Repricing Intelligence on Allegro: Always the Lowest Offer

Client Snapshot

A Polish garden-irrigation retailer with ~700 active Allegro offers, primarily Rain Bird and Hunter products.

The Challenge

The client's entire strategy depended on being the cheapest offer on every listing. But Allegro's API was not available to them, listings were protected by CAPTCHA, and — the real problem — the same physical product appeared under dozens of slightly different titles across sellers. A sprinkler head listed as "Hunter PGP-ADJ" by one seller and "Hunter PGP ADJ Ultra rotor" by another is the same SKU, and any repricing system that can't see that is worthless.

Manual matching was consuming 15+ hours a week and still missed offers.

Objectives

• Identify, for each of the client's 700 offers, the genuinely competing offers
• Capture the current lowest competitor price
• Feed a repricing rule (undercut by 1 grosz) without human intervention

The Actowiz Approach

The scraping was the easy half. The hard half was exact product identification, and that's where the engineering went.

We built a matching layer that normalized manufacturer part numbers out of free-text titles, then scored candidates on a combination of MPN match, brand, specification attributes, and title similarity. Matches above a confidence threshold entered the repricing feed automatically; borderline matches were queued for a one-time human decision and then remembered permanently.

Collection ran on a schedule with session management and human-paced request rates, respecting the platform's load rather than hammering it.

Compliance Note

Public listing data only. No account access, no automated purchasing, no CAPTCHA-farm circumvention. Request rates were kept conservative.