See whether customers come back after their first order.
Ask how each month's new customers kept ordering. Concierge groups customers by the month of their first order, counts who ordered again one, two and three months later, and hands you the table and a chart.
- or weekly customer groups
- Monthly
- or weekly customer groups
- the only data it needs
- Orders
- the only data it needs
- saved as a file to share
- Chart
- saved as a file to share
The problem
Are the customers you won in the spring still ordering in the fall? One repeat rate blends new and old customers together. This playbook splits customers by when they first ordered, so you can see whether newer groups come back better or worse than older ones.
What it reads.
Your orders
Every order in the window: who ordered, when, and whether it was cancelled.
A window that starts early
Groups start at a customer's first order inside the window, so the window has to start early enough to hold it.
Your calendar
Months and weeks follow your workspace's time zone.
Your definitions
If you have saved a definition for a term, it uses yours and says so.
The steps it follows.
- 1
Prepare the data
Concierge prepares a private copy of your orders and opens it on its own computer.
- 2
Set aside cancelled orders
Cancelled orders are left out unless you ask for them, and it prints how many.
- 3
Build the table
One row per first-order month (or week). Column 0 is the group's size; column 1 counts customers who ordered again a month later, and so on.
- 4
Show percentages and a chart
Retention as a share of each group, with group sizes beside it, plus a heatmap or line chart.
A worked example.
A made-up store, to show how the playbook works from question to answer.
- The store
- A coffee brand on Shopify that sells bags one at a time and on subscription, with orders going back to early 2024.
- The question
- "How well do customers who first ordered in 2025 come back, month by month?"
- What Concierge does
- Prepares the orders with a window that starts before January 2025, leaves out cancelled orders and prints how many, then builds one row for each first-order month from January 2025 on.
- What comes back
- A table with each group size in the first column and repeat customers in each column after it, the same table as percentages, and a heatmap saved as a PNG. It also points out that the most recent months have had less time to come back.
- How to read it
- Read across a row to see how one group fades over time. Read down a column, say month 3, to compare groups at the same age. If the newer rows run higher in the same column, newer customers are sticking better.
- A good next question
- "Do the same by week for customers from our March promotion." Or ask what those groups are worth with the customer lifetime value playbook.
This store is invented for illustration. It is not a Chargezen customer, and no results are shown: your answer depends on your own data.
Run it, then make it yours.
You don't need to name the playbook. Concierge reads each playbook's "When to use" line and opens the one that fits your question.
- In the Chargezen admin
- Type your question in Ask Concierge, for example: "How well do customers who first ordered in 2025 come back, month by month?" Concierge does the work and replies with the answer and any files.
- In ChatGPT or Claude
- Connect Chargezen to ChatGPT or Claude, then ask Concierge from the chat. The request goes to your workspace's own agent, with the same playbooks, and the reply comes back to the chat.
- In the Finn app for iPhone
- Type or dictate the question in the chat box on Home and send it to your agent. It's the same agent with the same playbooks.
- Change the steps
- Open Concierge > Build > Playbooks, pick Cohort retention and edit the instructions. Edits apply to new requests; work already under way keeps the version it started with.
- Use your definitions
- If you've saved a definition for a term the playbook uses, like what counts as an active customer, Concierge applies yours instead and prints it in the answer.
- Choose who uses it
- It starts switched on for your workspace's operator agents that can analyze store data. Switch it on for other agents under Agents, archive it, or click New Playbook to write your own.
What you get back.
The table
Every group's size and how many came back in each month after.
Percentages with sizes
So a tiny group's big percentage doesn't fool you.
A chart file
A heatmap or line chart as a PNG, ready for a team meeting.
Get answers you can check, without the spreadsheet afternoon.
Every Concierge workspace starts with these playbooks. Ask a question in plain English and see the numbers and the work behind them.
Common questions
Whether customers come back after their first order: for each month (or week) of first orders, how many of those customers ordered again one, two, three and more periods later.
When you want to know if retention is improving, compare customers won in different months or campaigns, or see how quickly customers drop off after their first order.
Because they have had fewer months to come back. The answer says so, and you should compare groups at the same age (the same column), not the same calendar month.
No. Concierge sees the title and "When to use" line of every published playbook switched on for the agent, and opens the full steps only when one fits your question. You can still say "use the Cohort retention playbook" to be sure.
Yes. It's your workspace's playbook: edit the steps, rename it, switch it on for other agents, archive it or delete it under Concierge > Build > Playbooks. A deleted default playbook stays deleted.
Running a playbook is a Concierge job, so it uses your workspace credits like any other job when Concierge runs on Chargezen-managed AI models. Concierge Starter is $20 a month with 20,000 credits. Pro is $199 a month with 199,000 credits, or $1,990 a year.
More playbooks and guides.
