We had a reporting dashboard that was throwing intermittent errors and affecting a few internal teams. Requirements weren't fully locked — product hadn't signed off on the final data model — so I spent the first week syncing with stakeholders to get alignment on what the fix should look like long-term. Once I had a clearer picture, I scoped the fix, got it reviewed by the team, and shipped it in week two. The dashboard stabilised and the internal teams were unblocked. It was a good example of making sure we didn't introduce more ambiguity by rushing.
Our internal reporting dashboard started throwing errors affecting three teams mid-sprint. Product hadn't locked the data model, so I made a call: I'd fix the broken query path with a scoped shim — not the full model rewrite — and ship it within 48 hours. I flagged the trade-off in a doc, got async sign-off from the one stakeholder whose data was most at risk, and shipped. Dashboard error rate dropped from 40% to under 2% within the day. Two weeks later, once the model was finalised, I replaced the shim cleanly. Iteration beat waiting.