Define what faster means
A faster first render, a lower API latency, and smoother scrolling are different problems. Name the metric, where measurement begins, and the conditions under which you collected it. One positive impression does not establish an improvement across the system.
Compare like with like
Explain whether data size, resources, network conditions, and sampling were comparable before and after the change. For a variable metric, describe repeated measurements and your choice of mean or percentile. If exact numbers were not retained, describe the method and trend without false precision.
Explain the investigation
'We added a cache' describes the final action. Explain how you ruled out other causes, which observation pointed to the bottleneck, and what experiment confirmed it. That reasoning shows how you might approach a new problem.
Name the limits of the improvement
A change may reduce latency while increasing consistency or maintenance costs. Explain how you watched for those costs, when you would roll back, and which situations the approach would not suit. An honest boundary makes the answer more useful.