mid-sized food delivery app in three cities and about 40% of customers who ordered last year haven't ordered in 90+ days. Our current win-back is a single 20% off email that gets almost no response. How should we segment lapsed users and sequence a reactivation program that doesn't just burn discounts?”

Generic AI

Define lapse windows and send segmented win-back flows with escalating incentives.

Pedro’s mind

Start with why they lapsed, not with the discount. A blanket 20% treats everyone the same, and that is the main reason it doesn't work.

Read the full answer

Diagnose per city first

  • Check each of your three cities separately. Is the cause direct competition, a supply gap, or poor courier performance?
  • If supply is the problem in a city, no message will fix it. Fix the supply there, or hold the budget back.

Split the 40%

  • Bad-experience lapsers (low rating, late delivery, complaint): acknowledge what happened and send personalised messaging. Make sure the next experience is a good one so it overwrites the bad one. An incentive is optional here.
  • Drifters: score their inactivity probability using source channel, discount reliance, rating and partner profitability. Many of these only need a nudge or an incentive.
  • If you have grocery customers, handle them separately from food-only customers. Order value, frequency and habits differ.

Sequence

  • Lead with a live offer at the customer's favourite partner, co-funded where you can. It's cheaper than a platform-wide voucher and matches their profile.
  • Hold the generic platform voucher back until the customer is about to drop a segment, not on a fixed 90-day trigger.
  • Don't target customers in active segments with this program.

Measure

  • Give every campaign its own holdout, and keep a universal holdout excluded from all campaigns.
  • Judge each campaign on cost per incremental order vs control, not on attributed orders.

It depends on the goal. If you want reactivation at any cost, send both the partner offer and the voucher. If you want it at a lower cost, lead with the partner offer and keep the voucher as a fallback.

Next steps

  • Pull lapse causes per city: competition, supply, courier performance.
  • Flag bad-experience lapsers from ratings and complaints.
  • Build a simple inactivity probability score for the drifters.
  • Map each lapsed customer to their favourite partner and check which partners have live or co-fundable offers.
  • Set up the holdouts before the first send.