How Potbelly Uses AI to Detect Unhappy Guests, Recover Bad Experiences, and Drive Retention

Momos Partner Since:

2024

Locations:

456

Integrations:

POS (Loyalty Transactions), Momos Surveys, Momos Inbox

Contents

Key Result with Momos

Background

The Challenge

The Turning Point

The Results

Why It Works

Conclusion

Key Result with Momos

Topics	KPIs
AI-triggered loyalty recovery	Guests identified as unhappy and offered loyalty-point recovery saw a +13.5pp uplift in 30-day return rate versus guests who received no recovery.
Automated recovery at scale	Momos detected 400–600 loyalty-actionable negative feedback submissions per week, with 50–60% of all feedback triggering recovery actions.
Higher return rates after recovery	Negative-feedback guests who received loyalty points returned at a 73–80% 30-day return rate, representing a ~20–30% relative improvement.
Closing the engagement gap	AI-powered recovery helped address the gap between unhappy guests, who typically visit ~2.4–2.5x per 30 days, and baseline loyalty guests, who visit ~4.4–4.6x per 30 days.

Background

Potbelly is a nationwide sandwich chain with a loyal customer base and a well-established loyalty program. With thousands of transactions processed daily and a high volume of guest interactions across locations, Potbelly had the infrastructure to capture feedback; but was missing a way to close the loop with precision.

By 2024, Potbelly had the data to know when guests were unhappy, but not the system to win them back. Surveys and cases captured negative experiences, and loyalty points could be issued manually, but the path from feedback to recovery was fragmented. There was no automated way to detect an at-risk guest, trigger the right recovery action, and turn a bad experience into a reason to return.

Potbelly did not just need a way to apologize to unhappy guests. It needed a system that could detect them, recover them, and prove that recovery translated into retention.

The Challenge

Feedback and recovery lived in separate systems
Potbelly lacked a closed-loop recovery system. Guest feedback revealed who was unhappy, but it was not tied to recovery or return behavior. That made it nearly impossible to prove whether Potbelly was driving retention.
No way to track if recovered guests came back
Without a direct link between guest feedback and in-store purchases, the team had no reliable way to confirm whether a recovered guest actually returned. The data existed but the framework to connect it didn't.
Disengaged guests going unaddressed
Guests who submitted negative feedback visited roughly half as often as satisfied loyalty members . This wasn't random. It was a consistent, structural engagement gap being left open every week.
Recovery actions without proof of impact
Loyalty points were issued reactively, with no measurement to confirm they changed behavior. The team needed evidence that recovery translated into return visits — not just closed tickets.

The Turning Point

Potbelly partnered with Momos to build a structured, AI recovery engine on top of its existing loyalty infrastructure — connecting feedback to recovery and tracking the outcome.

Momos Inbox as the recovery hub: All guest feedback — surveys and support cases — was routed through Momos. From there, 50–60% of submissions were flagged as recovery-eligible, generating 400–600 actionable items per week.
Connecting feedback to returns: MMomos built the data foundation for an AI-powered recovery engine. Feedback submissions were matched to guest profiles, loyalty actions, and post-feedback transactions, creating a measurable pipeline from unhappy guest signal to recovery action to return visit.
Recovery intelligence: Momos turned guest recovery into a measurable retention program. Instead of simply issuing points and hoping guests returned, Potbelly could now see whether recovery actions drove incremental visits — and quantify the business impact of winning unhappy guests back.
Consistent points-based recovery at scale: By routing all negative cases and survey responses through Momos, Potbelly moved from a manual, inconsistent recovery process to a repeatable engine that reached hundreds of at-risk guests every week.

The Results

Metric	Impact
+13.5pp Average 30-Day Return Rate Uplift	Guests who received loyalty points after negative feedback returned at 73–80% rates
50–60% of Feedback Is Loyalty-Actionable	Out of 650–1,000 weekly feedback submissions, 400–600 qualify for recovery action — giving Potbelly a consistent, high-volume pipeline of recovery opportunities every week
Structural Engagement Gap Identified	Negative feedback users visit ~2× less than baseline loyalty users.
Points-Driven Recovery Scales Across Months	Across Nov–Dec, return rates for guests who received points climbed consistently — proving the effect is repeatable, not a one-month anomaly

Why It Works

Recovery only works if it's measurable.
Potbelly already had a loyalty program. What was missing was proof that points changed behavior. Momos built the measurement framework that connected the action to the outcome — and the data confirmed it works.
Unhappy guests aren't lost — they're underserved.
Guests who submitted negative feedback still visited 2.4–2.5 times per month. They hadn't churned. A targeted loyalty action at the right moment is enough to close the gap between a disengaged guest and a returning one.
Volume matters.
With 400–600 loyalty-actionable feedback items per week, Potbelly isn't running a recovery program for edge cases. It's running a systematic retention engine that touches hundreds of at-risk guests every week.
Fixing the feedback-to-action pipeline yields compounding returns.
When recovery actions are consistent and fast, return rates improve month over month — not just in isolated incidents.

Conclusion

Potbelly did not need to reinvent their loyalty program. It needed to connect the loyalty infrastructure it already had to the guest feedback signals it was already collecting with AI.

With Momos, Potbelly turned that fragmented process into a closed-loop recovery engine: automatically detect unhappy guests, AI triggers the right loyalty action, and measure when those guests came back.

The impact was clear. Guests who received loyalty points after a negative experience returned at rates 30% more. At 400–600 recovery-eligible feedback events per week, that uplift compounds quickly.

For multi-location brands, the retention opportunity is already sitting inside their own data. The next advantage is not just collecting more feedback. It is using AI to identify the guests most at risk, recover them in the moment, and turn a bad experience into a reason to return.

See all