Every brand has bad days. A promotion sells out in an hour, a carrier loses a pallet, or a holiday gift shows up a week after the holiday. And over the holidays, a lot of brands have several of those days at once.

How you respond is called service recovery, and I think it matters more to your loyalty program than most teams plan for. Some of the moments I remember most from running a program weren't wins at all. They were the times something went wrong for a member, and we got to decide how to show up.

One of the hardest ones I dealt with wasn't a shipping problem. I was running an event for our members where the grand prize was a trip to our headquarters, and we covered the flights, the hotel and spending money for the trip. I had given every winner my cell number, and one of them called me in tears. They had just gotten difficult personal news and physically couldn't make the trip. We couldn't change that, but we shipped them a package that recreated as much of the experience as we could, so they still felt part of it.

Another time, our social listening caught someone who wasn't feeling well posting that our product was their comfort food. I went to our CMO, and we reached out and held the product for them at their nearest store, at no charge. Nothing had gone wrong with an order. But it was a moment when someone was vulnerable, and showing up then left a lasting impression.

I once heard Peter Greenberg, the travel journalist, talk about this at a conference, and it has stuck with me ever since. Loyalty is so often built in the moments a brand gets something wrong. It's never on purpose, but it happens, and the customer is left in a vulnerable spot because of it. The brands that own it and make it right in that moment are the ones people remember. And sometimes the brand didn't cause the problem at all. It just recognizes that a customer is having a hard time and steps in.

That's why I think recovery deserves real planning. It's one of the best chances a brand gets to show it cares, and the way you handle it can deepen a relationship more than any campaign can. The moment matters because of the relationship, and the loyalty program is what lets you show up that way every time, for every member, instead of only for the ones who happen to call. Most brands spend months designing how members earn and redeem, and almost no time designing what happens when something goes wrong. So when an order goes sideways, the response comes from a customer service queue that only hears from the people who complain. Recovery should be planned ahead of time, matched to the member, and started from your own order data, so even the members who never reach out feel taken care of. I've written about why your follow-through after Black Friday matters, and this post is about what to do when the follow-through itself goes wrong.

Bad days pile up during the holiday season

The volume alone is a reason to plan for it. In their 2025 returns research, retailers surveyed by the NRF and Happy Returns estimated that consumers would return about $850 billion in merchandise that year, and they expected 17% of holiday sales to come back [1]. In the same research, 71% of consumers said they're less likely to shop with a retailer again after a poor returns experience [1].

Delivery has gotten better, too. ShipMatrix data showed the national parcel carriers posted stronger on-time rates during the 2025 peak season than the year before, with FedEx Express at 95.3% for the peak season and USPS at 94.1% in December [2]. But that still means about one package in twenty showed up late, and across thousands of orders, that's a lot of customers having a bad week with your brand.

The service recovery paradox only goes so far

You've probably heard of the service recovery paradox. It's the idea that a customer who has a problem fixed well ends up more loyal than a customer who never had a problem at all. It's a nice idea, and the research is more careful than that.

A meta-analysis that pulled together years of studies on it found that a good recovery does lift satisfaction, but it didn't make a meaningful difference in whether customers planned to buy again or recommend the company [3]. Other research found the paradox is most likely to show up when the problem isn't severe, it's the first time it has happened, and the customer sees it as a one-off outside the company's control [4].

So recovery can save a relationship after one late package, but it can't make up for the same problem happening again and again. If the same thing keeps going wrong, the fix belongs in operations, and the program's job is to look after customers while that work gets done.

An apology works better when it costs you something

One of the most useful studies on this came from Uber. Researchers sent different apology emails to about 1.5 million riders who'd had a late trip, then followed what those riders spent over the next 84 days [5]. A single late trip cut future spending by 5% to 10%. An apology that came with a $5 coupon paid off for Uber, even after the cost of the coupon. But apologies followed by more bad trips did more damage than not apologizing at all, and riders who got a promise to do better and then had another late trip spent less by day 84 than riders who never got an apology [5][10].

The researchers' takeaway was that an apology works because it's sometimes costly [10]. For a loyalty team, that's a helpful way to think about recovery. A points credit is a real cost, so the member can tell you meant it. You also control the value, and the member has to come back and buy from you to use it.

Your best members feel a bad recovery the most

This part surprised me when I first read it. Loyalty members can be more forgiving when something goes wrong, especially when they value the special treatment the program gives them [6]. But when a recovery goes badly, the damage is bigger with your strongest relationships. Research on airline passengers who complained to a consumer agency after a failed recovery found that the stronger the relationship, the stronger the sense of betrayal when the recovery felt unfair [7].

So the members you've worked hardest to keep are the ones with the most to lose when you get this wrong. That's a good reason to make better, faster recovery part of what a higher tier means.

Decide who gets what before the next rush

The worst time to design recovery is in the middle of the problem, and January returns are coming whether you plan for them or not. Showing up well once makes a nice story. Showing up well every time is what builds the relationship, and that takes a few rules agreed on ahead of time. So something to consider is a simple set built on four questions.

1. How bad was it? A package two days late is different from a damaged gift or a promotion that ran out before checkout.
2. Whose fault was it? Problems in your control, like an oversold promotion, call for more than a weather delay does.
3. Is it the first time? A repeat problem for the same customer needs a bigger and more personal response.
4. Who is the customer? A higher-tier member should get more, and get it sooner.

