When the right schema is no schema
I spent Sunday cleaning up BinderDex's structured data after a Semrush audit flagged a problem I had been ignoring: every card price guide was emitting Product, Offer, and InStock JSON-LD. BinderDex is not the merchant for those cards. TCGplayer is. Google does not care that the markup was well-formed; it cares that it was false.
The fix was deletion. I removed the entire card-schema component and dropped commerce markup from set lists, ranked lists, and free-tool pages. A price guide that aggregates and links out is a CollectionPage or Article, not a product listing. Pretending otherwise is the kind of thing that triggers manual actions and quiet ranking losses, and you will not know which one hit you first.
Let your 404s be crawlable
The second surprise was in robots.txt. I had been disallowing the retired binder-builder path, which sounds tidy until you realize it means Google keeps those 28,203 legacy URLs as "indexed, blocked by robots" indefinitely. They never age out because the crawler can never see the 404.
- disallow: ["/api/", "/binders/builder", "/explore/", "/sign-in", "/sign-up"]
+ disallow: ["/api/", "/explore/", "/sign-in", "/sign-up"]Removing /binders/builder from the disallow list lets crawlers observe the 404s and drop the URLs from the index on their own schedule. Blocking a retired path feels protective; it actually freezes the problem in place.
Same posture on titles: card pages were generating 70+ character search titles with redundant set names and price strings. I bounded the page-specific portion to 55 characters with a candidate waterfall that tries the full title, then a shorter set-code variant, then a truncated fallback. The root layout appends the brand suffix, so the page portion has to leave room.
Both fixes came from the same realization: crawlers are not adversaries to hide things from. They are readers you want to be honest with. False schema and blocked 404s are both lies of omission, and Google penalizes both in different ways.