September 19, 2026 · Method
If your site answers at both www and the bare domain, search engines can index two copies and crawl each one less. Here is the ten second check.
Type your domain two ways: with www and without. If both load the site, both can be indexed. Google's own documentation calls a canonical tag a hint rather than an order, so the copy you did not mean to publish can be the one it spends its visits on.
Open your site at the address you do not normally use. If you say narrowsweb.com, try www.narrowsweb.com. Watch the address bar. If it snaps to the other version, that is a redirect doing its job. If it sits there and serves the page, you have two copies.
The exact version, if you are comfortable in a terminal: curl -sI https://www.yourdomain.com/. You want a 301 and a location line pointing at the other address.
One rule at your host or CDN, pointing every request at one address, permanently. Not a page that says moved. Not a meta refresh. A 301, the standard signal that the move is final. Then check a deep page and a page with a query string. A rule that only covers the homepage is not a fix.
Ours served both until September 11. Search Console recorded 525 Google visits over the previous ninety days. 222 went to the copy nobody was meant to see, and 24 real pages had never been fetched at all, the pricing page among them. The fix took one rule. Finding it took reading the crawl report.
We set every site we launch to answer on one address, with the other redirecting. We check it on the day of the cutover and again a week later, because a DNS change can put the second copy back.
ChartedMethod
Same method, run for you on a cadence. Start with a free two-score check, sent back within one business day.