Push Notifications Without Annoying Users: Buen Provecho's Location-Based Strategy

Most apps treat push notifications as a blunt instrument—send them to everyone, hope for clicks, and worry about unsubscribes later. But what if your app can't always deliver what users want, even when they click? For Buen Provecho, a food waste marketplace operating across Latin America, push notifications needed to be smarter, more targeted, and respectful of the fine line between engagement and annoyance.

On the Levels Podcast, Guillermo Martinez, COO and co-founder of Buen Provecho, explained how they developed a location-based push notification strategy that keeps users engaged without burning them out—even when supply constraints mean they can't always fulfill demand.

The Core Problem

Unlike typical food delivery apps where restaurants have full menus available during operating hours, Buen Provecho only has what's in surplus on any given day. A bakery might have excess pastries in the morning but nothing by afternoon. A supermarket might have surplus produce one day and nothing the next.

This creates a fundamental challenge for user engagement. Generic push notifications—"Check out today's deals!"—will frustrate users who click through only to find nothing available near them. Do this enough times and users either disable notifications or uninstall entirely.

"Our biggest friend, our ally is push notifications because it was like an easy way, a low cost way to just get them and get in touch with them."

The team needed push notifications to work harder and smarter, targeting only users who actually had products available nearby.

Building Privacy-Conscious Location Targeting

The solution seemed obvious: use location data to send notifications only when stores near a user published products. But implementing this while respecting privacy proved tricky.

"What we did quite early on is recommend people stores that are near them, which was at the end, when we started to do that and we wanted to do that, it was quite hard. It's like, how do we do that without violating any kind of privacy concerns? which is hyper important."

Rather than tracking users' real-time location (creepy and battery-draining), they found an elegant workaround: use the last location where the user made a purchase. This provided enough geographic precision to target relevant notifications without continuously monitoring users' movements.

The system works on a roughly two-kilometer radius. When a store within that radius publishes new surplus products, users who've previously purchased nearby get notified. The approach balances relevance with privacy—users only get notifications for stores they can realistically visit.

The Cooling Period Strategy

But location targeting solved only half the problem. Users still needed protection from notification fatigue. The team implemented what Guillermo calls a "cooling period"—a strategic delay before sending another notification to someone who recently engaged.

"The people that engage that notification at the same time, we don't send them another one, for example, maybe 48 or 72 hours later. We try to go for people that didn't use a notification."

This creates a healthy notification rhythm: If a user clicks a notification and makes a purchase, they won't get another notification for 2-3 days. The system prioritizes re-engaging dormant users rather than overwhelming active ones.

The logic is counterintuitive but sound. Active users will check the app anyway—they don't need constant reminders. Dormant users who haven't engaged recently are the ones who benefit most from timely notifications about nearby products.

Different Notifications for Different Scenarios

Beyond location-based store notifications, Buen Provecho developed notification strategies for different user scenarios. When a user adds items to cart but doesn't complete the purchase, they trigger abandoned cart notifications. These remind users about products they showed interest in before inventory runs out.

For stores, they've created a favoriting system that fundamentally changes the notification dynamic. When users favorite a store, they opt into instant notifications the moment that store publishes new products.

"We have users that really like some particular stores and you can like favorite them. So you get a notification in the second that they publish something. So that actually is like a double-edged sword, right? Because active users get notified immediately. So they purchase instantly."

This creates an interesting tension. Power users who favorite multiple stores snap up inventory quickly, which can leave new users frustrated when they arrive minutes later to find sold-out products. It's a reminder that even sophisticated notification strategies can't solve fundamental supply constraints—they can only optimize around them.

The Technical Evolution

When Buen Provecho started, they didn't have these capabilities. Their WhatsApp group validation phase operated entirely manually—someone literally managed a group chat and posted products as they became available. But even then, they were learning about user behavior and notification preferences.

As they built their app, these insights shaped their notification infrastructure from the start. Rather than building generic broadcast notifications and iterating later, they designed for personalization and geographic targeting from day one.

"Yeah, I mean, at the start, obviously now you have a lot of like resources and tools and can use AI agents and many resources that can help you just work smarter and be more productive. But yeah, we just kind of like, know, the bootstrap mentality. We just did everything and that just worked, right?"

The manual work they did early—photographing products, managing the chat, understanding which items sold fastest—gave them the domain knowledge to build smart automation later.

The Personalization Roadmap

Location-based notifications work well, but Guillermo sees room for significant improvement. The next phase involves behavioral personalization: notifications based on past purchases, preferred product categories, and shopping patterns.

"Now we're going back to the app and we are actually developing AR recommendations, personalized on what you have been buying, what you like, how we can basically get you to purchase more, frank, and always keep you as a user."

Imagine getting notified only about the types of products you actually buy. If you consistently purchase baked goods but never buy produce, the system would prioritize bakery notifications over produce alerts. This level of personalization could significantly improve conversion rates while reducing notification volume.

They're also working on automatic notifications based on purchase patterns. If you typically buy from a particular store every Tuesday, the system might proactively notify you on Tuesday mornings when that store has new inventory.

Why This Matters for Other Apps

Buen Provecho's notification strategy offers lessons for any consumer app, especially those with variable supply or geographic constraints. The core principles translate across industries.

Respect context. Notifications should consider whether the user can actually act on them. Don't notify users about things they can't access—whether that's geographic distance, sold-out products, or simply bad timing.

Build cooling periods. More notifications don't necessarily drive more engagement. Sometimes less is more, especially if you're trying to re-engage dormant users rather than pestering active ones.

Make opting in explicit. Features like favoriting stores let users signal what they want to be interrupted about. These explicit opt-ins create permission for more aggressive notification strategies in narrow contexts.

Start manual to learn patterns. Buen Provecho's manual WhatsApp phase taught them how users actually behaved before they invested in complex automation. Manual processes often reveal insights that inform better automation later.

Balance immediate wins with long-term sustainability. It would be easy to blast notifications to everyone whenever any store publishes products. That might drive short-term engagement but would destroy retention as users got annoyed and disabled notifications. The patient approach of building sophisticated targeting creates sustainable engagement.

The Measurement Challenge

Like any feature, notification strategies need measurement. Buen Provecho tracks standard metrics: notification click-through rates, conversion rates, and opt-out rates. But they also monitor more nuanced indicators.

Time between notification and purchase matters. If users consistently take hours to act after a notification, that suggests poor timing or irrelevant targeting. Quick conversions indicate the notification arrived at the right moment with the right message.

They also track how notification engagement correlates with overall retention. Users who regularly engage with notifications but never complete purchases are a warning sign—they're interested but consistently disappointed, probably due to lack of available inventory in their area.

Key Takeaways

• Location-based targeting increases relevance by ensuring users only get notified about products they can actually access
• Cooling periods prevent burnout by giving users breaks between notifications rather than bombarding active users
• Privacy-conscious design builds trust using last purchase location rather than real-time tracking
• Explicit opt-ins enable aggressive strategies features like favoriting let users signal when they want immediate notifications
• Personalization is the next frontier behavioral data can make notifications even more relevant over time

Listen to the full conversation with Guillermo Martinez on the Levels Podcast to hear more about how Buen Provecho balances user engagement with supply constraints across Latin America.