Ask a marketer how a spend-and-get offer makes money, and you get a nice answer about lifetime value. Ask the CFO, and you get the real one. Plenty of people stop halfway.
We have priced that in for years. The 20th purchase nobody makes. The third deposit that slips a month. Breakage is the polite word for it, and it has quietly paid for a lot of good marketing.
The money saved was a nice-to-have. Everyone who finished came out the other side with a habit, and the habit was always the point.
Lenders now pay twice. Once at sign-up, and again for the behavior they want repeated. Card mail with a second bonus went from 0.7 million pieces to 9.2 million in a year.
The part I keep thinking about is who else responds to that design.
A habit appears to be human-only on the surface. An AI agent builds one for exactly the reason we do: because something in the offer makes coming back worth it.
Breakage did two jobs. It kept part of the bonus, and it taught the habit. AI Agents follow the same design.
The sign-up bonus still drives card opens.
Delta pays 80,000 miles after $4,000 of spending, then 20,000 more after another $2,000. Marriott pays 75,000 points after $3,000, then 50,000 more at $6,000.
The first Delta milestone pays 20 bonus miles per dollar. The second pays 10. The habit costs half as much as the sign-up.
Capital One does it another way. It puts $300 of the Venture reward into its own travel site, so the first booking happens there.
| January to July 2025 | 0.7M | |
| January to July 2026 | 9.2M |
Design the welcome period as a sequence of earning moments so value continues after the first payoff instead of ending with it.
I don’t think there is a right or wrong here. The takeaway is: how could I add to or redesign my existing sign-up bonus to build a habit for the human and/or agent?
Saving and spending are two different relationships. One mailer builds a savings balance. The other buys a place in the customer’s daily payments.
They cost different amounts. The debit version pays out purchase by purchase. The savings version pays only when the customer finishes all three deposits.
KeyBank tested time. One mailer paid $700 for keeping $10,000 in checking for 120 days. The other paid $600, then $300 more for keeping it there a year.
Citizens aimed to balance an action consumers believe they can adopt with what will build a long-term financial habit.
Which habit in the first 90 days tells us a customer will still be here in year three?
A person repeats a behavior when the reward makes repeating it worth the effort. That is the whole design behind a second milestone, and Citizens priced it at $300.
Software works the same way. An agent does more of whatever earns more. Put an ongoing reward inside the product and the agent has its reason to come back, the same reason a customer has.
Look at the habits in this edition from that angle. Each one asks a person for effort and asks software for a rule.
The welcome bonus persuades a person once. The ongoing reward keeps working for the customer and for the software they hand the card to.
| The habit an issuer pays for | What it asks of a person | What it asks of software |
|---|---|---|
| $500 into savings, 3 months in a row | A reminder every month | One standing transfer |
| 100 debit purchases in 60 days | Roughly 2 a day, counted by hand | A counter |
| $10,000 in checking for 365 days | A year of leaving it alone | A balance rule |
| Another $2,000 on the card | Watching the statement | A threshold |
| Reaching $6,000 on the card | Adding it up as you go | A total that updates itself |
Which reward in our product gives a customer a reason to come back in month 13?
In this section, I highlight the other side of my own case so you can get the fully rounded argument and make the decision that makes sense for your LOB.
Simple offers attract more people. One number, one hurdle, one payout. Every extra step adds a line to the offer and a reason to pass.
The people who finish may have stayed anyway. Anyone who saves for three months in a row wanted to save.
The software argument is a design case today. A human’s habit assumes its longer term. An AI Agent has no reason to stick with our product after full value extraction. Much of this world is still unexplored and will surely force us to redesign our value props in other ways, too.
Pay for a habit when your product earns more every time somebody uses it. Then build the ongoing reward into the product itself, because that is the piece that keeps working after the welcome period ends.
A person comes back because the reward makes coming back worth it. An AI agent comes back for exactly the same reason. One design earns both, and a win-win wins every single time.