Once you have those answers, the response follows. Small, first-time issues can get an automatic update and a small points credit, sent before the customer has to ask. Repeat issues and high-tier members should get something bigger and a human touch, and it's smart to fix the cause where you can and be careful about promising it won't happen again. Claims from one-time buyers can go through a quick check, since research on opportunistic claims found they're more common in one-off transactions than in established relationships [8]. And every recovery should have a cap and a log, so the program never ends up rewarding a habit of complaining.

Let your order data start the recovery

On the brand side, recovery usually lived with customer service, which meant it only reached the people who reached out. A lot of unhappy customers never do. They just stop buying, and members can go quiet faster than most teams expect. This is where your loyalty program earns its keep, because it already knows who the member is, what they've bought and what tier they're in, so it can help you show up before they ever have to ask.

DoorDash is a good example of the alternative. For select DashPass members in the US, on eligible orders, if an order arrives 10 or more minutes after the latest time in the delivery window, the member automatically gets a $10 credit as an apology, so nobody has to file a complaint to get it [9]. The opposite approach is to add friction to keep costs down. Delta's 20-minute bag guarantee still exists, but the window to claim it shrank from three days to two hours after arrival [11]. That may save money, but it also tells members the promise is harder to use than it looks.

For a consumer brand, the same idea works with a shipment that arrives damaged or a promotion that sells out before a member can use it. And the triggers are often already in your systems, like a late tracking scan, a canceled line item, a damage report or a return that's taking too long. Something to consider is which of those events your loyalty platform can see today, because the ones it can see are the ones you can respond to automatically. I wrote about why the data behind your program matters in this post on what transactions alone can't tell you.

Measure recovery like any other program change

Recovery costs money, so finance will want to know whether it works, and you can find out on a small group first. Give the recovery to most of the customers who had a problem, and hold a small group back. If holding back an apology feels wrong for your brand, you can test two levels of recovery against each other instead.

Set a pre-period so you know each group's normal buying before the issue, then a post-period of the same length, so you're comparing like with like. The Uber study used 84 days [5]. Then compare repeat purchase rate and net spend, after the cost of the recovery.

You can also compare three groups over time. One group had an issue and got a recovery, one had an issue and didn't, and one never had an issue at all. That tells you how much a bad day really costs your brand, and how much of it recovery wins back. The holdout setup is the same one I walk through in proving a loyalty program is working, and The Most Profitable Customer Is the One You Didn't Pay covers how much of a result you would have gotten anyway.

Put your bad-day plan on one page

Before the season gets any busier, I'd sit down with customer service, ecommerce and whoever runs the program and get the plan onto one page. Start with which order problems your loyalty program can actually see today, and which ones it can't. Then write down who decides what a customer gets, how your top-tier members are handled differently, and how many recoveries one customer can receive in a season. Make sure the plan covers the customers who never contact you, too, because they never show up in the complaint queue. And agree up front on how you'll know whether recovery brought people back, so you have an answer ready when finance asks in the new year.

If you want help writing that page, reach out to our team and we can work through it with you.

Frequently Asked Questions

What is customer service recovery?

Customer service recovery is how a company responds after something goes wrong for a customer, like a late delivery, a stockout or a damaged order. The goal is to fix the problem and keep the relationship, and it usually includes an apology, a fix and sometimes compensation.

What is service recovery in a loyalty program?

In a loyalty program, service recovery is how the brand uses the program to respond when an order goes wrong. The program can make that response faster and more personal, for example with an automatic points credit or a perk based on the member's tier, sent before the member has to ask.

Does the service recovery paradox really work?

Only partly. A meta-analysis found that a good recovery raises satisfaction but doesn't reliably increase repurchase intent. Other research found the paradox is most likely after a first, minor problem that the customer sees as a one-off. Repeated failures can't be fixed with apologies alone.

Should loyalty members get better service recovery than other customers?

Generally, yes. Members who value the program can be more forgiving, but a poor recovery hurts strong relationships more. Higher tiers should get faster and more meaningful responses.

How do you keep service recovery offers from being abused?

Start recovery from order data instead of complaints where you can, set a limit on recoveries per customer per period, keep a log, and use a quick check for claims from one-time buyers.

How do you measure whether service recovery works?

Among customers who had an issue, hold back a small group or test two levels of recovery against each other. Compare each group's repeat purchase rate and net spend after the cost of recovery, using a pre-period and a post-period of the same length.

Sources

1. NRF and Happy Returns, 2025 returns research, Oct 15, 2025.
2. FreightWaves, ShipMatrix data, Jan 12, 2026.
3. de Matos, Henrique and Rossi, "Service Recovery Paradox: A Meta-Analysis," Journal of Service Research, 2007.
4. Magnini, Ford, Markowski and Honeycutt, "The service recovery paradox: justifiable theory or smoldering myth?" Journal of Services Marketing, 2007.
5. Halperin, Ho, List and Muir, "Toward an Understanding of the Economics of Apologies," NBER Working Paper 25676, March 2019.
6. Schumann, Wünderlich and Evanschitzky, Journal of Retailing, 2014.
7. Grégoire and Fisher, "Customer betrayal and retaliation: when your best customers become your worst enemies," Journal of the Academy of Marketing Science, 2008.
8. Wirtz and McColl-Kennedy, Journal of the Academy of Marketing Science, 2010.
9. DoorDash, VIP Delivery terms and conditions.
10. Halperin, Ho, List and Muir, "How firms should apologise," VoxEU column.
11. Thrifty Traveler, Mar 4, 2026.