Business Problem
In a higher-ticket device flow, some customers did not drop off because they lost interest, but because they could not pass verification for the specific product they selected. From a business point of view, this is one of the most painful moments in the funnel: the user has intent, has already invested effort in the process, and the order still ends in nothing.
An earlier alternative-offer mechanism I had already built proved that these situations could be recovered. The next step was to build a more systematic solution, one that could automate the recovery path instead of handling it as a one-off rescue.
Scope of Work
I delivered this feature end-to-end across admin, web, and mobile application layers.
The implementation included:
- new database tables and relations with migrations,
- a job queue for generating and processing offers,
- integration with an algorithm running on GCP,
- automatic SKU and price proposal generation,
- an email template for the alternative offer,
- an admin panel for managing the process,
- support across web and app flows,
- automatic cancellation when external API errors occur,
- storing
baseandhardlimits at offer generation time.
Technical Decisions
The core challenge was not only generating an alternative, but doing it in a scalable and operationally safe way. That is why I split the logic into clear stages: input persistence, asynchronous proposal generation, validation, customer communication, and exception handling.
From the business angle, it was not enough to show "something cheaper". The system had to generate an offer that still made commercial sense. That is why pricing constraints were stored at generation time, and why the flow could automatically cancel itself when an external API failed instead of leaving broken half-finished offers for the team to fix manually.
Impact
This implementation turned difficult edge cases after failed verification into a repeatable revenue-recovery workflow. Instead of ending the process with a hard stop, the platform can return to the customer with a meaningful, automatically generated alternative.
The most important verified context is already there: the earlier alternative-offer mechanism was recovering around PLN 50k in revenue per month, and this system was built specifically to scale that result further and cover additional lost-order scenarios.
