Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for e-commerce managers

Customer Win-Back Campaign Agent

Bring lapsed customers back to a second or third order without giving away margin.

Customer Win-Back Campaign Agent: what goes in, what the agent does and what you get

What it does

Every month a store has customers who bought once or twice and then went quiet. Nobody has time to look at each one, so they get nothing, or they all get the same blanket discount that cuts into margin. This agent finds customers who have not ordered in a set number of days and groups them by what they last bought. For each group it picks an offer that stays inside your margin limit, such as free shipping or a replenishment reminder instead of a deep discount. It drafts the message, then waits for you. After sending it watches orders and replies. If a group does not respond, it tests a different offer for that group only. It never sends anything itself. Edge case: a customer who returned their last order is moved out of the offer list and into a service follow-up.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
ApprovedYes, continueYes, continueNoNo 1 STARTS WHEN Monthly win-back run begins 2 USES A TOOL Pull customers with no order in the lapse window(default 90 days) 3 DOES Remove unsubscribed, refunded and open-complaintcustomers 4 DOES Group the rest by last purchase category and ordervalue 5 DOES Pick an offer per group that keeps margin above thefloor 6 USES A TOOL Draft one message per group with the offer andproduct links 7 YOU APPROVE Manager approves each send 8 USES A TOOL Track orders and replies for 14 days 9 CHECKS THE RESULT Did each group reach the target reorder rate? If not: Swap in the next offer in the ladder for thegroups below target and redraft their messages. Back tostep 5. 10 CHECKS THE RESULT Did any offer push a group's margin below the floor? If not: Withdraw that offer and choose a cheaper one.Back to step 5. 11 RESULT Win-back results report and next-month list
Read the steps as a list
  1. Monthly win-back run begins
  2. Pull customers with no order in the lapse window (default 90 days)
  3. Remove unsubscribed, refunded and open-complaint customers
  4. Group the rest by last purchase category and order value
  5. Pick an offer per group that keeps margin above the floor
  6. Draft one message per group with the offer and product links
  7. Manager approves each sendThe agent waits here for your OK.
  8. Track orders and replies for 14 days
  9. Did each group reach the target reorder rate?If not: Swap in the next offer in the ladder for the groups below target and redraft their messages. Back to step 5.
  10. Did any offer push a group's margin below the floor?If not: Withdraw that offer and choose a cheaper one. Back to step 5.
  11. Win-back results report and next-month list

How it decides

Groups customers by last category bought, matches each group to the offer with the best expected return that still clears the margin floor, and changes the offer only for groups whose response stays below target.

  • Never offer a discount that takes order margin below the floor (default 25%)
  • Group with under 40 customers gets a reminder only, no discount
  • Stop contacting a customer after two messages with no open
  • If a group reorder rate is under 2% after 14 days, move to the next offer in the ladder

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Days without an order before a customer counts as lapsed (default 90)
  • Margin floor for any offer (default 25%)
  • Offer ladder in order of cost
  • Target reorder rate per group (default 2%)
  • Maximum messages per customer (default 2)

What keeps you in control

It always asks you first

  • Each group's message and offer before sending
  • Any discount above the starting step

Hard limits

  • Never send to anyone who requested no contact
  • Never exceed the stated discount ladder

It stops when

  • Done: all groups sent once, results reported
  • Stop: reorder rate hits target
  • Stop: customer opts out or complains

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensIn March the agent finds 1,240 customers with no order in 90 days and splits them into five groups. The coffee-filter group gets a free-shipping offer. After 14 days it reads a 0.8% reorder rate, below the 2% target, so it drafts a 10% offer. It checks margin: 27%, above the 25% floor. The manager approves, and the second send lifts reorders to 3.1%.

More agents for e-commerce managers