Why I asked to pause a product-search launch

The release decision

In March 2026, Search Console reported 1,015 Poor URLs and zero Good URLs. We also had a new product-search feature ready to launch. I took the trend to my manager. We discussed shipping search while I worked on performance.

My concern was diagnosis. If the numbers changed again after search went live, I’d have another major change to account for. We agreed to pause the release and get the store back to a healthy baseline first.

What I changed

I added Redis object caching, replaced our image CDN with native WebP conversion, and went through the scripts to remove or delay loads where I could. Removing the CDN also took a recurring charge off the books.

What Search Console reported

The March count was the baseline I used in the launch discussion. By May 9 both device reports showed only Good URLs. Search Console rounded those later totals, and mobile and desktop were separate views of the store.

The pause cost us two weeks on the search schedule. When we did ship, we had a baseline we could use to investigate any new regression.