Keeping a luxury retail clienteling system running, improving, and catching what breaks — one year of ongoing product operations
After an initial CRM rollout, someone needed to own the system day to day — triaging enhancement requests, catching data issues before they became customer-facing problems, and finding ways to put the system to better use. That ongoing ownership ran for about a year, covering a live clienteling platform used by sales associates across an international luxury retail brand.
Enhancement and maintenance requests came in continuously from business users — client handover, tagging, appointment reminders, booking, messaging. I managed this as a live backlog through Jira and Confluence: triaging, prioritizing against business impact, and coordinating with technical vendors to get things shipped.
Not every issue arrives as a ticket. This one surfaced as a pattern: a growing number of customers had no sales associate assigned to their account.
Over 100 migrated customers had been left without an assigned sales associate. The root cause traced back to office-based QR codes that were missing a binding parameter, so the usual customer-to-associate link never formed. Rather than fixing accounts one by one, I pulled the full affected list and coordinated a systematic review of all customer-SA bindings with the sales team — closing the gap for good, with no recurrence afterward.
Beyond fixing what broke, part of the work was finding new value in the system. I coordinated the build of an offline appointment feature inside the CRM's Client 360 view, then defined task-management requirements to reach a segment that had never been systematically worked before: light-dormant customers, with no purchase in three to five years — over 2,000 accounts.
Diagrams above are redrawn and generalized — no client system screenshots, real customer data, or brand names are shown.
A year of continuous ownership: an 88% closure rate on an active 190+ item backlog, a systemic data issue found and closed for good rather than patched case by case, and a previously-unworked customer segment turned into 71 confirmed repeat orders — tracked, not assumed.